M-SIFT: A new method for Vehicle Logo Recognition

Size: px
Start display at page:

Download "M-SIFT: A new method for Vehicle Logo Recognition"

Transcription

1 01 IEEE International Conference on Vehicular Electronics an Safety July 4-7, 01. Istanbul, Turkey M-SIFT: A new metho for Vehicle Logo Recognition Apostolos Psyllos, Christos-Nikolaos Anagnostopoulos, Eleftherios Kayafas Abstract - In this paper, a new algorithm for Vehicle Logo Recognition is propose, on the basis of an enhance Scale Invariant Feature Transform (Merge-SIFT or M-SIFT). The algorithm is assesse on a set of 1500 logo images that belong to 10 istinctive vehicle manufacturers. A series of experiments are conucte, splitting the 1500 images to a training set (atabase) an to a testing set (uery). It is shown that the M- SIFT approach, which is propose in this paper, boosts the recognition accuracy compare to the stanar SIFT metho. The reporte results inicate an average of 94.6% true recognition rate in vehicle logos, while the processing time remains low (~0.8sec). M I. INTRODUCTION any vision-base Intelligent Transport Systems for etecting, tracking or recognizing vehicles in image seuences are referre in the literature. Vehicle type classification is a task that has been aeuately aresse in the literature [1-6]. However, Vehicle Manufacturer Recognition (VMR) is a crucial subject, with relatively limite research reporte in the fiel. This is ue to the fact that manufacturer recognition is an inherently har problem ue to the wie variety in the appearance of vehicles. This task becomes even more ifficult since image acuisition is performe in outoor environment where illumination is ambient. Scale Invariant Feature Transforms (SIFT) were introuce by Lowe [7-9] an they are invariant to rotation, translation an scale variation between images an partially invariant to affine istortion, illumination variance an noise. Research relate to fully invariant features has been publishe by Brown an Lowe [10], Mikolajczyk an Schmit [11]. Image matching is a funamental problem in computer vision which occurs in many computer vision applications regaring a variety of fiels, incluing image retrieval for security enforcement an robot navigation. A common approach to accurate image matching is known as the invariant keypoint or the extraction of the point of interest from the images compare. It involves ientifying Manuscript receive June 1, 01. A. Psyllos is with NTUA, Sc. of Applie Mathematical an Physical Science, Athens, Greece ( psyllos@central.ntua.gr). C. N. Anagnostopoulos is with University of the Aegean, Mytilene, Greece. (Corresponing author: ; fax: ; e- mail: canag@ct.aegean.gr E. Kayafas is with NTUA, Sc. of Electrical & Computer Engineering, Athens, Greece ( kayafas@cs.ntua.gr). the points that can be reliably extracte from ifferent images of the same object or the same category of objects. Earlier research into invariant keypoints inclues the Harris corner etector [1] an the keypoints invariant to rotation an translation [13], [14]. Dlangenkov an Belongie [15] utilize Scale Invariant Feature Transform (SIFT) features, using a vehicle atabase of rear-view vehicle images. Čonos [16] eals with a vehicle moel recognition problem of frontal view images, proposing a SIFT-base escriptor for feature extraction. The rawback of the latter approach is the extensive computational cost reporte in the respective paper. Petrovic an Cootes [17] presente an interesting approach base on ege graient an match refinement algorithm for vehicle moel recognition an verification. A comparative knowlege acuisition system, consisting of several object recognition moules for the rear view of vehicles appears in Maemoto et al. [18]. The VMR problem is resolve through vehicle logo recognition (VLR) an the recognition task reuires the successful extraction of the small logo area from the original vehicle image. Usually, the process involves a license plate location moule, followe by coarse to fine methos to ientify the logo area using symmetry an/or ege statistics in the image. Then logo recognition is either performe through neural networks [19-1] or template matching. Wang et al. [3] presente a metho for logo recognition using template matching an ege orientation histograms with goo results. In this paper, a SIFT-base Vehicle Logo Recognition schema is escribe, whose aim is to obtain reliable Vehicle Make Recognition from the logo image. An enhance SIFT matching moule etects an extracts the points of interest (keypoints) in the uery logo image, escribes them an matches them with keypoints store in a logo image atabase. The whole process is fine-tune by clustering the matche keypoints, using Generalise Hough Transform [4], an geometric verification by means of an affine transformation. The propose metho is assesse on a atabase containing 1500 logo images, which were processe using our previous work escribe in [1], [], [5] using the Meialab Vehicle Database [6]. II. VEHICLE LOGO RECOGNITION SIFT is the state-of-the-art in the fiel of image recognition an is use in a wie range of image retrieval /1/$ IEEE 61

2 applications. It is base on the iea of representing images by a set of invariant keypoint escriptors using graient orientation histograms. The invariant features (points of interest), or just keypoints, are istinguishe as local peaks in the image scale-space representation an filtere to preserve only those that are likely to remain stable over transformations. For each invariant feature a SIFT keypoint escriptor is calculate having the following properties: a) smooth changes in location, orientation an scale o not cause raical changes in the feature vector, b) it is fairly compact, expressing an orientation histogram from a patch of pixels in the neighborhoo of the invariant feature, using a 18 imensional vector, c) it is resilient to affine eformations, such as those cause by weak perspective effects (for small angles of eviation), an thus it coul be efficient for logo recognition in non-controlle conitions. In this paper, in the initial stage of computation, scale an orientation invariant features are etecte an extracte from the uery image, exploring the scale-space structure of the image as escribe in [7] an [7]. In the next sections, our contribution in eveloping an enhance Scale Invariant Feature Transform (Merge-SIFT or M-SIFT) is escribe, pointing out the feature matching an hypothesis valiation proceures in the caniate points. A. Feature Matching For each feature i in the uery image, the escriptor is use to fin its Nearest-Neighbor (NN) matches among all the features store from all the images in a atabase calle NN atabase (see Figure 1). The nearest neighbors are ientifie minimizing an L Eucliean metric: N Qi D i(j) = (k k ) (1) k= 1 where Q i (i.e. 1,..., N ) is the i th escriptor for the uery image, D i (j) (i.e D 1,D...,D N ) is the i th escriptor for image j an N=18. The total number of nearest neighbors is calculate from Euation (1), where carinality is the number of elements in a set an γ is an appropriate threshol value. NN i= carinality( arg Qi D i(j) < ) () Therefore, the number of NNs in the atabase for each feature (keypoint) epens on the threshol selection. In Figure, the parameter τ is plotte versus threshol (γ), where τ is escribe by the euation (3): τ = NNi KP KP where NN i is the number of nearest neighbors foun from Euation () for each feature i in the uery image, KP is the number of keypoints in the uery image an KP b is the total number of keypoints in NN atabase. b (3) B. Hypothesis Valiation 1) Feature Clustering A typical logo image contains approximately 100 features, numerous of which are not important to the matching proceure. This is ue to the fact that some features correspon to ambient illumination reflections, shaows or noise. To maximize the performance of VLR for small, illcapture or noise logos, logo ientification shoul occur with the fewest possible feature matches. It was foun that it is possible to have reliable recognition with as few as 3 feature matches. This step etermines the reasonable NN matches foun in the previous step an verifies whether the uery image represents a logo image store in the atabase. Nearest Neighbor features are clustere using Generalise Hough Transform (GHT) [4], [8], in orer to efine the parameters for a similarity transformation between the uery an the atabase features. Then, an approximate affine transformation follows, which maps each NN from the uery image onto the matche NN feature from the atabase image. Fig. 1. Keypoint Nearest - Neighbour Matching Scheme. For every invariant feature in the atabase there is a recor that contains the image from which the feature comes from as well as the feature parameters position, orientation, scale an the 18-imensional escriptor. Query Q an atabase D feature vector similarity transformation geometry is shown in Figure 3. In this proceure, λ is a normalization coefficient in orer to achieve a reasonable size for the vector length compare to the image size (efault λ=10). 6

