Understanding People Pointing: The Perseus System. Roger E. Kahn and Michael J. Swain. University of Chicago East 58th Street.

Size: px
Start display at page:

Download "Understanding People Pointing: The Perseus System. Roger E. Kahn and Michael J. Swain. University of Chicago East 58th Street."

Transcription

1 Understanding People Pointing: The Perseus System Roger E. Kahn and Michael J. Swain Department of Computer Science University of Chicago 1100 East 58th Street Chicago, IL Abstract In this paper we present Perseus, a purposive visual system used by our robot, CHIP, to locate objects being pointed at by people. Perseus uses knowledge about the task and environment at all levels of processing to more accurately and eciently perform visual tasks. 1 Introduction One of the tasks our robot, CHIP, performs quite well is to pick up trash around our oces and throw it away[6]. Sometimes we get tired of watching it clean up the entire room and decide we want it to pick up one particular piece of trash. Providing a verbal description of the trash's location to CHIP is awkward; it is far more natural to simply point at it. For CHIP to nd objects this way it needs to notice people, recognize when they are pointing, determine which area they are pointing to, and nd the object in that area. In this paper we present Perseus, a visual architecture implemented for CHIP, that enables it to perform this task. One important aspect of Perseus is that processing at all levels is inuenced by knowledge about the task and environment. Thus, unnecessary computations are avoided and we are able to tune the operations used to the system's specic needs. Perseus receives the higher level knowledge through its interface with a reactive execution system, such as the RAP system[5]. 2 The system Perseus is a purposive vision system[1] composed of six components (see gure 1): a set of retinotopic feature maps[16, 13, 12, 17, 3, 14], a short term visual memory (STVM), a collection of object representations (ORs), markers[3], a visual routine library[18, 3], and a long term visual memory (LTVM). A reactive execution system activates the visual routines providing higher level information about the task being performed to each component of Perseus. This information is used to select regions of attention and to parameterize visual operators. Computations at even the Early Maps Intensity Edge Color Motion Disparity Object Representations Lights Floor Trash / Pepsi People Markers Head Hand Trash Short Term Visual Memory Segmentation Map Long Term Visual Memory Symbols Visual Routines Identify-Object Find-Object-Pointed-At Figure 1: The Perseus system in the pointing task lowest level in the system, the feature maps, are inuenced by the task being performed. 2.1 Feature maps Feature maps are retinotopic descriptions of the scene that provide information about the location of features to ORs and markers. This information is used to segment, track, locate, and recognize objects. To maximize the usefulness of feature maps, ORs and markers parameterize each map based on their goals. The feature maps used in the pointing task are edge (Sobel), motion (normal ow [9]), disparity (NEC stereo algorithm[4]), and intensity. Other feature maps, such as color histogram backprojection[15, 8, 7] and texture measures have proven useful in other tasks. Each map can be parameterized with a resolution and an area to operate over. Additional parameters are shown in Figure 2. The parameterization of feature maps not only increases their sensitivity to the particular feature values of interest to the task, but it also makes certain types of feature maps possible that could not be used if parameterization was not allowed. For instance, the

2 Map Color Edge Motion Disparity Intensity Parameters Histogram Sobel threshold Min velocity, Gradient threshold Disparity range Upper and lower intensity bounds Figure 2: Feature map parameters color histogram backprojection algorithm requires a histogram to compute the feature map. 2.2 Object representations In Perseus object representations are structures containing data that describes an object and methods for segmenting, tracking, recognizing, and locating it. Their purpose is to acquire information about the realworld object they represent. This is done by invoking their methods. The methods call visual operators and store the results in the OR. Thus, ORs are not immutable. They can be specialized as methods acquire information about the object they represent. This specialization includes acquiring knowledge about the object's location, trajectory, color, etc. Furthermore, the addition of knowledge to an OR may inuence how its methods interpret the scene. ORs are initially instantiated either by a visual routine or are recalled from LTVM. Once an OR is instantiated, visual routines send it messages requesting information or specifying that an action be performed. Some of the messages an OR may receive include: locate object within area, segment object, grab object's color histogram, and place a marker on object. The methods available for an OR can be quite specic to that object. For instance, the person representation may be able to perform face recognition. However, many methods will be common to a large number of ORs, and the underlying operations performed by them will often be the same. We may have many ORs that use color histogram backprojection to locate an object. ORs use feature maps to determine which parts of the scene are most important to the operations they are performing. When using the maps, an OR may parameterize them so they better serve the task at hand. For instance, when a Pepsi can OR is asked for its location it may instantiate the color map with a Pepsi histogram and focus its search on peaks in that map. When multiple ORs parameterize a single feature map, the feature map process creates separate maps for each of them. Thus ORs cannot interfere with each other's maps, and the number of feature maps computed at any given time is limited by the number of ORs using them. We have found that the number of objects simultaneously being attended to by the agent is typically small in real world problems. Thus the total number of maps requiring computation at any given time is reasonably small. Also, ORs may instantiate other ORs and send requests to them. For example, when the segmentation method of a person OR is invoked, it instantiates a oor OR and invokes its segmentation method. The person OR uses the oor's segmentation to eliminate possible locations of the person. ORs can also instantiate markers[3] to track important parts of the scene. 2.3 Short term visual memory The STVM is a retinotopic segmentation map. When an OR performs a segmentation, the STVM is updated. Each point in the STVM is marked as belonging to an object that has claimed it in a segmentation, or as free. Any OR may access the STVM to aid in a computation. For instance when an OR is performing a segmentation it can use the STVM to eliminate points based on whether they belong to another object or not. 2.4 Markers Markers are created by ORs to track objects or subparts of an object (like a person's hand), much as those used by Chapman[3]. Once a marker is tracking an object, visual routines and ORs may ask the marker for its position. If the marker has not lost track of the object, it returns its position in retinotopic coordinates, otherwise it returns a message saying so. When a marker is instantiated it is parameterized with a starting location and a tracking method. An example of a tracking method is a function that tracks the centroid of a blob in a feature map using a Kalman lter. As with ORs, it is important for markers to be able to track a large number of objects with a relatively small number of tracking methods so that this architecture can be used for a wide variety of tasks. 2.5 Visual routines The interface between Perseus and the higher level system is a library of task specic visual routines[18, 3]. Visual routines use markers, ORs, and LTVM to perform functions for the higher level system. Typical visual routines may nd an object a person is pointing to, or identify a person in the room. Visual routines may be passed parameters when they are called. One type of parameter for visual routines are symbols that the LTVM associates with particular ORs. This allows the symbolic, higher level system to call visual routines with particular ORs. Additionally, other types of data such as ORs and mark-

