Moving Object Detection Using Frame Difference, Background Subtraction And SOBS For Video Surveillance Application

Transcription

1 Moving Object Detection Using Frame Difference, Background Subtraction And SOBS For Video Surveillance Application Pritee Gupta 1, Yashpal Singh 2, Manoj Gupta 3 1 Ph.D Student, Bhagwant university, Ajmer & Assistant Professor, CSE Deptt,, I.T.S Engg. College, Gr. Noida (U.P), India 2 Associate Professor & Head IT Deptt, Bundelkhand Institute of Engg. &Technology, Jhansi, India 3 Associate Professor, CSE Deptt, School of Engineering and Technology, Sharda University, Gr. Noida (U.P) 1 23march.pritee@gmail.com 2 yash_biet@yahoo.co.in 3 India manoj.cst@gmail.com Abstract Visual surveillance has been a very active research topic in the last few years due to its growing importance in security, law enforcement, and military applications. Detection of moving objects in video streams is the first relevant step of information extraction in many computer vision applications. Aside from the intrinsic usefulness of being able to segment video streams into moving and background components, detecting moving objects provides a focus of attention for recognition, classification, and activity analysis, making these later steps more efficient. In this work three methods as background subtraction, frame difference and SOBS methods are used to detect moving object from input video and report experimental results. The experimental results show the most suitable method that runs quickly, accurately and fits for the real-time detection. Keywords Frame separation, Background subtraction,, Frame subtraction, SOBS, Motion detection I. INTRODUCTION Surveillance is the monitoring of behavior. Systems surveillance is the process of monitoring the behavior of people, objects or processes within systems for conformity to expected or desired norms in trusted systems for security or social control. The word surveillance is commonly used to describe observation from a distance by means of electronic equipment or other technological means. Fig. 1 Example of CCTV camera At a basic level, computers are a surveillance target because large amounts of personal information are stored on them. Anyone who can access or remove a computer can retrieve information. If someone is able to install software on a computer system, they can turn the computer into a surveillance device. CCTV is a collection of video cameras used for video surveillance. CCTV is generally used in areas where there is an increased need for security, such as banks, airports and town centres. A basic CCTV system comprises of the Camera, lens and power supply. Recording device, VCR or a digital video recorder and monitor. Closed-circuit television (CCTV) is the use of video cameras to transmit a signal to a specific place, on a limited set of monitors [1]. The main tasks in visual surveillance systems include motion detection, object classification, tracking. Our focus here is on the detection phase of a general visual surveillance system using static cameras. 151

2 The usual approach for moving object detection is through background subtraction that consists in maintaining an up-to date model of the background and detecting moving objects as those that deviate from such a model. The background image is not fixed but must adapt to: Illumination changes, sudden (such as clouds), Motion changes, camera oscillations, highfrequencies background objects (such as tree branches, sea waves, and similar) Changes in the background geometry. II. LITERATURE SURVEY The three ways for detecting motion in input image sequences is discussed in this paper as: (a) Frame Difference (b) Background subtraction (c) SOBS (a) Frame Difference This method is through the difference between two consecutive images to determine the presence of moving objects. Fig. 2 Example of frame difference method The Frame difference is arguably the simplest form of background subtraction. Frame differencing, also known as temporal difference, uses the video frame at time t-1 as the background model for the frame at time t. This technique is sensitive to noise and variations in illumination, and does not consider local consistency properties of the change mask. This method also fails to segment the non-background objects if they stop moving. Since it uses only a single previous frame, frame differencing may not be able to identify the interior pixels of a large, uniformly-colored moving object. This is commonly known as the aperture problem. It has strong adaptability, but it is generally difficult to obtain a complete outline of moving object, liable to appear the empty phenomenon, as a result the detection of moving object is not accurate. (b) Background subtraction The basic scheme of background subtraction is to subtract the image from a reference image that models the background scene. Typically, the basic steps of the algorithm are as follows: Background modeling constructs a reference image representing the background. Threshold selection determines appropriate threshold values used in the subtraction operation to obtain a desired detection rate. Subtraction operation or pixel classification classifies the type of a given pixel, i.e., the pixel is the part of background (including ordinary background and shaded background), or it is a moving object. After background image B(X, Y) is obtained, subtract the background image B(X, Y) from the current frame FK(X, Y). If the pixel difference is greater than the set threshold T, then determines that the pixels appear in the moving object, otherwise, as the background pixels [5]. The moving object can be detected after threshold operation. Its expression is as follows: DK(X, Y) = 1 if ( FK(X, Y) B(X, Y) > T) (1) = 0 others 152