3 threshol, this pair participates in a Hough similarity transformation an a vote is ae to the corresponing cell in the Hough array. The atabase logo image with the highest number of votes is consiere to be the most similar with the uery image an it is forware to the geometrical consistency check with affine transformation (homography). Fig.. Reuce Nearest Neighbour parameter (τ) plotte versus istance threshol (γ). ' ' (x, )=(x + λscos θ, + λssinθ ) (4) ' ' (x, y )=(x + λs cos θ, y + λs sinθ ) (5) In orer to be consistent with a similarity transformation, points (x, y ), (x, y ), (x, y ), (x, y ) in Euations (4) an (5) shoul satisfy the matrix relationship (homography) shown in Euation (6). x 1 0 cos x x 0 1 sin y (6) x 1 0 Tx x x 0 1 Ty y where Λ is the scale change, φ is the rotation angle an T x, T y are the translation values of the similarity transformation. GHT ientifies clusters of features with a consistent interpretation by using each feature to vote for all the logo images that are consistent with the feature. Therefore, a four imensioznal hash table (accumulator) entry can be create, preicting T x, T y, Λ an φ parameters from the match hypothesis, through the similarity transformation, escribe in Euation (6). T x an T y get values between [-a, +a] an [-b +b], where axb is the image imension. In aition, Λ an φ get values in [-4, +4] an [-π,+π] ras, respectively. Bins size of Δφ= π/6 ras, ΔΛ=, ΔTx =a/4 an Δ Ty =b/4 are use for orientation, scale an translations, respectively. The above bin sizes are rather broa an allow clustering even in the case of substantial geometric istortion, ue to a change in 3D point of view. This preiction has large error bouns, as the similarity transformation implie by these 4 parameters is only an approximation to the full 6 egrees-of-freeom posespace for a 3D object an also it oes not account for any nonrigi eformations. When clusters of features are foun to vote for the same pose of a vehicle logo, the probability of the interpretation being correct is much higher than for any single feature. Therefore, when the NN istance between a pair of uery an atabase feature is less than a certain ) Geometrical Verification In this step, the locations of the keypoints in the uery an the atabase image are verifie for geometrical consistency. Those keypoints that fit well to a homography transformation are calle inliers, while those that are inconsistent to it are calle outliers. A keypoint is consiere to be an inlier or outlier accoring to its conformity to an affine geometrical transformation, which takes into account small ifferences in translation, rotation an scaling between the two images. Euation (7) efines the affine transformation for a set of uery an ata points x x a x 0 1 y a x x a 3 (7) 0 0 x 0 1 y a V... x N N N x x N N Vy N 0 0 x 0 1 x In the above euation, a i (i = 1,, 3, 4) are the warping parameters an V x, V y are the translation parameters of the affine transformation. If Euation (7) is consiere as a system of A*b=c, then solving the respective euations, there is a least suare solution for the parameters b as seen in Euation (8). This euation minimizes the sum of the suare istances from the uery to the atabase image keypoint positions. T 1 T b = [A A] A c (18) Fig. 3. Query an moel keypoint vector similarity transformation geometry. 63

4 In orer to select which caniate points shoul be use for affine transformation, the RANSAC metho [9] is applie to a pair of images by ranomly choosing three pairs of matche points of interest, since this is the minimum reuire for an affine transformation. These three pairs of points of interest are fitte to an affine transformation, which is then applie to the rest of the points of interest (keypoints) of the two images. Inliers are ientifie by selecting those feature points in one image that are within a certain threshol istance from its matche feature of the other image an minimizing the fitting error. This process was repeate until the inlier count reaches a maximum or the fitting error is below a fixe value. The transformation with the highest inlier count is labele as the best one to map the uery image onto the image of the atabase. The total fitting error, ρ, between uery an logo atabase images uner affine transformation is: Ab c ρ = (9) w where w are the number of keypoints in matrix A, so if ρ in e. (9) excees a threshol, we reject the respective match. Here we have use a threshol of value ρ = 0.05*L x where L x is the maximum imension of the uery image. The above proceure was repeate, for all clusters foun in the Generalise Hough Transform an the atabase image with the maximum number of inliers was selecte. When two or more matches have the same number of inlier points, in orer to choose the best match, the atabase image with the minimum number of outliers is selecte. However, if the maximum clusters size is less than three points, the unconitional match hypothesis is aopte. In this case, the moel with the maximum value of feature matches in the uery image ecies for the logo recognition. The whole process for feature etection, matching an hypothesis valiation is shown in Figure 4, while an inicative example of the NN matches is given in Figure 5. C. Experimental Part 1) Database set The propose metho is assesse on a set of 1500 logo images that have been successfully segmente using our previous work escribe in [1], [] using the Meialab LPR Database [6]. The Meialab LPR Database contains frontal views of vehicles, both stationary an in motion, capture by a Nikon Coolpix 885 ajuste to acuire 104x768 pixel images. From the total of 1500 logo images, 400 have been selecte to form the logo atabase as in []. These images correspon to 10 classes of selecte vehicle manufacturers. Each class contains 40 images an for each one, the keypoints have been etecte an the escriptors have been store. Then, a reference image was selectively chosen by an expert an the remaining 39 samples were registere accoring to that reference view using the homography calculate by the RANSAC algorithm. Only areas belonging to the common parts of the images were selecte an the keypoint escriptors were re-reference to the new position, scale an orientation. In this way, the number of keypoints for every manufacturer logo is substantially increase by merging the keypoints, thus making the recognition process more robust in illumination conitions. Finally, a atabase containing merge keypoints for every manufacturer is create. Fig.4. SIFT Match Hypothesis Testing Flow Chart. ) Query set The remaining 1100 logo images form the uery set for our experiments. For every image, the keypoints an their respective escriptors have been calculate following the SIFT proceure. Then, these escriptors were matche against the escriptors of the atabase set, with the parameter γ set to γ=1. These values correspon to the turning point in the curve shown in Figure. For Eucliean NN search, a KD-Tree ata structure is employe with search time complexity shown in Euation 10. k 1 ( ) O(kN k ) (10) where k is the tree imension an N the number of tree noes [30],[31]. For the experiments in this paper, the typical values are: k = 18 an N ~ (400 atabase images with approximately 100 keypoints per image). The image in the atabase whose escriptors get the maximum 64

5 number of votes in Hough Transformation array, is consiere to be the one most similar to the uery image an is then checke with RANSAC for geometric consistency. The recognition statistics are given analytically in Table I. There is an overall 94.6% correct recognition an 5.4% false recognition. The number of Hough-clustere keypoints is approximately 6% of the total keypoints etecte, while the number of inliers is about 50% of the Hough-clustere, as presente in Table II. The recognition spee is shown for comparison with stanar SIFT an M-SIFT, in Table III. A correct match occurs when the two keypoints correspon to the same physical location an a false match when the two keypoints come from ifferent physical locations. The total number of corresponences for the given ataset is known a priori an has been calculate by estimating the homography between each pair of images, for an affine transformation. Total corresponences are the total groun truth matches which are greater that the total matches (false + true) foun. Using 1-Precision versus Recall, we can evaluate the total recognition performance, for a range of experimental conitions, varying the threshol value γ, for NN matching. The results from these experiments using the stanar SIFT features an the enhance (merge) ones, are plotte together for comparison in Figure 6. every moel was substantially increase, thus making the recognition process more robust in outoor conitions, such as low-contrast, inaeuate lighting, reflections, clouy weather etc. TABLE I M-SIFT RECOGNITION RATE Manufacturer True False Alfa Romeo Aui Bmw Citroen Fiat 10 8 Peugeot Renault Seat Toyota Volkswagen Total Average (%) 94.6% 5.4% TABLE II KP SIZE DETECTED VS. HOUGH CLUSTER SIZE & RANSAC INLIERS Query Image Total Hough RANSAC NN Cluster Size Inliers Alfa Romeo Aui BMW Citroen Fiat Peugeot Renault Seat Toyota VW 1 10 Total TABLE III VEHICLE LOGO RECOGNITION RATE Methoology Total Recognition Time (ms) SIFT 850 M-SIFT 100 Fig. 5. Database Moel NN matches for a sample uery image. III. CONCLUSIONS This paper provies a escription of a practical application for vehicle make recognition, base on an extening well known image processing routines, such as SIFT. Specifically, a moification of SIFT has been use, where the keypoints from similar objects were merge after homography transformation to the same coorinates an scales. With this techniue, the number of keypoints for Fig. 6. Logo recognition performance for merge SIFT features an stanar SIFT, respectively. 1-Precision an Recall calculate with various threshols for NN matching. 65