3 ers may be passed to visual routines. This commonly occurs when one visual routine calls another. 2.6 Long term visual memory The LTVM associates symbols with ORs to allow the higher level system to reference objects when it calls visual routines. Each OR is typed and contains a comparison method used to see if two ORs represent the same object in the scene. The comparison method may use dierent operations depending upon the information contained in each OR. For instance, if two soda can representations exist that have previously been sent a message to segment-and-grab-histogram, color histogram intersection may be used to compare the two representations. However, if the histogram had not been acquired, but an edge representation for each can is known instead, the Haussdorf distance may be used for the comparison. By performing comparisons between objects the higher level system is able to determine if it has seen a particular object before. 3 Example: Finding trash The Perseus system has been used to identify trash pointed to by people. We assume that a stationary camera is observing a scene containing a single person. The pointing gesture is dened to occur when the person extends either of his arms to his side and holds it still for a while. This system has been implemented on CHIP and used to locate objects pointed to by visitors to our laboratory. The high level system needs to nd an object the person is pointing to. To do this it calls the visual routine Find-Object-Pointed-At. This visual routine requires two arguments, an OR for a person (person) and an OR for the object pointed at (trash). LTVM contains generic ORs of people and trash, i.e. ORs that do not have details such as a color histogram of the trash in their data, but do have information such as bounds on the size of trash. The high level system passes the symbols associated with these ORs to Find- Object-Pointed-At and it accesses the ORs via LTVM. Methods of person and trash will be invoked by the visual routine to ll in details about the objects they represent. Thus once these ORs are passed to Find-Object-Pointed-At they are no longer thought of as generic ORs, but instead as particular ORs being used in this particular task. Eventually, very specic information will be acquired by the OR's methods and stored in them. For instance, the location of the person's head will eventually be acquired by one of person's methods and stored in the person OR. It should be noted that if the higher level system's goal were to nd a particular type of trash being pointed at by a particular person, it would use that trash's OR rather than the generic one. The data contained in it would inuence how it was found in the scene. For instance, if we know the person will point at a Pepsi can, we could use the pepsi-can OR. This would result in the particular colors in Pepsi cans being used to further restrict the search in the area pointed to. 3.1 Finding the person Find-Object-Pointed-At rst invokes person's ndlocation method. This method currently assumes that only people move in the scene, thus motion is used to nd a person. It parameterizes the motion feature map to detect medium velocity and waits for a large area of motion in the map, see gure 3. Once motion is detected, a marker is placed on the centroid of the largest moving connected component. The marker is parameterized to track the centroid of this component in the motion map. The marker is returned. 3.2 Segmenting the person Next Find-Object-Pointed-At invokes person's ndbody-parts method which segments the person and places markers on the person's head and hands. The rst step in the method for nding the hands and head is to segment the person. The disparity map and motion maps are used to nd an average disparity of the person. This is done by averaging the disparity at each point in the scene that moves (we are assuming only one person is in the scene). Then the disparity map is parameterized to detect disparities within a range corresponding to a typical person's width around this average. The person is segmented by extracting connected components from the resulting disparity map near the marker tracking the person's body. Since the lights in the room saturate our cameras, the stereo algorithm may give erroneous information about their disparity. Therefore, before segmenting, person instantiates an OR for the lights (lights) and uses it to ignore the lights in the scene. Another dif- culty is that the oor has the same disparity as the person's feet where they are standing. person instantiates a oor OR (oor) so it can ignore the oor as well. Once lights is instantiated person invokes its segment method. This method sets the intensity feature map to threshold the image for high intensities. The STVM is updated with the saturated points from the intensity map. Person then instantiates oor and calls its segment method. This method assumes that the oor is textureless, so to get a rough segmentation all we must do is to nd the lowest edges in the scene. oor parameterizes the edge map and segments the oor as the pixels below the lowest edge in each row in the

4 Figure 3: The motion feature map, the short term visual memory, and the segmented person map. This is the same method Horswill uses for navigation of his robot Polly[10]. STVM is then updated with the resulting segmentation. Figure 3 shows the STVM after the lights and oor have been segmented. Person uses the disparity map and the STVM to segment the person by accepting regions in the disparity map that are near the marker on the person and are not part of the oor or lights, see gure Finding the head The nd-body-parts method now nds the head by locating the highest point in the person segmentation that is approximately above the marker tracking the person's body. If the head is found a marker is placed on it and parameterized to track the head as the highest point in its connected component in the motion map. This tracking method is the same as the one used to track the centroid of the person's body in section 3.1 except it is parameterized to track the top rather than the centroid of the component. 3.4 Finding the hands Now nd-body-parts nds the hands by nding the leftmost and rightmost points in the person segmentation. These points are required to be approximately to the side of the marker tracking the person's body. They are also required to be connected to the body by segments that are thin enough to be arms. Each hand found has a marker placed on it, but both hands do not have to be found. The hand markers are parameterized to track the leftmost/rightmost point in its connected component in the motion map. This tracking method is the same as the one used to track the centroid of the person's body in section 3.1 except it is parameterized to track the side rather than the centroid of the component. 3.5 Finding the area pointed to Find-Object-Pointed-At now monitors the position of the head and hand markers. If either is lost it relocates them as described above. If either hand marker maintains a steady position far to the side of the person for a short time, the pointing gesture has occurred. Find-Object-Pointed-At computes the line of sight from the head to the hand. This line approximately denotes the area occluded for the person by their hand. Next, Perseus computes a two degree wide cone centered around this line, starting at the hand, and projecting away from the person. The area contained within this cone is the region pointed at. 3.6 Finding the trash Find-Object-Pointed-At now invokes trash's locatein-region method with the search cone as a parameter. This method locates trash by searching for small, densely textured areas in the scene. A more sophisticated method could be used, but for the purpose of this experiment this denition is sucient (the dense texture denition locates cans, styrofoam cups, and crumpled pieces of paper on textureless surfaces like the oor of our lab). trash parameterizes the edge map to locate ne textures over the search area. If trash is found a marker is placed on it and parameterized to track the centroid of the region in the texture map; otherwise a message describing the situation is returned. The method used to track the trash is the same as the one used to track the centroid of the person's body in section 3.1 except it is parameterized to use the edge map rather than the motion map. If trash is not located the cone's width is increased until either it is found or the cone becomes too wide. Figure 4 shows the search area, head marker, hand marker, and trash marker. 3.7 Identifying the trash Find-Object-Pointed-At returns with a message to the higher level system saying if the trash has been found. The higher level system now determines if it knows the name of this trash. It does this by calling the Identify-Object visual routine with trash as an argument. Identify-Object compares its passed OR with

5 Pointing trial Person Figure 5: Success of Perseus over successive trials. The number in each box describes the angle the trash was from the line of sight; it is in bold type if Perseus correctly found it. Blank entries indicate Perseus gave up. Figure 4: Perseus has found a Pepsi can the other ORs in LTVM of the same type using their comparison method. LTVM has ORs of Pepsi cans (pepsi-can), and Mountain Dew cans (mountain-dew-can). These ORs contain color histograms of the cans they represent and their method for comparison involves performing a histogram intersection[15]. When trash is compared with these ORs, the comparison method asks trash for its color histogram. The rst time trash receives this request its grab-histogram method is invoked which nds the convex hull of the textured region in the edge map and histograms the segmentation. In the future we will use a more sophisticated segmentation method, but in our restricted domain this procedure works well. Once the histogram is known, future segmentations are not required. Trash is compared with pepsi-can and mountain-dew-can. If either of these comparisons has a suciently high color intersection it is considered a match and returned. 4 Experiments This system has been tested with a variety of people pointing at objects in numerous positions on the oor. Typically when someone visits our lab we show them the system, and have found that it can successfully nd objects pointed to by our guests. Here we describe the system's success with ve people who have never used it before. Each person was told to place a piece of trash on the ground where CHIP could see it. Then he or she was asked to stand anywhere in CHIP's eld of view and point at the trash for at least three seconds. The subjects were not told to hold their arm in any particular position, yet we found that in almost all cases the line of sight from their head to their hand success- fully described the region they were pointing at. Table 5 shows the angle between the search line Perseus computes and the trash pointed at. Numbers in bold denote objects that Perseus correctly located; while angles in normal type refer to failures. Blanks in this table occur in places where Perseus gave up. It is interesting to note that the system has no problem nding trash lying at a dierent distance from the camera than the person. This is because lines project to lines under perspective projection, and therefore the line of sight from the head to the hand to the trash is still a line in the image plane. The only failure in this situation is when the trash lies between the person and CHIP, which occurs because Perseus currently can only nd the person's hand when it is to the side of his body. Our current implementation exhibits three other failure types. The rst type occurs when Perseus nds the person's head and hands correctly, but the trash pointed at is not near the search cone. This only happened 1 time in the above 40 runs, which suggests that people often do point along the line from their eye to their hand. The second failure type occurs when Perseus incorrectly locates the person's body parts. This type of error is almost always the result of an incorrect segmentation of the person. When a person stands too close to CHIP their legs may be included in the oor segmentation because of the oor OR's simplistic assumption that any region adjoining the bottom of the image is the oor. Another problem occurs when the person stands too close to other objects. This causes the disparity-based segmentation to fail. The nal type of failure results from our oversimplied trash representation. Non-trash objects will occasionally be found by the trash OR if they lie within the region pointed to.