3 3 rd International Conference on System Modeling & Advancement in Research Trends (SMART) Fig. 3 Example of Background subtraction method Background subtraction method is very sensitive to the changes in the external environment. The methods with a background model based on a single scalar value can guarantee adaptation to slow illumination changes, but cannot cope with multi-valued background distributions. As such, they will be prone to errors whenever those situations arise. Processing time required to detect the object using this technique is low but accuracy may not be good enough ([2], [3],[4],[5],[6]). (c) SOBS (Self organizing background subtraction method) There are many challenges in developing a good background subtraction algorithm. First, it must be robust against changes in illumination. Second, it should avoid detecting non-stationary background objects such as moving leaves, rain, snow, and shadows cast by moving objects. Some of the well-known issues in background maintenance are as light changes, moving background, cast shadows, bootstrapping:,camouflage etc. The main problem of the background subtraction approach to moving object detection is its extreme sensitivity to dynamic scene changes due to lighting and extraneous events. While the background model eventually adapts to these holes, they generate false alarms for a short period of time ([7], [8],[9],[10]). Therefore, it is highly desirable to construct an approach to motion detection based on a background model that automatically adapts to changes in a self-organizing manner and without a priori knowledge. This is biologically inspired problem-solving method based on visual attention mechanisms. This approach defines a method for the generation of an active attention focus to monitor dynamic scenes for surveillance purposes ([11],[12],[13],[14]). The idea is to build the background model by learning in a selfmany background organizing manner variations, i.e., background motion cycles, seen as trajectories of pixels in time. Based on the learnt background model through a map of motion and stationary patterns, this can detect motion and selectively update the background model. Each node computes a function of the weighted linear combination of incoming inputs, where weights resemble the neural network learning. Doing so, each node could be represented by a weight vector, obtained collecting the weights related to incoming links. In the following, the set of weight vectors will be called a model. An incoming pattern is mapped to the node whose model is most similar to the pattern, and weight vectors in a neighborhood of such node are updated. Therefore, the network behaves as a competitive neural network that implements a winner take- all function with an associated mechanism, that modifies the local synaptic plasticity of the neurons, allowing learning to be restricted spatially to the local neighborhood of the most active neurons. For each color pixel, consider a neuronal map consisting of n*n weight vectors. Each incoming sample is mapped to the weight vector that is closest according to a suitable distance measure, and the weight vectors in its neighborhood are updated. The whole set of weight vectors acts as a background model, that is used for background subtraction in order to identify moving pixels ([15], 16],[17]). 153

4 IV. RESULT AND DISCUSSION Fig. 4 (Left) Simple image and the (right) neuronal map structure III. PROPOSED WORK IMPLEMENTATION In this work the aim is to build such a surveillance system, which will detect motion and tracking even if the moving background, gradual illumination variations and camouflage and shadow into the background,thus achieves robust detection for different types of videos taken with stationary cameras. To fulfill this aim, a strong computing software called Matlab is used. Matlab provides image Acquisition and Image Processing Toolboxes which facilitate us in creating a good code. Frame processing is the first step after selecting the input video. The purpose of this step is to prepare the modified video frames by removing noise and unwanted object s in the frame in order to increase the amount of information gained from the frame and the sensitivity of the algorithm. Preprocessing is a process of collecting simple image processing tasks that change the raw input video info a format. This can be processed by subsequent steps. Experimental results for moving object detection using the proposed methods have been produced for input video, that represent typical situations critical for video surveillance systems, and present qualitative results obtained with the proposed three methods. V. SEQUENCE PERSON S MOVEMENT MONITOR Sequence Person s movement monitor is an outdoor sequence consisting of 31 frames of 320 x 240 spatial resolutions, acquired at a frequency of 30 fps (frames per second). The scene consist a open area, where a man comes out and then goes in the other side. This represents an example of easy sequence, in that lighting conditions are quite stable and moving objects are well contrasted with the background, however, strong shadows cast by moving objects can be observed in the entire sequence. Several points for moving object detection using the given approach have been produced for input image. (a) Fig.5 Block diagram of proposed methodology (b) 154