6 The logo recognition system emonstrate goo performance, yieling approximately 95% recognition rate, when applie to a atabase of alreay segmente logo images. One thousan an one hunre (1100) uery logo images were processe with the enhance SIFT-base approach, which gives better recognition accuracy compare to the stanar SIFT proceure. The propose experiments performs better compare to our previous work, in which a Probabilistic Neural Network (PNN) was use [0], [1] with 87% true recognition rate an extens the work escribe in []. The recognition spee is rather fast (~1 sec), which is suitable for real-time applications. To further boost performance an robustness, there is an ongoing research to combine logo recognition with vehicle moel recognition (e.g. Aui A3). A short time vieo that emonstrates some preliminary results in real-time execution can be ownloae from [33]. Moreover, another extension of this work might be the possibility of ealing with a wier range of viewpoints or 3-D recognition, as well as recognition of more complex scenes incluing many vehicles an uner a wier variety of illumination conitions. REFERENCES [1] Weber M., Welling M., Perona P. Unsupervise Learning of Moels for Recognition, Lecture Notes in Computer Science 184, Springer- Verlag, pp. 18 3, 000. [] Kato T., Ninomiya Y., an Masaki I., Preceing vehicle recognition base on learning from sample images, IEEE Transactions on Intelligent Transportation Systems, vol. 3(4), pp. 5 60, 00. [3] Lai A.H.S., Yung N.H.C., Vehicle-type ientification through automate virtual loop assignment an block-base irection-biase motion estimation, IEEE Trans. Intelligent Transportation Systems, 1(), pp , 000. [4] Lai A.H.S., Fung G. S. K. an Yung N.H.C., Vehicle Type Classification from Visual-Base Dimension Estimation, 001 IEEE Intelligent Transportation Systems Conference, Oaklan (CA), USA, pp , 001. [5] S. Gupte, O. Masou, R. F. K. Martin, N. P. Papanikolopoulos, Detection an Classification of Vehicles, IEEE Trans. on Intelligent Transportation Systems, Vol. 3, No. 1, March 00, pp [6] J. W. Hsieh, S. H. Yu, Y. S. Chen an W.F. Hu, Automatic traffic surveillance system for vehicle tracking an classification, IEEE Trans. on Intelligent Transportation Systems, Vol. 7, No., June 006, pp [7] Lowe D., Object recognition from local scale-invariant features, International Conference on Computer Vision, Corfu, Greece (September 1999), pp [8] Lowe D.,, Local feature view clustering for 3D object recognition, 001 IEEE Computer Society Conference on Computer Vision an Pattern Recognition (CVPR'01), vol. 1, pp , 001. [9] Lowe D., Distinctive image features from scale-invariant keypoints, International Journal of Computer Vision, vol. (60), pp , 004. [10] Brown M. an Lowe D., Invariant features from interest point groups, In British Machine Vision Conference, BMVC 00, Cariff, Wales (September 00), pp [11] Mikolajczyk K. an Schmi C. An Affine Invariant Interest Point Detector, European Conference on Computer Vision, vol 1, pp , 00. [1] Harris C. an Stephens M., A combine corner an ege etector, Fourth Alvey Vision Conference, Manchester, UK, pp , [13] Siggelkow S., Feature Histograms for Content-Base Image Retrieval, PhD Thesis, Albert-Luwigs-University Frieiburg, December 00. [14] Schulz-Mirbach H., Invariant Features for Gray Scale Images, DAGM Symposium I, pp.1-14,1995. [15] Dlagnekov L., Belongie S., Recognizing Cars, University of California San Diego, Tech. Rep. CS , 005. [16] Čonos M., Recognition of vehicle make from a frontal view, Diploma Thesis, Faculty of Electrical Engineering, Czech Technical University, 006. [17] Petrovic V. S. an Cootes T. F., "Analysis of Features for Rigi Structure Vehicle Type Recognition", British Machine Vision Conference, vol., pp , 004. [18] Maemoto H., Okuma S., Yano Y., Parametric Vehicle Recognition Using Knowlege Acuisition System, IEEE International Conference on Systems, Man an Cybernetics, vol. 4, pp , 004. [19] Lee H. J., Neural Network Approach to Ientify Moel of Vehicles, ISNN 006, Lecture Notes in Computer Science 3973, 006, Springer-Verlag, pp [0] Psyllos A., Anagnostopoulos C.N., Kayafas E., Loumos V. Image Processing & Artificial Neural Networks for Vehicle Make an Moel Recognition, 10th International Conference on Applications of Avance Technologies in Transportation, May 8-30, Athens, Greece, 008. [1] Psyllos A., Anagnostopoulos C.,N., Kayafas E., Vehicle Authentication from Digital Image Measurements, 16th IMEKO TC4 Symposium & 13th Workshop on ADC Moeling an Testing Florence, Italy, September - 4, 008. [] Psyllos A., Anagnostopoulos C.,N., Kayafas, Vehicle Logo Recognition using a SIFT-base Enhance Matching Scheme IEEE Trans on Intelligent Transportation Systems, vol. 11, no., 010, pp [3] Wang Y., Liu Z., Xiao F., A fast coarse-to-fine vehicle logo etection an recognition metho, IEEE International Conference on Robotics an Biomimetics, pp , Sanya, China, Dec. 007 [4] Ballar, D., Generalizing the Hough transform to etect arbitrary patterns, Pattern Recognition, 3( ), pp ,1981. [5] Anagnostopoulos C., Anagnostopoulos I., Loumos V., Kayafas E. (006), "A license plate recognition algorithm for intelligent transportation system applications", IEEE Transactions on Intelligent Transportation Systems, 7 (3), [6] The Meialab License Plate Recognition Database: meialab.ntua.gr/research/lpratabase.html last accesse March 01. [7] Lineberg T., Scale-space theory: A basic tool for analysing structures at ifferent scales, Journal of Applie Statistics, 1(), pp [8] Grimson, E., Object Recognition by Computer: The Role of Geometric Constraints, The MIT Press: Cambrige, MA, [9] Fischler M., Bolles R., Ranom Sample Consensus: A Paraigm for Moel Fitting with Applications to Image Analysis an Automate Cartography, Comm. of the ACM, vol. (4), pp , [30] Arya S., Mount D., Approximate Nearest Neighbor Searching, Proceeings 4th Ann. ACM-SIAM Symposium on Discrete Algorithms} (SODA), pp , [31] Frieman J., Bentley J., Finkel, R., An algorithm for fining best matches in logarithmic expecte time, ACM Transactions on Mathematics Software,vol. 3(3), pp. 09 6, [3] Y. Ke, R., Sukthankar, PCA-SIFT: A More Distinctive Representation for Local Image Descriptors, Conference on Computer Vision an Pattern Recognition, vol. pp , 004. [33] Vieo Demo: semo.mpg, last accesse March

Vehicle Logo Recognition using Image Matching and Textural Features

Vehicle Logo Recognition using Image Matching and Textural Features Vehicle Logo Recognition using Image Matching and Textural Features Nacer Farajzadeh Faculty of IT and Computer Engineering Azarbaijan Shahid Madani University Tabriz, Iran n.farajzadeh@azaruniv.edu Negin

More information

Shift-map Image Registration

Shift-map Image Registration Shift-map Image Registration Svärm, Linus; Stranmark, Petter Unpublishe: 2010-01-01 Link to publication Citation for publishe version (APA): Svärm, L., & Stranmark, P. (2010). Shift-map Image Registration.