6 As can be seen by the data, the system has been quite successful. It has proved to be robust to variations in the clothing people use, the hand they point with, their height, and many other variables. In the above experiments it correctly located the object 68% of the time, and 80% of the time the object was with 20 degrees of the search area. 5 Future work The failures described in Section 4 have resulted primarily from an over-simplication of our ORs, not from the Perseus architecture. Currently we are modifying the methods used by the person and oor ORs to be more robust, and we are rewriting the trash OR from scratch. The oor OR is being improved so that it can be used on textured oors. It will electronically verge the cameras to place their horoptor on the ground plane. Segmentation is then be done with a zero disparity lter[2]. This allows us to nd textured oors, and does not fail if a person is standing close to the camera. We are also experimenting with other methods for segmenting the person that can be used in conjunction with the current disparity segmentation procedure. In particular we have found that histogramming the intensity of each point in the image over time and performing gure/ground segmentation based upon how each pixel in the current frame diers from the mode of its histogram can improve our segmentation [11]. 6 Summary and conclusions A system based upon Perseus, a purposive vision architecture, has been presented and used to locate objects being pointed at in a non-engineered environment in real-time. The system has been tested with numerous people pointing at objects before various backgrounds while wearing a variety of clothes. The success of the system lies in its ability to use information about on the task and environment at every level of the architecture: feature maps are parameterized by the objects that are important to the system's goals, a short term visual memory allows object representations to know where other objects relevant to the task are, markers track objects with dierent methods depending on what they are tracking, the information extracted about objects in the scene is determined by the visual routines being run, and, of course, the visual routines are selected based on the goals of the higher level system. Because of this, Perseus is able to avoid unneeded computations, optimize operations for the particular goals of the system, and perform calculations at the lowest level that are not possible without task specic information. 7 Acknowledgements The authors would like to thank R. James Firby and Peter N. Prokopowicz for their contributions toward the design of our system and the Perseus architecture. References [1] J. Aloimonos. Purposive and qualitative active vision. International Conference on Pattern Recognition, pages 346{360, [2] P. J. Burt, P. Anandan, K. Hana, and G. van der Wal. A front end vision processor for unmanned vehicles. Technical report, David Sarno Research Center, [3] D. Chapman. Vision, Instruction, and Action. MIT Press, [4] I. J. Cox, S. Hingorani, B. M. Maggs, and S. B. Rao. Stereo without disparity gradient smoothing: a bayesian sensor fusion solution. British Machine Vision Conference, pages 337{346, [5] R. J. Firby. Adaptive Execution in Complex Dynamic Worlds. PhD thesis, Yale, [6] R. J. Firby, R. E. Kahn, P. N. Prokopowicz, and M. J. Swain. Collecting trash: A test of purposive vision. Workshop on Vision for Robots, August [7] B. V. Funt and G. D. Finlayson. Color constant color indexing. IEEE Pattern Analysis and Machine Intelligence, in press. [8] G. Healey and D. Slater. Using illumination invariant color histogram descriptors for recognition. Computer Vision and Pattern Recognition, pages 355{360, [9] B. K. P. Horn and B. G. Schunck. Determining optical ow. Articial Intelligence, 17(1-3):185{203, Augest [10] I. Horswill. Polly: A vision-based articial agent. American Association for Articial Intelligence, [11] R. E. Kahn and M. J. Swain. Background detection for segmentation. Animate Agent Project Working Note 8, University of Chicago, Augest [12] J. Malik and P. Perona. Preattentive texture discrimination with early vision mechanisms. Optical Society of America, 7(5):923{932, [13] D. Marr. Vision. W. H. Freeman and Company, San Francisco California, [14] R. Milanese. Detecting Salient Regions in an Image: From Biological Evidence to Computer Implementation. PhD thesis, University of Geneva, [15] M. J. Swain and D. H. Ballard. Color indexing. International Journal of Computer Vision, 7:11{32, [16] A. Treisman and G. Gelade. A feature-integration theory of attention. Cognitive Psychology, 12:97{136, [17] J. K. Tsotsos. Analyzing vision at the complexity level. Behavioral and Brain Sciences, 13:423{469, [18] S. Ullman. Visual routines. Cognition, 18:97{159, 1984.

and, if satised, the task is considered complete and the and one of the methods with a satised test is selected.

and, if satised, the task is considered complete and the and one of the methods with a satised test is selected. An Architecture for Vision and Action R. James Firby, Roger E. Kahn, Peter N. Prokopowicz and Michael J. Swain Department of Computer Science University of Chicago 1100 E. 58th St., Chicago, IL 60637 rby,kahn,peterp,swain@cs.uchicago.edu

More information

Design Specication. Group 3

Design Specication. Group 3 Design Specication Group 3 September 20, 2012 Project Identity Group 3, 2012/HT, "The Robot Dog" Linköping University, ISY Name Responsibility Phone number E-mail Martin Danelljan Design 072-372 6364 marda097@student.liu.se

More information

A Real Time System for Detecting and Tracking People. Ismail Haritaoglu, David Harwood and Larry S. Davis. University of Maryland

A Real Time System for Detecting and Tracking People. Ismail Haritaoglu, David Harwood and Larry S. Davis. University of Maryland W 4 : Who? When? Where? What? A Real Time System for Detecting and Tracking People Ismail Haritaoglu, David Harwood and Larry S. Davis Computer Vision Laboratory University of Maryland College Park, MD

More information

Local qualitative shape from stereo. without detailed correspondence. Extended Abstract. Shimon Edelman. Internet:

Local qualitative shape from stereo. without detailed correspondence. Extended Abstract. Shimon Edelman. Internet: Local qualitative shape from stereo without detailed correspondence Extended Abstract Shimon Edelman Center for Biological Information Processing MIT E25-201, Cambridge MA 02139 Internet: edelman@ai.mit.edu

More information

Active Motion Detection and Object Tracking. Joachim Denzler and Dietrich W.R.Paulus.