More information

NEW METHOD FOR FINDING A REFERENCE POINT IN FINGERPRINT IMAGES WITH THE USE OF THE IPAN99 ALGORITHM 1. INTRODUCTION 2.

NEW METHOD FOR FINDING A REFERENCE POINT IN FINGERPRINT IMAGES WITH THE USE OF THE IPAN99 ALGORITHM 1. INTRODUCTION 2. JOURNAL OF MEDICAL INFORMATICS & TECHNOLOGIES Vol. 13/009, ISSN 164-6037 Krzysztof WRÓBEL, Rafał DOROZ * fingerprint, reference point, IPAN99 NEW METHOD FOR FINDING A REFERENCE POINT IN FINGERPRINT IMAGES

More information

Shift-map Image Registration

Shift-map Image Registration Shift-map Image Registration Linus Svärm Petter Stranmark Centre for Mathematical Sciences, Lun University {linus,petter}@maths.lth.se Abstract Shift-map image processing is a new framework base on energy

More information

Object Recognition Using Colour, Shape and Affine Invariant Ratios

Object Recognition Using Colour, Shape and Affine Invariant Ratios Object Recognition Using Colour, Shape an Affine Invariant Ratios Paul A. Walcott Centre for Information Engineering City University, Lonon EC1V 0HB, Englan P.A.Walcott@city.ac.uk Abstract This paper escribes

More information

Online Appendix to: Generalizing Database Forensics

Online Appendix to: Generalizing Database Forensics Online Appenix to: Generalizing Database Forensics KYRIACOS E. PAVLOU an RICHARD T. SNODGRASS, University of Arizona This appenix presents a step-by-step iscussion of the forensic analysis protocol that

More information

Classifying Facial Expression with Radial Basis Function Networks, using Gradient Descent and K-means

Classifying Facial Expression with Radial Basis Function Networks, using Gradient Descent and K-means Classifying Facial Expression with Raial Basis Function Networks, using Graient Descent an K-means Neil Allrin Department of Computer Science University of California, San Diego La Jolla, CA 9237 nallrin@cs.ucs.eu

More information

Multi-camera tracking algorithm study based on information fusion

Multi-camera tracking algorithm study based on information fusion International Conference on Avance Electronic Science an Technolog (AEST 016) Multi-camera tracking algorithm stu base on information fusion a Guoqiang Wang, Shangfu Li an Xue Wen School of Electronic

More information

Comparison of Methods for Increasing the Performance of a DUA Computation

Comparison of Methods for Increasing the Performance of a DUA Computation Comparison of Methos for Increasing the Performance of a DUA Computation Michael Behrisch, Daniel Krajzewicz, Peter Wagner an Yun-Pang Wang Institute of Transportation Systems, German Aerospace Center,

More information

Object Recognition with Invariant Features

Object Recognition with Invariant Features Object Recognition with Invariant Features Definition: Identify objects or scenes and determine their pose and model parameters Applications Industrial automation and inspection Mobile robots, toys, user

More information

Fast Window Based Stereo Matching for 3D Scene Reconstruction

Fast Window Based Stereo Matching for 3D Scene Reconstruction The International Arab Journal of Information Technology, Vol. 0, No. 3, May 203 209 Fast Winow Base Stereo Matching for 3D Scene Reconstruction Mohamma Mozammel Chowhury an Mohamma AL-Amin Bhuiyan Department

More information

THE BAYESIAN RECEIVER OPERATING CHARACTERISTIC CURVE AN EFFECTIVE APPROACH TO EVALUATE THE IDS PERFORMANCE

THE BAYESIAN RECEIVER OPERATING CHARACTERISTIC CURVE AN EFFECTIVE APPROACH TO EVALUATE THE IDS PERFORMANCE БСУ Международна конференция - 2 THE BAYESIAN RECEIVER OPERATING CHARACTERISTIC CURVE AN EFFECTIVE APPROACH TO EVALUATE THE IDS PERFORMANCE Evgeniya Nikolova, Veselina Jecheva Burgas Free University Abstract:

More information

CAMERAS AND GRAVITY: ESTIMATING PLANAR OBJECT ORIENTATION. Zhaoyin Jia, Andrew Gallagher, Tsuhan Chen

CAMERAS AND GRAVITY: ESTIMATING PLANAR OBJECT ORIENTATION. Zhaoyin Jia, Andrew Gallagher, Tsuhan Chen CAMERAS AND GRAVITY: ESTIMATING PLANAR OBJECT ORIENTATION Zhaoyin Jia, Anrew Gallagher, Tsuhan Chen School of Electrical an Computer Engineering, Cornell University ABSTRACT Photography on a mobile camera

More information

Instance-level recognition

Instance-level recognition Instance-level recognition 1) Local invariant features 2) Matching and recognition with local features 3) Efficient visual search 4) Very large scale indexing Matching of descriptors Matching and 3D reconstruction

More information

A Neural Network Model Based on Graph Matching and Annealing :Application to Hand-Written Digits Recognition

A Neural Network Model Based on Graph Matching and Annealing :Application to Hand-Written Digits Recognition ITERATIOAL JOURAL OF MATHEMATICS AD COMPUTERS I SIMULATIO A eural etwork Moel Base on Graph Matching an Annealing :Application to Han-Written Digits Recognition Kyunghee Lee Abstract We present a neural

More information

A Plane Tracker for AEC-automation Applications

A Plane Tracker for AEC-automation Applications A Plane Tracker for AEC-automation Applications Chen Feng *, an Vineet R. Kamat Department of Civil an Environmental Engineering, University of Michigan, Ann Arbor, USA * Corresponing author (cforrest@umich.eu)

More information

Refinement of scene depth from stereo camera ego-motion parameters

Refinement of scene depth from stereo camera ego-motion parameters Refinement of scene epth from stereo camera ego-motion parameters Piotr Skulimowski, Pawel Strumillo An algorithm for refinement of isparity (epth) map from stereoscopic sequences is propose. The metho

More information

Instance-level recognition

Instance-level recognition Instance-level recognition 1) Local invariant features 2) Matching and recognition with local features 3) Efficient visual search 4) Very large scale indexing Matching of descriptors Matching and 3D reconstruction

More information

Image Segmentation using K-means clustering and Thresholding

Image Segmentation using K-means clustering and Thresholding Image Segmentation using Kmeans clustering an Thresholing Preeti Panwar 1, Girhar Gopal 2, Rakesh Kumar 3 1M.Tech Stuent, Department of Computer Science & Applications, Kurukshetra University, Kurukshetra,

More information

New Geometric Interpretation and Analytic Solution for Quadrilateral Reconstruction

New Geometric Interpretation and Analytic Solution for Quadrilateral Reconstruction New Geometric Interpretation an Analytic Solution for uarilateral Reconstruction Joo-Haeng Lee Convergence Technology Research Lab ETRI Daejeon, 305 777, KOREA Abstract A new geometric framework, calle

More information

Yet Another Parallel Hypothesis Search for Inverse Entailment Hiroyuki Nishiyama and Hayato Ohwada Faculty of Sci. and Tech. Tokyo University of Scien