Active Motion Detection and Object Tracking. Joachim Denzler and Dietrich W.R.Paulus. 0 Active Motion Detection and Object Tracking Joachim Denzler and Dietrich W.R.Paulus denzler,paulus@informatik.uni-erlangen.de The following paper was published in the Proceedings on the 1 st International

More information

Collecting Trash: A Test of Purposive Vision. R. James Firby, Roger E. Kahn, Peter N. Prokopowicz, and Michael J. Swain,

Collecting Trash: A Test of Purposive Vision. R. James Firby, Roger E. Kahn, Peter N. Prokopowicz, and Michael J. Swain, Collecting Trash: A Test of Purposive Vision R. James Firby, Roger E. Kahn, Peter N. Prokopowicz, and Michael J. Swain, Department of Computer Science University of Chicago Ryerson 152, 1100 E. 58th St.,

More information

Center for Automation Research, University of Maryland. The independence measure is the residual normal

Center for Automation Research, University of Maryland. The independence measure is the residual normal Independent Motion: The Importance of History Robert Pless, Tomas Brodsky, and Yiannis Aloimonos Center for Automation Research, University of Maryland College Park, MD, 74-375 Abstract We consider a problem

More information

CS4670: Computer Vision

CS4670: Computer Vision CS4670: Computer Vision Noah Snavely Lecture 6: Feature matching and alignment Szeliski: Chapter 6.1 Reading Last time: Corners and blobs Scale-space blob detector: Example Feature descriptors We know

More information

3. International Conference on Face and Gesture Recognition, April 14-16, 1998, Nara, Japan 1. A Real Time System for Detecting and Tracking People

3. International Conference on Face and Gesture Recognition, April 14-16, 1998, Nara, Japan 1. A Real Time System for Detecting and Tracking People 3. International Conference on Face and Gesture Recognition, April 14-16, 1998, Nara, Japan 1 W 4 : Who? When? Where? What? A Real Time System for Detecting and Tracking People Ismail Haritaoglu, David

More information

Bitangent 3. Bitangent 1. dist = max Region A. Region B. Bitangent 2. Bitangent 4

Bitangent 3. Bitangent 1. dist = max Region A. Region B. Bitangent 2. Bitangent 4 Ecient pictograph detection Dietrich Buesching TU Muenchen, Fakultaet fuer Informatik FG Bildverstehen 81667 Munich, Germany email: bueschin@informatik.tu-muenchen.de 1 Introduction Pictographs are ubiquitous

More information

A Two Stage Real{Time Object Tracking System. Universitat Erlangen{Nurnberg. Martensstr. 3, D{91058 Erlangen, Germany

A Two Stage Real{Time Object Tracking System. Universitat Erlangen{Nurnberg. Martensstr. 3, D{91058 Erlangen, Germany 0 A Two Stage Real{Time Object Tracking System J. Denzler, H. Niemann Lehrstuhl fur Mustererkennung (Informatik 5) Universitat Erlangen{Nurnberg Martensstr. 3, D{91058 Erlangen, Germany email: fdenzler,niemanng@informatik.uni-erlangen.de

More information

Proceedings of the 6th Int. Conf. on Computer Analysis of Images and Patterns. Direct Obstacle Detection and Motion. from Spatio-Temporal Derivatives

Proceedings of the 6th Int. Conf. on Computer Analysis of Images and Patterns. Direct Obstacle Detection and Motion. from Spatio-Temporal Derivatives Proceedings of the 6th Int. Conf. on Computer Analysis of Images and Patterns CAIP'95, pp. 874-879, Prague, Czech Republic, Sep 1995 Direct Obstacle Detection and Motion from Spatio-Temporal Derivatives

More information

A Model of Dynamic Visual Attention for Object Tracking in Natural Image Sequences

A Model of Dynamic Visual Attention for Object Tracking in Natural Image Sequences Published in Computational Methods in Neural Modeling. (In: Lecture Notes in Computer Science) 2686, vol. 1, 702-709, 2003 which should be used for any reference to this work 1 A Model of Dynamic Visual

More information

of human activities. Our research is motivated by considerations of a ground-based mobile surveillance system that monitors an extended area for

of human activities. Our research is motivated by considerations of a ground-based mobile surveillance system that monitors an extended area for To Appear in ACCV-98, Mumbai-India, Material Subject to ACCV Copy-Rights Visual Surveillance of Human Activity Larry Davis 1 Sandor Fejes 1 David Harwood 1 Yaser Yacoob 1 Ismail Hariatoglu 1 Michael J.

More information

Optical Flow-Based Person Tracking by Multiple Cameras

Optical Flow-Based Person Tracking by Multiple Cameras Proc. IEEE Int. Conf. on Multisensor Fusion and Integration in Intelligent Systems, Baden-Baden, Germany, Aug. 2001. Optical Flow-Based Person Tracking by Multiple Cameras Hideki Tsutsui, Jun Miura, and

More information

Automatic Tracking of Moving Objects in Video for Surveillance Applications

Automatic Tracking of Moving Objects in Video for Surveillance Applications Automatic Tracking of Moving Objects in Video for Surveillance Applications Manjunath Narayana Committee: Dr. Donna Haverkamp (Chair) Dr. Arvin Agah Dr. James Miller Department of Electrical Engineering

More information

Computer and Machine Vision

Computer and Machine Vision Computer and Machine Vision Lecture Week 4 Part-2 February 5, 2014 Sam Siewert Outline of Week 4 Practical Methods for Dealing with Camera Streams, Frame by Frame and De-coding/Re-encoding for Analysis

More information

Notes 9: Optical Flow

Notes 9: Optical Flow Course 049064: Variational Methods in Image Processing Notes 9: Optical Flow Guy Gilboa 1 Basic Model 1.1 Background Optical flow is a fundamental problem in computer vision. The general goal is to find

More information

Flow Estimation. Min Bai. February 8, University of Toronto. Min Bai (UofT) Flow Estimation February 8, / 47

Flow Estimation. Min Bai. February 8, University of Toronto. Min Bai (UofT) Flow Estimation February 8, / 47 Flow Estimation Min Bai University of Toronto February 8, 2016 Min Bai (UofT) Flow Estimation February 8, 2016 1 / 47 Outline Optical Flow - Continued Min Bai (UofT) Flow Estimation February 8, 2016 2

More information

Measurement of 3D Foot Shape Deformation in Motion

Measurement of 3D Foot Shape Deformation in Motion Measurement of 3D Foot Shape Deformation in Motion Makoto Kimura Masaaki Mochimaru Takeo Kanade Digital Human Research Center National Institute of Advanced Industrial Science and Technology, Japan The

More information

A Survey of Light Source Detection Methods

A Survey of Light Source Detection Methods A Survey of Light Source Detection Methods Nathan Funk University of Alberta Mini-Project for CMPUT 603 November 30, 2003 Abstract This paper provides an overview of the most prominent techniques for light

More information

Fundamentals of Stereo Vision Michael Bleyer LVA Stereo Vision

Fundamentals of Stereo Vision Michael Bleyer LVA Stereo Vision Fundamentals of Stereo Vision Michael Bleyer LVA Stereo Vision What Happened Last Time? Human 3D perception (3D cinema) Computational stereo Intuitive explanation of what is meant by disparity Stereo matching

More information

What is Computer Vision?

What is Computer Vision? Perceptual Grouping in Computer Vision Gérard Medioni University of Southern California What is Computer Vision? Computer Vision Attempt to emulate Human Visual System Perceive visual stimuli with cameras

More information

Department of Electrical Engineering, Keio University Hiyoshi Kouhoku-ku Yokohama 223, Japan

Department of Electrical Engineering, Keio University Hiyoshi Kouhoku-ku Yokohama 223, Japan Shape Modeling from Multiple View Images Using GAs Satoshi KIRIHARA and Hideo SAITO Department of Electrical Engineering, Keio University 3-14-1 Hiyoshi Kouhoku-ku Yokohama 223, Japan TEL +81-45-563-1141

More information

Integrated Object Recognition with Extended Hamming Distance

Integrated Object Recognition with Extended Hamming Distance Integrated Object Recognition with Extended Hamming Distance Vladimir A. Kulyukin Abraham Bookstein School of Computer Science Center for Information and Language Studies DePaul University The University

More information

The SUMO Speaker Series for Undergraduates. The Art Gallery Problem

The SUMO Speaker Series for Undergraduates. The Art Gallery Problem The SUMO Speaker Series for Undergraduates (food from Pizza Chicago) Wednesday, April 21 4:40-5:30, room 380C The Art Gallery Problem Amy Pang Abstract: Imagine you are the curator of an art gallery with

More information

BSB663 Image Processing Pinar Duygulu. Slides are adapted from Selim Aksoy

BSB663 Image Processing Pinar Duygulu. Slides are adapted from Selim Aksoy BSB663 Image Processing Pinar Duygulu Slides are adapted from Selim Aksoy Image matching Image matching is a fundamental aspect of many problems in computer vision. Object or scene recognition Solving

More information

3D-Tracking of Head and Hands for Pointing Gesture Recognition in a Human-Robot Interaction Scenario

3D-Tracking of Head and Hands for Pointing Gesture Recognition in a Human-Robot Interaction Scenario 3D-Tracking of Head and Hands for Pointing Gesture Recognition in a Human-Robot Interaction Scenario Kai Nickel Edgar Seemann Rainer Stiefelhagen Interactive Systems Labs Universitt Karlsruhe (TH) Germany

More information

Self Lane Assignment Using Smart Mobile Camera For Intelligent GPS Navigation and Traffic Interpretation

Self Lane Assignment Using Smart Mobile Camera For Intelligent GPS Navigation and Traffic Interpretation For Intelligent GPS Navigation and Traffic Interpretation Tianshi Gao Stanford University tianshig@stanford.edu 1. Introduction Imagine that you are driving on the highway at 70 mph and trying to figure

More information

Human Motion Detection and Tracking for Video Surveillance

Human Motion Detection and Tracking for Video Surveillance Human Motion Detection and Tracking for Video Surveillance Prithviraj Banerjee and Somnath Sengupta Department of Electronics and Electrical Communication Engineering Indian Institute of Technology, Kharagpur,

More information

SUMMARY: DISTINCTIVE IMAGE FEATURES FROM SCALE- INVARIANT KEYPOINTS

SUMMARY: DISTINCTIVE IMAGE FEATURES FROM SCALE- INVARIANT KEYPOINTS SUMMARY: DISTINCTIVE IMAGE FEATURES FROM SCALE- INVARIANT KEYPOINTS Cognitive Robotics Original: David G. Lowe, 004 Summary: Coen van Leeuwen, s1460919 Abstract: This article presents a method to extract

More information

Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition, pages 80-85, Variable-Scale Smoothing and Edge Detection Guided b

Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition, pages 80-85, Variable-Scale Smoothing and Edge Detection Guided b Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition, pages 80-85, 1998. Variable-Scale Smoothing and Edge Detection Guided by Stereoscopy Clark F. Olson Jet Propulsion Laboratory,

More information

Attentive Face Detection and Recognition? Volker Kruger, Udo Mahlmeister, and Gerald Sommer. University of Kiel, Germany,

Attentive Face Detection and Recognition? Volker Kruger, Udo Mahlmeister, and Gerald Sommer. University of Kiel, Germany, Attentive Face Detection and Recognition? Volker Kruger, Udo Mahlmeister, and Gerald Sommer University of Kiel, Germany, Preuerstr. 1-9, 24105 Kiel, Germany Tel: ++49-431-560496 FAX: ++49-431-560481 vok@informatik.uni-kiel.de,

More information

PRELIMINARY RESULTS ON REAL-TIME 3D FEATURE-BASED TRACKER 1. We present some preliminary results on a system for tracking 3D motion using

PRELIMINARY RESULTS ON REAL-TIME 3D FEATURE-BASED TRACKER 1. We present some preliminary results on a system for tracking 3D motion using PRELIMINARY RESULTS ON REAL-TIME 3D FEATURE-BASED TRACKER 1 Tak-keung CHENG derek@cs.mu.oz.au Leslie KITCHEN ljk@cs.mu.oz.au Computer Vision and Pattern Recognition Laboratory, Department of Computer Science,

More information

Proximity States and Motion Vectors. Behavior Modes and Global Locations. Correlation set Relative Location. Correlation Peak Assessment

Proximity States and Motion Vectors. Behavior Modes and Global Locations. Correlation set Relative Location. Correlation Peak Assessment Integrating a behavior-based approach to active stereo vision with an intelligent control architecture for mobile robots David Kortenkamp, Eric Huber and Glenn Wasson Metrica Inc. Texas Robotics and Automation

More information

To appear in ECCV-94, Stockholm, Sweden, May 2-6, 1994.

To appear in ECCV-94, Stockholm, Sweden, May 2-6, 1994. To appear in ECCV-94, Stockholm, Sweden, May 2-6, 994. Recognizing Hand Gestures? James Davis and Mubarak Shah?? Computer Vision Laboratory, University of Central Florida, Orlando FL 3286, USA Abstract.

More information

1st frame Figure 1: Ball Trajectory, shadow trajectory and a reference player 48th frame the points S and E is a straight line and the plane formed by

1st frame Figure 1: Ball Trajectory, shadow trajectory and a reference player 48th frame the points S and E is a straight line and the plane formed by Physics-based 3D Position Analysis of a Soccer Ball from Monocular Image Sequences Taeone Kim, Yongduek Seo, Ki-Sang Hong Dept. of EE, POSTECH San 31 Hyoja Dong, Pohang, 790-784, Republic of Korea Abstract

More information

Marcel Worring Intelligent Sensory Information Systems

Marcel Worring Intelligent Sensory Information Systems Marcel Worring worring@science.uva.nl Intelligent Sensory Information Systems University of Amsterdam Information and Communication Technology archives of documentaries, film, or training material, video

More information

ATRANSPUTER BASED LASER SCANNING SYSTEM

ATRANSPUTER BASED LASER SCANNING SYSTEM ATRANSPUTER BASED LASER SCANNING SYSTEM Mr G TIZWKESBURY, Dr D J HARRISON, Dr D A SANDERS, Prof J. BILLINGSLEY, Dr J E L Hollis Automation & Robotics Research Laboratory, School of Systems Engineering,

More information

The SIFT (Scale Invariant Feature

The SIFT (Scale Invariant Feature The SIFT (Scale Invariant Feature Transform) Detector and Descriptor developed by David Lowe University of British Columbia Initial paper ICCV 1999 Newer journal paper IJCV 2004 Review: Matt Brown s Canonical

More information

An Interactive Technique for Robot Control by Using Image Processing Method

An Interactive Technique for Robot Control by Using Image Processing Method An Interactive Technique for Robot Control by Using Image Processing Method Mr. Raskar D. S 1., Prof. Mrs. Belagali P. P 2 1, E&TC Dept. Dr. JJMCOE., Jaysingpur. Maharashtra., India. 2 Associate Prof.