Yet Another Parallel Hypothesis Search for Inverse Entailment Hiroyuki Nishiyama and Hayato Ohwada Faculty of Sci. and Tech. Tokyo University of Scien Yet Another Parallel Hypothesis Search for Inverse Entailment Hiroyuki Nishiyama an Hayato Ohwaa Faculty of Sci. an Tech. Tokyo University of Science, 2641 Yamazaki, Noa-shi, CHIBA, 278-8510, Japan hiroyuki@rs.noa.tus.ac.jp,

More information

Vision-based Multi-Robot Simultaneous Localization and Mapping

Vision-based Multi-Robot Simultaneous Localization and Mapping Vision-base Multi-Robot Simultaneous Localization an Mapping Hassan Hajjiab an Robert Laganière VIVA Research Lab School of Information Technology an Engineering University of Ottawa, Ottawa, Canaa, K1N

More information

Research Article Research on Law s Mask Texture Analysis System Reliability

Research Article Research on Law s Mask Texture Analysis System Reliability Research Journal of Applie Sciences, Engineering an Technology 7(19): 4002-4007, 2014 DOI:10.19026/rjaset.7.761 ISSN: 2040-7459; e-issn: 2040-7467 2014 Maxwell Scientific Publication Corp. Submitte: November

More information

A PSO Optimized Layered Approach for Parametric Clustering on Weather Dataset

A PSO Optimized Layered Approach for Parametric Clustering on Weather Dataset Vol.3, Issue.1, Jan-Feb. 013 pp-504-508 ISSN: 49-6645 A PSO Optimize Layere Approach for Parametric Clustering on Weather Dataset Shikha Verma, 1 Kiran Jyoti 1 Stuent, Guru Nanak Dev Engineering College

More information

Kinematic Analysis of a Family of 3R Manipulators

Kinematic Analysis of a Family of 3R Manipulators Kinematic Analysis of a Family of R Manipulators Maher Baili, Philippe Wenger an Damien Chablat Institut e Recherche en Communications et Cybernétique e Nantes, UMR C.N.R.S. 6597 1, rue e la Noë, BP 92101,

More information

Text Particles Multi-band Fusion for Robust Text Detection

Text Particles Multi-band Fusion for Robust Text Detection Text Particles Multi-ban Fusion for Robust Text Detection Pengfei Xu, Rongrong Ji, Hongxun Yao, Xiaoshuai Sun, Tianqiang Liu, an Xianming Liu School of Computer Science an Engineering Harbin Institute

More information

Deep Spatial Pyramid for Person Re-identification

Deep Spatial Pyramid for Person Re-identification Deep Spatial Pyrami for Person Re-ientification Sławomir Bąk Peter Carr Disney Research Pittsburgh, PA, USA, 15213 {slawomir.bak,peter.carr}@isneyresearch.com Abstract Re-ientification refers to the task

More information

Robust Camera Calibration for an Autonomous Underwater Vehicle

Robust Camera Calibration for an Autonomous Underwater Vehicle obust Camera Calibration for an Autonomous Unerwater Vehicle Matthew Bryant, Davi Wettergreen *, Samer Aballah, Alexaner Zelinsky obotic Systems Laboratory Department of Engineering, FEIT Department of

More information

Multimodal Stereo Image Registration for Pedestrian Detection

Multimodal Stereo Image Registration for Pedestrian Detection Multimoal Stereo Image Registration for Peestrian Detection Stephen Krotosky an Mohan Trivei Abstract This paper presents an approach for the registration of multimoal imagery for peestrian etection when

More information

Using Vector and Raster-Based Techniques in Categorical Map Generalization

Using Vector and Raster-Based Techniques in Categorical Map Generalization Thir ICA Workshop on Progress in Automate Map Generalization, Ottawa, 12-14 August 1999 1 Using Vector an Raster-Base Techniques in Categorical Map Generalization Beat Peter an Robert Weibel Department

More information

MORA: a Movement-Based Routing Algorithm for Vehicle Ad Hoc Networks

MORA: a Movement-Based Routing Algorithm for Vehicle Ad Hoc Networks : a Movement-Base Routing Algorithm for Vehicle A Hoc Networks Fabrizio Granelli, Senior Member, Giulia Boato, Member, an Dzmitry Kliazovich, Stuent Member Abstract Recent interest in car-to-car communications

More information

Transient analysis of wave propagation in 3D soil by using the scaled boundary finite element method

Transient analysis of wave propagation in 3D soil by using the scaled boundary finite element method Southern Cross University epublications@scu 23r Australasian Conference on the Mechanics of Structures an Materials 214 Transient analysis of wave propagation in 3D soil by using the scale bounary finite

More information

A Framework for Dialogue Detection in Movies

A Framework for Dialogue Detection in Movies A Framework for Dialogue Detection in Movies Margarita Kotti, Constantine Kotropoulos, Bartosz Ziólko, Ioannis Pitas, an Vassiliki Moschou Department of Informatics, Aristotle University of Thessaloniki

More information

VEHICLE MAKE AND MODEL RECOGNITION BY KEYPOINT MATCHING OF PSEUDO FRONTAL VIEW

VEHICLE MAKE AND MODEL RECOGNITION BY KEYPOINT MATCHING OF PSEUDO FRONTAL VIEW VEHICLE MAKE AND MODEL RECOGNITION BY KEYPOINT MATCHING OF PSEUDO FRONTAL VIEW Yukiko Shinozuka, Ruiko Miyano, Takuya Minagawa and Hideo Saito Department of Information and Computer Science, Keio University

More information

Characterizing Decoding Robustness under Parametric Channel Uncertainty

Characterizing Decoding Robustness under Parametric Channel Uncertainty Characterizing Decoing Robustness uner Parametric Channel Uncertainty Jay D. Wierer, Wahee U. Bajwa, Nigel Boston, an Robert D. Nowak Abstract This paper characterizes the robustness of ecoing uner parametric

More information

A shortest path algorithm in multimodal networks: a case study with time varying costs

A shortest path algorithm in multimodal networks: a case study with time varying costs A shortest path algorithm in multimoal networks: a case stuy with time varying costs Daniela Ambrosino*, Anna Sciomachen* * Department of Economics an Quantitative Methos (DIEM), University of Genoa Via

More information

Invariant Features from Interest Point Groups

Invariant Features from Interest Point Groups Invariant Features from Interest Point Groups Matthew Brown and David Lowe {mbrown lowe}@cs.ubc.ca Department of Computer Science, University of British Columbia, Vancouver, Canada. Abstract This paper

More information

Estimating Velocity Fields on a Freeway from Low Resolution Video

Estimating Velocity Fields on a Freeway from Low Resolution Video Estimating Velocity Fiels on a Freeway from Low Resolution Vieo Young Cho Department of Statistics University of California, Berkeley Berkeley, CA 94720-3860 Email: young@stat.berkeley.eu John Rice Department

More information

5th International Conference on Advanced Design and Manufacturing Engineering (ICADME 2015)

5th International Conference on Advanced Design and Manufacturing Engineering (ICADME 2015) 5th International Conference on Avance Design an Manufacturing Engineering (ICADME 25) Research on motion characteristics an application of multi egree of freeom mechanism base on R-W metho Xiao-guang

More information

Research Article Inviscid Uniform Shear Flow past a Smooth Concave Body

Research Article Inviscid Uniform Shear Flow past a Smooth Concave Body International Engineering Mathematics Volume 04, Article ID 46593, 7 pages http://x.oi.org/0.55/04/46593 Research Article Invisci Uniform Shear Flow past a Smooth Concave Boy Abullah Mura Department of

More information

Dense Disparity Estimation in Ego-motion Reduced Search Space

Dense Disparity Estimation in Ego-motion Reduced Search Space Dense Disparity Estimation in Ego-motion Reuce Search Space Luka Fućek, Ivan Marković, Igor Cvišić, Ivan Petrović University of Zagreb, Faculty of Electrical Engineering an Computing, Croatia (e-mail:

More information

A Comparison of SIFT, PCA-SIFT and SURF

A Comparison of SIFT, PCA-SIFT and SURF A Comparison of SIFT, PCA-SIFT and SURF Luo Juan Computer Graphics Lab, Chonbuk National University, Jeonju 561-756, South Korea qiuhehappy@hotmail.com Oubong Gwun Computer Graphics Lab, Chonbuk National

More information

Object Recognition and Tracking for Scene Understanding of Outdoor Mobile Robot

Object Recognition and Tracking for Scene Understanding of Outdoor Mobile Robot Object Recognition an Tracking for Scene Unerstaning of Outoor Mobile Robot My-Ha Le Faculty of Electrical an Electronic Engineering Ho Chi Minh City University of Technology an Eucation No.1 Vo Van Ngan

More information

1/5/2014. Bedrich Benes Purdue University Dec 12 th 2013 INRIA Imagine. Modeling is an open problem in CG

1/5/2014. Bedrich Benes Purdue University Dec 12 th 2013 INRIA Imagine. Modeling is an open problem in CG Berich Benes Purue University Dec 12 th 213 INRIA Imagine Inverse Proceural Moeling (IPM) Motivation IPM Classification Case stuies IPM of volumetric builings IPM of stochastic trees Urban reparameterization

More information

APPLYING GENETIC ALGORITHM IN QUERY IMPROVEMENT PROBLEM. Abdelmgeid A. Aly

APPLYING GENETIC ALGORITHM IN QUERY IMPROVEMENT PROBLEM. Abdelmgeid A. Aly International Journal "Information Technologies an Knowlege" Vol. / 2007 309 [Project MINERVAEUROPE] Project MINERVAEUROPE: Ministerial Network for Valorising Activities in igitalisation -

More information

1/5/2014. Bedrich Benes Purdue University Dec 6 th 2013 Prague. Modeling is an open problem in CG

1/5/2014. Bedrich Benes Purdue University Dec 6 th 2013 Prague. Modeling is an open problem in CG Berich Benes Purue University Dec 6 th 213 Prague Inverse Proceural Moeling (IPM) Motivation IPM Classification Case stuies IPM of volumetric builings IPM of stochastic trees Urban reparameterization IPM

More information

A Versatile Model-Based Visibility Measure for Geometric Primitives

A Versatile Model-Based Visibility Measure for Geometric Primitives A Versatile Moel-Base Visibility Measure for Geometric Primitives Marc M. Ellenrieer 1,LarsKrüger 1, Dirk Stößel 2, an Marc Hanheie 2 1 DaimlerChrysler AG, Research & Technology, 89013 Ulm, Germany 2 Faculty

More information

Fast Fractal Image Compression using PSO Based Optimization Techniques

Fast Fractal Image Compression using PSO Based Optimization Techniques Fast Fractal Compression using PSO Base Optimization Techniques A.Krishnamoorthy Visiting faculty Department Of ECE University College of Engineering panruti rishpci89@gmail.com S.Buvaneswari Visiting

More information

Non-homogeneous Generalization in Privacy Preserving Data Publishing

Non-homogeneous Generalization in Privacy Preserving Data Publishing Non-homogeneous Generalization in Privacy Preserving Data Publishing W. K. Wong, Nios Mamoulis an Davi W. Cheung Department of Computer Science, The University of Hong Kong Pofulam Roa, Hong Kong {wwong2,nios,cheung}@cs.hu.h

More information

AnyTraffic Labeled Routing

AnyTraffic Labeled Routing AnyTraffic Labele Routing Dimitri Papaimitriou 1, Pero Peroso 2, Davie Careglio 2 1 Alcatel-Lucent Bell, Antwerp, Belgium Email: imitri.papaimitriou@alcatel-lucent.com 2 Universitat Politècnica e Catalunya,

More information

Feature Based Registration - Image Alignment

Feature Based Registration - Image Alignment Feature Based Registration - Image Alignment Image Registration Image registration is the process of estimating an optimal transformation between two or more images. Many slides from Alexei Efros http://graphics.cs.cmu.edu/courses/15-463/2007_fall/463.html

More information

filtering LETTER An Improved Neighbor Selection Algorithm in Collaborative Taek-Hun KIM a), Student Member and Sung-Bong YANG b), Nonmember

filtering LETTER An Improved Neighbor Selection Algorithm in Collaborative Taek-Hun KIM a), Student Member and Sung-Bong YANG b), Nonmember 107 IEICE TRANS INF & SYST, VOLE88 D, NO5 MAY 005 LETTER An Improve Neighbor Selection Algorithm in Collaborative Filtering Taek-Hun KIM a), Stuent Member an Sung-Bong YANG b), Nonmember SUMMARY Nowaays,

More information

Patch-based Object Recognition. Basic Idea

Patch-based Object Recognition. Basic Idea Patch-based Object Recognition 1! Basic Idea Determine interest points in image Determine local image properties around interest points Use local image properties for object classification Example: Interest

More information

Cluster Center Initialization Method for K-means Algorithm Over Data Sets with Two Clusters

Cluster Center Initialization Method for K-means Algorithm Over Data Sets with Two Clusters Available online at www.scienceirect.com Proceia Engineering 4 (011 ) 34 38 011 International Conference on Avances in Engineering Cluster Center Initialization Metho for K-means Algorithm Over Data Sets

More information

Skyline Community Search in Multi-valued Networks

Skyline Community Search in Multi-valued Networks Syline Community Search in Multi-value Networs Rong-Hua Li Beijing Institute of Technology Beijing, China lironghuascut@gmail.com Jeffrey Xu Yu Chinese University of Hong Kong Hong Kong, China yu@se.cuh.eu.h

More information

Generalized Edge Coloring for Channel Assignment in Wireless Networks

Generalized Edge Coloring for Channel Assignment in Wireless Networks Generalize Ege Coloring for Channel Assignment in Wireless Networks Chun-Chen Hsu Institute of Information Science Acaemia Sinica Taipei, Taiwan Da-wei Wang Jan-Jan Wu Institute of Information Science

More information

Unknown Radial Distortion Centers in Multiple View Geometry Problems

Unknown Radial Distortion Centers in Multiple View Geometry Problems Unknown Raial Distortion Centers in Multiple View Geometry Problems José Henrique Brito 1,2, Rolan Angst 3, Kevin Köser 3, Christopher Zach 4, Pero Branco 2, Manuel João Ferreira 2, Marc Pollefeys 3 1

More information

Introduction. Introduction. Related Research. SIFT method. SIFT method. Distinctive Image Features from Scale-Invariant. Scale.