More information

Image-Based Memory of Environment. homing uses a similar idea that the agent memorizes. [Hong 91]. However, the agent nds diculties in arranging its

Image-Based Memory of Environment. homing uses a similar idea that the agent memorizes. [Hong 91]. However, the agent nds diculties in arranging its Image-Based Memory of Environment Hiroshi ISHIGURO Department of Information Science Kyoto University Kyoto 606-01, Japan E-mail: ishiguro@kuis.kyoto-u.ac.jp Saburo TSUJI Faculty of Systems Engineering

More information

EECS150 - Digital Design Lecture 14 FIFO 2 and SIFT. Recap and Outline

EECS150 - Digital Design Lecture 14 FIFO 2 and SIFT. Recap and Outline EECS150 - Digital Design Lecture 14 FIFO 2 and SIFT Oct. 15, 2013 Prof. Ronald Fearing Electrical Engineering and Computer Sciences University of California, Berkeley (slides courtesy of Prof. John Wawrzynek)

More information

Colour Segmentation-based Computation of Dense Optical Flow with Application to Video Object Segmentation

Colour Segmentation-based Computation of Dense Optical Flow with Application to Video Object Segmentation ÖGAI Journal 24/1 11 Colour Segmentation-based Computation of Dense Optical Flow with Application to Video Object Segmentation Michael Bleyer, Margrit Gelautz, Christoph Rhemann Vienna University of Technology

More information

Multi-view stereo. Many slides adapted from S. Seitz

Multi-view stereo. Many slides adapted from S. Seitz Multi-view stereo Many slides adapted from S. Seitz Beyond two-view stereo The third eye can be used for verification Multiple-baseline stereo Pick a reference image, and slide the corresponding window

More information

Chapter 12 3D Localisation and High-Level Processing

Chapter 12 3D Localisation and High-Level Processing Chapter 12 3D Localisation and High-Level Processing This chapter describes how the results obtained from the moving object tracking phase are used for estimating the 3D location of objects, based on the

More information

Tracking Multiple Objects in 3D. Coimbra 3030 Coimbra target projection in the same image position (usually

Tracking Multiple Objects in 3D. Coimbra 3030 Coimbra target projection in the same image position (usually Tracking Multiple Objects in 3D Jo~ao P. Barreto Paulo Peioto Coimbra 33 Coimbra 33 Jorge Batista Helder Araujo Coimbra 33 Coimbra 33 Abstract In this paper a system for tracking multiple targets in 3D

More information

A method for depth-based hand tracing

A method for depth-based hand tracing A method for depth-based hand tracing Khoa Ha University of Maryland, College Park khoaha@umd.edu Abstract An algorithm for natural human-computer interaction via in-air drawing is detailed. We discuss

More information

What have we leaned so far?

What have we leaned so far? What have we leaned so far? Camera structure Eye structure Project 1: High Dynamic Range Imaging What have we learned so far? Image Filtering Image Warping Camera Projection Model Project 2: Panoramic

More information

Available online at ScienceDirect. Procedia Computer Science 22 (2013 )

Available online at   ScienceDirect. Procedia Computer Science 22 (2013 ) Available online at www.sciencedirect.com ScienceDirect Procedia Computer Science 22 (2013 ) 945 953 17 th International Conference in Knowledge Based and Intelligent Information and Engineering Systems

More information

Miniature faking. In close-up photo, the depth of field is limited.

Miniature faking. In close-up photo, the depth of field is limited. Miniature faking In close-up photo, the depth of field is limited. http://en.wikipedia.org/wiki/file:jodhpur_tilt_shift.jpg Miniature faking Miniature faking http://en.wikipedia.org/wiki/file:oregon_state_beavers_tilt-shift_miniature_greg_keene.jpg

More information

Dominant plane detection using optical flow and Independent Component Analysis

Dominant plane detection using optical flow and Independent Component Analysis Dominant plane detection using optical flow and Independent Component Analysis Naoya OHNISHI 1 and Atsushi IMIYA 2 1 School of Science and Technology, Chiba University, Japan Yayoicho 1-33, Inage-ku, 263-8522,

More information

Computer Vision. Introduction

Computer Vision. Introduction Computer Vision Introduction Filippo Bergamasco (filippo.bergamasco@unive.it) http://www.dais.unive.it/~bergamasco DAIS, Ca Foscari University of Venice Academic year 2016/2017 About this course Official

More information

COMPARATIVE STUDY OF DIFFERENT APPROACHES FOR EFFICIENT RECTIFICATION UNDER GENERAL MOTION

COMPARATIVE STUDY OF DIFFERENT APPROACHES FOR EFFICIENT RECTIFICATION UNDER GENERAL MOTION COMPARATIVE STUDY OF DIFFERENT APPROACHES FOR EFFICIENT RECTIFICATION UNDER GENERAL MOTION Mr.V.SRINIVASA RAO 1 Prof.A.SATYA KALYAN 2 DEPARTMENT OF COMPUTER SCIENCE AND ENGINEERING PRASAD V POTLURI SIDDHARTHA

More information

Finally: Motion and tracking. Motion 4/20/2011. CS 376 Lecture 24 Motion 1. Video. Uses of motion. Motion parallax. Motion field

Finally: Motion and tracking. Motion 4/20/2011. CS 376 Lecture 24 Motion 1. Video. Uses of motion. Motion parallax. Motion field Finally: Motion and tracking Tracking objects, video analysis, low level motion Motion Wed, April 20 Kristen Grauman UT-Austin Many slides adapted from S. Seitz, R. Szeliski, M. Pollefeys, and S. Lazebnik

More information

A Qualitative Analysis of 3D Display Technology

A Qualitative Analysis of 3D Display Technology A Qualitative Analysis of 3D Display Technology Nicholas Blackhawk, Shane Nelson, and Mary Scaramuzza Computer Science St. Olaf College 1500 St. Olaf Ave Northfield, MN 55057 scaramum@stolaf.edu Abstract

More information

Proc. Int. Symp. Robotics, Mechatronics and Manufacturing Systems 92 pp , Kobe, Japan, September 1992

Proc. Int. Symp. Robotics, Mechatronics and Manufacturing Systems 92 pp , Kobe, Japan, September 1992 Proc. Int. Symp. Robotics, Mechatronics and Manufacturing Systems 92 pp.957-962, Kobe, Japan, September 1992 Tracking a Moving Object by an Active Vision System: PANTHER-VZ Jun Miura, Hideharu Kawarabayashi,

More information

GARGOYLE: An Environment for e, Context-Sensitive Active Vision

GARGOYLE: An Environment for e, Context-Sensitive Active Vision From: AAAI-96 Proceedings. Copyright 1996, AAAI (www.aaai.org). All rights reserved. GARGOYLE: An Environment for e, Context-Sensitive Active Vision eter N. Prokopowicz, Michael J. Swain, R. James Firby,

More information

Local Feature Detectors

Local Feature Detectors Local Feature Detectors Selim Aksoy Department of Computer Engineering Bilkent University saksoy@cs.bilkent.edu.tr Slides adapted from Cordelia Schmid and David Lowe, CVPR 2003 Tutorial, Matthew Brown,

More information

COLOR FIDELITY OF CHROMATIC DISTRIBUTIONS BY TRIAD ILLUMINANT COMPARISON. Marcel P. Lucassen, Theo Gevers, Arjan Gijsenij

COLOR FIDELITY OF CHROMATIC DISTRIBUTIONS BY TRIAD ILLUMINANT COMPARISON. Marcel P. Lucassen, Theo Gevers, Arjan Gijsenij COLOR FIDELITY OF CHROMATIC DISTRIBUTIONS BY TRIAD ILLUMINANT COMPARISON Marcel P. Lucassen, Theo Gevers, Arjan Gijsenij Intelligent Systems Lab Amsterdam, University of Amsterdam ABSTRACT Performance

More information

Chapter 11 Arc Extraction and Segmentation

Chapter 11 Arc Extraction and Segmentation Chapter 11 Arc Extraction and Segmentation 11.1 Introduction edge detection: labels each pixel as edge or no edge additional properties of edge: direction, gradient magnitude, contrast edge grouping: edge