Introduction. Introduction. Related Research. SIFT method. SIFT method. Distinctive Image Features from Scale-Invariant. Scale. Distinctive Image Features from Scale-Invariant Keypoints David G. Lowe presented by, Sudheendra Invariance Intensity Scale Rotation Affine View point Introduction Introduction SIFT (Scale Invariant Feature

More information

Learning Polynomial Functions. by Feature Construction

Learning Polynomial Functions. by Feature Construction I Proceeings of the Eighth International Workshop on Machine Learning Chicago, Illinois, June 27-29 1991 Learning Polynomial Functions by Feature Construction Richar S. Sutton GTE Laboratories Incorporate

More information

Map-Enhanced UAV Image Sequence Registration and Synchronization of Multiple Image Sequences

Map-Enhanced UAV Image Sequence Registration and Synchronization of Multiple Image Sequences Map-Enhanced UAV Image Sequence Registration and Synchronization of Multiple Image Sequences Yuping Lin and Gérard Medioni Computer Science Department, University of Southern California 941 W. 37th Place,

More information

6 Gradient Descent. 6.1 Functions

6 Gradient Descent. 6.1 Functions 6 Graient Descent In this topic we will iscuss optimizing over general functions f. Typically the function is efine f : R! R; that is its omain is multi-imensional (in this case -imensional) an output

More information

Generalized Edge Coloring for Channel Assignment in Wireless Networks

Generalized Edge Coloring for Channel Assignment in Wireless Networks TR-IIS-05-021 Generalize Ege Coloring for Channel Assignment in Wireless Networks Chun-Chen Hsu, Pangfeng Liu, Da-Wei Wang, Jan-Jan Wu December 2005 Technical Report No. TR-IIS-05-021 http://www.iis.sinica.eu.tw/lib/techreport/tr2005/tr05.html

More information

Particle Swarm Optimization Based on Smoothing Approach for Solving a Class of Bi-Level Multiobjective Programming Problem

Particle Swarm Optimization Based on Smoothing Approach for Solving a Class of Bi-Level Multiobjective Programming Problem BULGARIAN ACADEMY OF SCIENCES CYBERNETICS AND INFORMATION TECHNOLOGIES Volume 17, No 3 Sofia 017 Print ISSN: 1311-970; Online ISSN: 1314-4081 DOI: 10.1515/cait-017-0030 Particle Swarm Optimization Base

More information

A new fuzzy visual servoing with application to robot manipulator

A new fuzzy visual servoing with application to robot manipulator 2005 American Control Conference June 8-10, 2005. Portlan, OR, USA FrA09.4 A new fuzzy visual servoing with application to robot manipulator Marco A. Moreno-Armenariz, Wen Yu Abstract Many stereo vision

More information

A Classification of 3R Orthogonal Manipulators by the Topology of their Workspace

A Classification of 3R Orthogonal Manipulators by the Topology of their Workspace A Classification of R Orthogonal Manipulators by the Topology of their Workspace Maher aili, Philippe Wenger an Damien Chablat Institut e Recherche en Communications et Cybernétique e Nantes, UMR C.N.R.S.

More information

Video-based Characters Creating New Human Performances from a Multi-view Video Database

Video-based Characters Creating New Human Performances from a Multi-view Video Database Vieo-base Characters Creating New Human Performances from a Multi-view Vieo Database Feng Xu Yebin Liu? Carsten Stoll? James Tompkin Gaurav Bharaj? Qionghai Dai Hans-Peter Seiel? Jan Kautz Christian Theobalt?

More information

Design and Analysis of Optimization Algorithms Using Computational

Design and Analysis of Optimization Algorithms Using Computational Appl. Num. Anal. Comp. Math., No. 3, 43 433 (4) / DOI./anac.47 Design an Analysis of Optimization Algorithms Using Computational Statistics T. Bartz Beielstein, K.E. Parsopoulos,3, an M.N. Vrahatis,3 Department

More information

Feature Detectors and Descriptors: Corners, Lines, etc.

Feature Detectors and Descriptors: Corners, Lines, etc. Feature Detectors and Descriptors: Corners, Lines, etc. Edges vs. Corners Edges = maxima in intensity gradient Edges vs. Corners Corners = lots of variation in direction of gradient in a small neighborhood

More information

CONTENT-BASED RETRIEVAL OF DEFECT IMAGES. Jukka Iivarinen and Jussi Pakkanen

CONTENT-BASED RETRIEVAL OF DEFECT IMAGES. Jukka Iivarinen and Jussi Pakkanen Proceeings of ACIVS 2002 (Avance Concepts for Intelligent Vision Systems), Ghent, Belgium, September 9-11, 2002 CONTENT-BASED RETRIEVAL OF DEFECT IMAGES Jukka Iivarinen an Jussi Pakkanen jukka.iivarinen@hut.fi,

More information

Queueing Model and Optimization of Packet Dropping in Real-Time Wireless Sensor Networks

Queueing Model and Optimization of Packet Dropping in Real-Time Wireless Sensor Networks Queueing Moel an Optimization of Packet Dropping in Real-Time Wireless Sensor Networks Marc Aoun, Antonios Argyriou, Philips Research, Einhoven, 66AE, The Netherlans Department of Computer an Communication

More information

Automatic Image Alignment (feature-based)

Automatic Image Alignment (feature-based) Automatic Image Alignment (feature-based) Mike Nese with a lot of slides stolen from Steve Seitz and Rick Szeliski 15-463: Computational Photography Alexei Efros, CMU, Fall 2006 Today s lecture Feature

More information

Real Time On Board Stereo Camera Pose through Image Registration*

Real Time On Board Stereo Camera Pose through Image Registration* 28 IEEE Intelligent Vehicles Symposium Einhoven University of Technology Einhoven, The Netherlans, June 4-6, 28 Real Time On Boar Stereo Camera Pose through Image Registration* Fai Dornaika French National

More information

Instance-level recognition part 2

Instance-level recognition part 2 Visual Recognition and Machine Learning Summer School Paris 2011 Instance-level recognition part 2 Josef Sivic http://www.di.ens.fr/~josef INRIA, WILLOW, ENS/INRIA/CNRS UMR 8548 Laboratoire d Informatique,

More information

Solution Representation for Job Shop Scheduling Problems in Ant Colony Optimisation

Solution Representation for Job Shop Scheduling Problems in Ant Colony Optimisation Solution Representation for Job Shop Scheuling Problems in Ant Colony Optimisation James Montgomery, Carole Faya 2, an Sana Petrovic 2 Faculty of Information & Communication Technologies, Swinburne University

More information

[2006] IEEE. Reprinted, with permission, from [Damith C. Herath, Sarath Kodagoda and Gamini Dissanayake, Simultaneous Localisation and Mapping: A

[2006] IEEE. Reprinted, with permission, from [Damith C. Herath, Sarath Kodagoda and Gamini Dissanayake, Simultaneous Localisation and Mapping: A [6] IEEE. Reprinte, with permission, from [Damith C. Herath, Sarath Koagoa an Gamini Dissanayake, Simultaneous Localisation an Mapping: A Stereo Vision Base Approach, Intelligent Robots an Systems, 6 IEEE/RSJ

More information

Bends, Jogs, And Wiggles for Railroad Tracks and Vehicle Guide Ways

Bends, Jogs, And Wiggles for Railroad Tracks and Vehicle Guide Ways Ben, Jogs, An Wiggles for Railroa Tracks an Vehicle Guie Ways Louis T. Klauer Jr., PhD, PE. Work Soft 833 Galer Dr. Newtown Square, PA 19073 lklauer@wsof.com Preprint, June 4, 00 Copyright 00 by Louis

More information

1 Surprises in high dimensions

1 Surprises in high dimensions 1 Surprises in high imensions Our intuition about space is base on two an three imensions an can often be misleaing in high imensions. It is instructive to analyze the shape an properties of some basic

More information

A Rapid Automatic Image Registration Method Based on Improved SIFT

A Rapid Automatic Image Registration Method Based on Improved SIFT Available online at www.sciencedirect.com Procedia Environmental Sciences 11 (2011) 85 91 A Rapid Automatic Image Registration Method Based on Improved SIFT Zhu Hongbo, Xu Xuejun, Wang Jing, Chen Xuesong,

More information

A Multi-scale Boosted Detector for Efficient and Robust Gesture Recognition

A Multi-scale Boosted Detector for Efficient and Robust Gesture Recognition A Multi-scale Booste Detector for Efficient an Robust Gesture Recognition Camille Monnier, Stan German, Anrey Ost Charles River Analytics Cambrige, MA, USA Abstract. We present an approach to etecting