More information

Introduction to visual computation and the primate visual system

Introduction to visual computation and the primate visual system Introduction to visual computation and the primate visual system Problems in vision Basic facts about the visual system Mathematical models for early vision Marr s computational philosophy and proposal

More information

Introducing Robotics Vision System to a Manufacturing Robotics Course

Introducing Robotics Vision System to a Manufacturing Robotics Course Paper ID #16241 Introducing Robotics Vision System to a Manufacturing Robotics Course Dr. Yuqiu You, Ohio University c American Society for Engineering Education, 2016 Introducing Robotics Vision System

More information

Feature descriptors. Alain Pagani Prof. Didier Stricker. Computer Vision: Object and People Tracking

Feature descriptors. Alain Pagani Prof. Didier Stricker. Computer Vision: Object and People Tracking Feature descriptors Alain Pagani Prof. Didier Stricker Computer Vision: Object and People Tracking 1 Overview Previous lectures: Feature extraction Today: Gradiant/edge Points (Kanade-Tomasi + Harris)

More information

Recap: Features and filters. Recap: Grouping & fitting. Now: Multiple views 10/29/2008. Epipolar geometry & stereo vision. Why multiple views?

Recap: Features and filters. Recap: Grouping & fitting. Now: Multiple views 10/29/2008. Epipolar geometry & stereo vision. Why multiple views? Recap: Features and filters Epipolar geometry & stereo vision Tuesday, Oct 21 Kristen Grauman UT-Austin Transforming and describing images; textures, colors, edges Recap: Grouping & fitting Now: Multiple

More information

Fast Local Mapping to Support Navigation and Object. Localization. R. James Firby David Christianson Tom McDougal. University of Chicago

Fast Local Mapping to Support Navigation and Object. Localization. R. James Firby David Christianson Tom McDougal. University of Chicago Fast Local Mapping to Support Navigation and Object Localization R. James Firby David Christianson Tom McDougal Articial Intelligence Laboratory Computer Science Department University of Chicago 1100 East

More information

Towards Autonomous Vision Self-Calibration for Soccer Robots

Towards Autonomous Vision Self-Calibration for Soccer Robots Towards Autonomous Vision Self-Calibration for Soccer Robots Gerd Mayer Hans Utz Gerhard Kraetzschmar University of Ulm, James-Franck-Ring, 89069 Ulm, Germany Abstract The ability to autonomously adapt

More information

Stereo. Outline. Multiple views 3/29/2017. Thurs Mar 30 Kristen Grauman UT Austin. Multi-view geometry, matching, invariant features, stereo vision

Stereo. Outline. Multiple views 3/29/2017. Thurs Mar 30 Kristen Grauman UT Austin. Multi-view geometry, matching, invariant features, stereo vision Stereo Thurs Mar 30 Kristen Grauman UT Austin Outline Last time: Human stereopsis Epipolar geometry and the epipolar constraint Case example with parallel optical axes General case with calibrated cameras

More information

2 The MiníMax Principle First consider a simple problem. This problem will address the tradeos involved in a two-objective optimiation problem, where

2 The MiníMax Principle First consider a simple problem. This problem will address the tradeos involved in a two-objective optimiation problem, where Determining the Optimal Weights in Multiple Objective Function Optimiation Michael A. Gennert Alan L. Yuille Department of Computer Science Division of Applied Sciences Worcester Polytechnic Institute

More information

ECE 172A: Introduction to Intelligent Systems: Machine Vision, Fall Midterm Examination

ECE 172A: Introduction to Intelligent Systems: Machine Vision, Fall Midterm Examination ECE 172A: Introduction to Intelligent Systems: Machine Vision, Fall 2008 October 29, 2008 Notes: Midterm Examination This is a closed book and closed notes examination. Please be precise and to the point.

More information

Recognition. Clark F. Olson. Cornell University. work on separate feature sets can be performed in

Recognition. Clark F. Olson. Cornell University. work on separate feature sets can be performed in Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition, pages 907-912, 1996. Connectionist Networks for Feature Indexing and Object Recognition Clark F. Olson Department of Computer

More information

Particle Tracking. For Bulk Material Handling Systems Using DEM Models. By: Jordan Pease

Particle Tracking. For Bulk Material Handling Systems Using DEM Models. By: Jordan Pease Particle Tracking For Bulk Material Handling Systems Using DEM Models By: Jordan Pease Introduction Motivation for project Particle Tracking Application to DEM models Experimental Results Future Work References

More information

Towards the completion of assignment 1

Towards the completion of assignment 1 Towards the completion of assignment 1 What to do for calibration What to do for point matching What to do for tracking What to do for GUI COMPSCI 773 Feature Point Detection Why study feature point detection?

More information

An Efficient Need-Based Vision System in Variable Illumination Environment of Middle Size RoboCup

An Efficient Need-Based Vision System in Variable Illumination Environment of Middle Size RoboCup An Efficient Need-Based Vision System in Variable Illumination Environment of Middle Size RoboCup Mansour Jamzad and Abolfazal Keighobadi Lamjiri Sharif University of Technology Department of Computer

More information

Complex Sensors: Cameras, Visual Sensing. The Robotics Primer (Ch. 9) ECE 497: Introduction to Mobile Robotics -Visual Sensors

Complex Sensors: Cameras, Visual Sensing. The Robotics Primer (Ch. 9) ECE 497: Introduction to Mobile Robotics -Visual Sensors Complex Sensors: Cameras, Visual Sensing The Robotics Primer (Ch. 9) Bring your laptop and robot everyday DO NOT unplug the network cables from the desktop computers or the walls Tuesday s Quiz is on Visual

More information

A Statistical Consistency Check for the Space Carving Algorithm.

A Statistical Consistency Check for the Space Carving Algorithm. A Statistical Consistency Check for the Space Carving Algorithm. A. Broadhurst and R. Cipolla Dept. of Engineering, Univ. of Cambridge, Cambridge, CB2 1PZ aeb29 cipolla @eng.cam.ac.uk Abstract This paper

More information

Adaptive Zoom Distance Measuring System of Camera Based on the Ranging of Binocular Vision

Adaptive Zoom Distance Measuring System of Camera Based on the Ranging of Binocular Vision Adaptive Zoom Distance Measuring System of Camera Based on the Ranging of Binocular Vision Zhiyan Zhang 1, Wei Qian 1, Lei Pan 1 & Yanjun Li 1 1 University of Shanghai for Science and Technology, China

More information

Using temporal seeding to constrain the disparity search range in stereo matching

Using temporal seeding to constrain the disparity search range in stereo matching Using temporal seeding to constrain the disparity search range in stereo matching Thulani Ndhlovu Mobile Intelligent Autonomous Systems CSIR South Africa Email: tndhlovu@csir.co.za Fred Nicolls Department

More information

Using Game Theory for Image Segmentation

Using Game Theory for Image Segmentation Using Game Theory for Image Segmentation Elizabeth Cassell Sumanth Kolar Alex Yakushev 1 Introduction 21st March 2007 The goal of image segmentation, is to distinguish objects from background. Robust segmentation

More information

Detecting and Tracking a Moving Object in a Dynamic Background using Color-Based Optical Flow

Detecting and Tracking a Moving Object in a Dynamic Background using Color-Based Optical Flow www.ijarcet.org 1758 International Journal of Advanced Research in Computer Engineering & Technology (IJARCET) Detecting and Tracking a Moving Object in a Dynamic Background using Color-Based Optical Flow

More information

Transactions on Information and Communications Technologies vol 16, 1996 WIT Press, ISSN

Transactions on Information and Communications Technologies vol 16, 1996 WIT Press,   ISSN ransactions on Information and Communications echnologies vol 6, 996 WI Press, www.witpress.com, ISSN 743-357 Obstacle detection using stereo without correspondence L. X. Zhou & W. K. Gu Institute of Information

More information

CONCLUSION ACKNOWLEDGMENTS REFERENCES

CONCLUSION ACKNOWLEDGMENTS REFERENCES tion method produces commands that are suggested by at least one behavior. The disadvantage of the chosen arbitration approach is that it is computationally expensive. We are currently investigating methods

More information

2 ATTILA FAZEKAS The tracking model of the robot car The schematic picture of the robot car can be seen on Fig.1. Figure 1. The main controlling task

2 ATTILA FAZEKAS The tracking model of the robot car The schematic picture of the robot car can be seen on Fig.1. Figure 1. The main controlling task NEW OPTICAL TRACKING METHODS FOR ROBOT CARS Attila Fazekas Debrecen Abstract. In this paper new methods are proposed for intelligent optical tracking of robot cars the important tools of CIM (Computer

More information

Hand-Eye Calibration from Image Derivatives

Hand-Eye Calibration from Image Derivatives Hand-Eye Calibration from Image Derivatives Abstract In this paper it is shown how to perform hand-eye calibration using only the normal flow field and knowledge about the motion of the hand. The proposed

More information

Recognizing Apples by Piecing Together the Segmentation Puzzle

Recognizing Apples by Piecing Together the Segmentation Puzzle Recognizing Apples by Piecing Together the Segmentation Puzzle Kyle Wilshusen 1 and Stephen Nuske 2 Abstract This paper presents a system that can provide yield estimates in apple orchards. This is done

More information

Fully Automatic Endoscope Calibration for Intraoperative Use

Fully Automatic Endoscope Calibration for Intraoperative Use Fully Automatic Endoscope Calibration for Intraoperative Use Christian Wengert, Mireille Reeff, Philippe C. Cattin, Gábor Székely Computer Vision Laboratory, ETH Zurich, 8092 Zurich, Switzerland {wengert,

More information

Object Recognition in Living Creatures. Object Recognition. Goals of Object Recognition. Object Recognition with Computers. Object Recognition Issues

Object Recognition in Living Creatures. Object Recognition. Goals of Object Recognition. Object Recognition with Computers. Object Recognition Issues Object Recognition Object Recognition in Living Creatures Most important aspect of visual perception Least understood Young children can recognize large variety of objects Child can generalize from a few

More information

Moving Object Tracking in Video Using MATLAB

Moving Object Tracking in Video Using MATLAB Moving Object Tracking in Video Using MATLAB Bhavana C. Bendale, Prof. Anil R. Karwankar Abstract In this paper a method is described for tracking moving objects from a sequence of video frame. This method

More information

Document Image Restoration Using Binary Morphological Filters. Jisheng Liang, Robert M. Haralick. Seattle, Washington Ihsin T.

Document Image Restoration Using Binary Morphological Filters. Jisheng Liang, Robert M. Haralick. Seattle, Washington Ihsin T. Document Image Restoration Using Binary Morphological Filters Jisheng Liang, Robert M. Haralick University of Washington, Department of Electrical Engineering Seattle, Washington 98195 Ihsin T. Phillips

More information

Detecting Salient Contours Using Orientation Energy Distribution. Part I: Thresholding Based on. Response Distribution

Detecting Salient Contours Using Orientation Energy Distribution. Part I: Thresholding Based on. Response Distribution Detecting Salient Contours Using Orientation Energy Distribution The Problem: How Does the Visual System Detect Salient Contours? CPSC 636 Slide12, Spring 212 Yoonsuck Choe Co-work with S. Sarma and H.-C.

More information

Vision-based Mobile Robot Localization and Mapping using Scale-Invariant Features

Vision-based Mobile Robot Localization and Mapping using Scale-Invariant Features Vision-based Mobile Robot Localization and Mapping using Scale-Invariant Features Stephen Se, David Lowe, Jim Little Department of Computer Science University of British Columbia Presented by Adam Bickett

More information

COS Lecture 10 Autonomous Robot Navigation

COS Lecture 10 Autonomous Robot Navigation COS 495 - Lecture 10 Autonomous Robot Navigation Instructor: Chris Clark Semester: Fall 2011 1 Figures courtesy of Siegwart & Nourbakhsh Control Structure Prior Knowledge Operator Commands Localization

More information

Motion and Tracking. Andrea Torsello DAIS Università Ca Foscari via Torino 155, Mestre (VE)

Motion and Tracking. Andrea Torsello DAIS Università Ca Foscari via Torino 155, Mestre (VE) Motion and Tracking Andrea Torsello DAIS Università Ca Foscari via Torino 155, 30172 Mestre (VE) Motion Segmentation Segment the video into multiple coherently moving objects Motion and Perceptual Organization

More information

PERSON FOLLOWING AND MOBILE ROBOT VIA ARM-POSTURE DRIVING USING COLOR AND STEREOVISION. Bogdan Kwolek

PERSON FOLLOWING AND MOBILE ROBOT VIA ARM-POSTURE DRIVING USING COLOR AND STEREOVISION. Bogdan Kwolek PERSON FOLLOWING AND MOBILE ROBOT VIA ARM-POSTURE DRIVING USING COLOR AND STEREOVISION Bogdan Kwolek Rzeszów University of Technology, W. Pola 2, 35-959 Rzeszów, Poland Abstract: This paper describes an

More information

Colorado School of Mines. Computer Vision. Professor William Hoff Dept of Electrical Engineering &Computer Science.

Colorado School of Mines. Computer Vision. Professor William Hoff Dept of Electrical Engineering &Computer Science. Professor William Hoff Dept of Electrical Engineering &Computer Science http://inside.mines.edu/~whoff/ 1 Introduction to 2 What is? A process that produces from images of the external world a description

More information

Backpack: Detection of People Carrying Objects Using Silhouettes

Backpack: Detection of People Carrying Objects Using Silhouettes Backpack: Detection of People Carrying Objects Using Silhouettes Ismail Haritaoglu, Ross Cutler, David Harwood and Larry S. Davis Computer Vision Laboratory University of Maryland, College Park, MD 2742

More information

People detection and tracking using stereo vision and color

People detection and tracking using stereo vision and color People detection and tracking using stereo vision and color Rafael Munoz-Salinas, Eugenio Aguirre, Miguel Garcia-Silvente. In Image and Vision Computing Volume 25 Issue 6 (2007) 995-1007. Presented by

More information

HISTOGRAMS OF ORIENTATIO N GRADIENTS

HISTOGRAMS OF ORIENTATIO N GRADIENTS HISTOGRAMS OF ORIENTATIO N GRADIENTS Histograms of Orientation Gradients Objective: object recognition Basic idea Local shape information often well described by the distribution of intensity gradients

More information

Coarse-to-fine image registration

Coarse-to-fine image registration Today we will look at a few important topics in scale space in computer vision, in particular, coarseto-fine approaches, and the SIFT feature descriptor. I will present only the main ideas here to give

More information

Stereo Wrap + Motion. Computer Vision I. CSE252A Lecture 17

Stereo Wrap + Motion. Computer Vision I. CSE252A Lecture 17 Stereo Wrap + Motion CSE252A Lecture 17 Some Issues Ambiguity Window size Window shape Lighting Half occluded regions Problem of Occlusion Stereo Constraints CONSTRAINT BRIEF DESCRIPTION 1-D Epipolar Search

More information