More information

SLAM in Dynamic Environments via ML-RANSAC

SLAM in Dynamic Environments via ML-RANSAC *Manuscript Clic here to view line References SLAM in Dynamic Environments via ML-RANSAC Masou S. Bahraini 1,2, Mohamma Bozorg 2, Ahma B. Ra 1* 1 School of Mechatronic Systems Engineering, Simon Fraser

More information

Features Points. Andrea Torsello DAIS Università Ca Foscari via Torino 155, Mestre (VE)

Features Points. Andrea Torsello DAIS Università Ca Foscari via Torino 155, Mestre (VE) Features Points Andrea Torsello DAIS Università Ca Foscari via Torino 155, 30172 Mestre (VE) Finding Corners Edge detectors perform poorly at corners. Corners provide repeatable points for matching, so

More information

Tensor Decomposition of Dense SIFT Descriptors in Object Recognition

Tensor Decomposition of Dense SIFT Descriptors in Object Recognition Tensor Decomposition of Dense SIFT Descriptors in Object Recognition Tan Vo 1 and Dat Tran 1 and Wanli Ma 1 1- Faculty of Education, Science, Technology and Mathematics University of Canberra, Australia

More information

Fuzzy Clustering in Parallel Universes

Fuzzy Clustering in Parallel Universes Fuzzy Clustering in Parallel Universes Bern Wisweel an Michael R. Berthol ALTANA-Chair for Bioinformatics an Information Mining Department of Computer an Information Science, University of Konstanz 78457

More information

New Version of Davies-Bouldin Index for Clustering Validation Based on Cylindrical Distance

New Version of Davies-Bouldin Index for Clustering Validation Based on Cylindrical Distance New Version of Davies-Boulin Inex for lustering Valiation Base on ylinrical Distance Juan arlos Roas Thomas Faculta e Informática Universia omplutense e Mari Mari, España correoroas@gmail.com Abstract

More information

Dual Arm Robot Research Report

Dual Arm Robot Research Report Dual Arm Robot Research Report Analytical Inverse Kinematics Solution for Moularize Dual-Arm Robot With offset at shouler an wrist Motivation an Abstract Generally, an inustrial manipulator such as PUMA

More information

Specular 3D Object Tracking by View Generative Learning

Specular 3D Object Tracking by View Generative Learning Specular 3D Object Tracking by View Generative Learning Yukiko Shinozuka, Francois de Sorbier and Hideo Saito Keio University 3-14-1 Hiyoshi, Kohoku-ku 223-8522 Yokohama, Japan shinozuka@hvrl.ics.keio.ac.jp

More information

Scale Invariant Feature Transform

Scale Invariant Feature Transform Why do we care about matching features? Scale Invariant Feature Transform Camera calibration Stereo Tracking/SFM Image moiaicing Object/activity Recognition Objection representation and recognition Automatic

More information

Real Time Tracking via Sparse Representation

Real Time Tracking via Sparse Representation 203 8th International Conference on Communications an Networing in China (CHINACOM) Real Time Tracing via Sparse Representation Hong-Mei Zhang, Xian-Sui Wei, Tao Huang 2, Yan He, Xiang-Li Zhang, JinYe

More information

WLAN Indoor Positioning Based on Euclidean Distances and Fuzzy Logic

WLAN Indoor Positioning Based on Euclidean Distances and Fuzzy Logic WLAN Inoor Positioning Base on Eucliean Distances an Fuzzy Logic Anreas TEUBER, Bern EISSFELLER Institute of Geoesy an Navigation, University FAF, Munich, Germany, e-mail: (anreas.teuber, bern.eissfeller)@unibw.e

More information

Evolutionary Optimisation Methods for Template Based Image Registration

Evolutionary Optimisation Methods for Template Based Image Registration Evolutionary Optimisation Methos for Template Base Image Registration Lukasz A Machowski, Tshilizi Marwala School of Electrical an Information Engineering University of Witwatersran, Johannesburg, South

More information

X y. f(x,y,d) f(x,y,d) Peak. Motion stereo space. parameter space. (x,y,d) Motion stereo space. Parameter space. Motion stereo space.

X y. f(x,y,d) f(x,y,d) Peak. Motion stereo space. parameter space. (x,y,d) Motion stereo space. Parameter space. Motion stereo space. 3D Shape Measurement of Unerwater Objects Using Motion Stereo Hieo SAITO Hirofumi KAWAMURA Masato NAKAJIMA Department of Electrical Engineering, Keio Universit 3-14-1Hioshi Kouhoku-ku Yokohama 223, Japan

More information

Blind Data Classification using Hyper-Dimensional Convex Polytopes

Blind Data Classification using Hyper-Dimensional Convex Polytopes Blin Data Classification using Hyper-Dimensional Convex Polytopes Brent T. McBrie an Gilbert L. Peterson Department of Electrical an Computer Engineering Air Force Institute of Technology 9 Hobson Way

More information

Feature Extraction and Rule Classification Algorithm of Digital Mammography based on Rough Set Theory

Feature Extraction and Rule Classification Algorithm of Digital Mammography based on Rough Set Theory Feature Extraction an Rule Classification Algorithm of Digital Mammography base on Rough Set Theory Aboul Ella Hassanien Jafar M. H. Ali. Kuwait University, Faculty of Aministrative Science, Quantitative

More information

A NEW FEATURE BASED IMAGE REGISTRATION ALGORITHM INTRODUCTION

A NEW FEATURE BASED IMAGE REGISTRATION ALGORITHM INTRODUCTION A NEW FEATURE BASED IMAGE REGISTRATION ALGORITHM Karthik Krish Stuart Heinrich Wesley E. Snyder Halil Cakir Siamak Khorram North Carolina State University Raleigh, 27695 kkrish@ncsu.edu sbheinri@ncsu.edu

More information

Message Transport With The User Datagram Protocol

Message Transport With The User Datagram Protocol Message Transport With The User Datagram Protocol User Datagram Protocol (UDP) Use During startup For VoIP an some vieo applications Accounts for less than 10% of Internet traffic Blocke by some ISPs Computer

More information

Rough Set Approach for Classification of Breast Cancer Mammogram Images

Rough Set Approach for Classification of Breast Cancer Mammogram Images Rough Set Approach for Classification of Breast Cancer Mammogram Images Aboul Ella Hassanien Jafar M. H. Ali. Kuwait University, Faculty of Aministrative Science, Quantitative Methos an Information Systems

More information

Instance-level recognition II.

Instance-level recognition II. Reconnaissance d objets et vision artificielle 2010 Instance-level recognition II. Josef Sivic http://www.di.ens.fr/~josef INRIA, WILLOW, ENS/INRIA/CNRS UMR 8548 Laboratoire d Informatique, Ecole Normale

More information

Architecture Design of Mobile Access Coordinated Wireless Sensor Networks

Architecture Design of Mobile Access Coordinated Wireless Sensor Networks Architecture Design of Mobile Access Coorinate Wireless Sensor Networks Mai Abelhakim 1 Leonar E. Lightfoot Jian Ren 1 Tongtong Li 1 1 Department of Electrical & Computer Engineering, Michigan State University,

More information

Intensive Hypercube Communication: Prearranged Communication in Link-Bound Machines 1 2

Intensive Hypercube Communication: Prearranged Communication in Link-Bound Machines 1 2 This paper appears in J. of Parallel an Distribute Computing 10 (1990), pp. 167 181. Intensive Hypercube Communication: Prearrange Communication in Link-Boun Machines 1 2 Quentin F. Stout an Bruce Wagar

More information

An Algorithm for Building an Enterprise Network Topology Using Widespread Data Sources

An Algorithm for Building an Enterprise Network Topology Using Widespread Data Sources An Algorithm for Builing an Enterprise Network Topology Using Wiesprea Data Sources Anton Anreev, Iurii Bogoiavlenskii Petrozavosk State University Petrozavosk, Russia {anreev, ybgv}@cs.petrsu.ru Abstract

More information