5 fact that this model learns background motion trajectories captured by different weight vectors and, therefore, different color clusters, and to the adoption of distance, which allows a quite fair discrimination among colors. (c) VI. CONCLUSION AND ONGOING WORK (d) Fig.6 Input video is detected and results are produced using methods as (a) main window (b) Frame difference (c) Background subtraction (d) Neuronal Map Background subtraction method is robust method rather than frame difference and SOBS method Frame difference method has major flaw of this method is that for objects with uniformly distributed intensity values, the pixels are interpreted as part of the background. Another problem is that objects must be continuously moving. SOBS method gives good result but processing time is very high. Values for the distance thresholds should be chosen such that High values for allow to limit selectivity in the update of the background model during the calibration phase, enabling the inclusion into the initial background model of several observed pixel intensity variations. Lower values for should be chosen to obtain a more accurate background model in the online phase, we can observe that the object is almost perfectly detected, despite his camouflage and moving background. This is due to the In video surveillance, there are many interference factors such as target changes, complex scenes, and target deformation in the moving object tracking.in this paper for a Input video stream background subtraction Frame difference & neural mapping methods(sobs) of moving object detection methods are used. Experimental results shows that the Background subtraction method using dynamic threshold is most suitable method for perfectly detection of moving object in a video stream. Future work will address on techniques to get better results to improve the human detection methods and occlusion handling in surveillance applications. REFERENCES [1] R. T. Collins, A. J. Lipton, T. Kanade, H. Fujiyoshi, D. Duggins, Y. Tsin, D. Tolliver, N. Enomoto, O. Hasegawa, P. Burt, and L. Wixson, A system for video surveillance and monitoring, Tech.Rep. CMU-RI-TR-00-12, The Robotics Inst., Carnegie Mellon Univ.,Pittsburgh, PA, [2] M.Dimitrijevic, "Human body pose detection using Bayesian spatiotemporal Templates, " 2007 International Conference on Intelligent and Advanced Systems, 2008, pp [3] Tao Jianguo, Yu Changhong, "Real-Time Detection and Tracking of Moving Object," Intelligent Information Technology Application, UTA '08. Second International Symposium on Volume 2, Dec Page(s): [4] J. L. Barron, D. J. Fleet, and S. S. Beauchemin, Performance of optical flow techniques, Int. J. Comput. Vis., vol. 12, no. 1, pp , [5] M. Piccardi, Background subtraction techniques: a review, in Proc. IEEE Int. Conf. Systems, Man, Cybernetics, 2004, pp [61 Niu Lianqiang and Nan Jiang, "A moving objects detection algorithm based on improved background subtraction," Intelligent Systems Design and Applications, ISDA '08. Eighth International Conference on Volume 3, Nov Page(s):

6 [7] Du-Ming Tsai and Shia-Chih Lai, "Independent Component Analysis Based Background Subtraction for Indoor Surveillance," hnage Processing,IEEE Transactions on Volume 18, Issue 1, Jan Page(s): [8] K.Kinoshita, M.Enokidani, M. Izumida and K.Murakami, "Tracking of a Moving Object Using One-Dimensional Optical Flow with a Rotating Observer," Control, Automation, Robotics and Vision, ICARCV '06. 9th International Conference on 5-8 Dec Page(s): 1-6 [9]N.J.Bauer and P.N.Pathirana,"Object focused simultaneous estimation of optical flow and state dynamics," Intelligent Sensors, Sensor Networks and Information Processing, ISSNIP International Conference on Dec Page(s):6l - 66 [10] Zhen Tang and Zhenjiang Miao, "Fast Background Subtraction and Shadow Elimination Using hnproved Gaussian Mixture Model," Haptic, Audio and Visual Environments and Garnes, HAVE IEEE International Workshop on Oct Page(s):38 41 [11 Mr. Mahesh C. Pawaskar1, Mr. N. S.Narkhede2 and Mr. Saurabh S. Athalye1Detection Of Moving Object Based On Background Subtraction, International Journal of Emerging Trends & Technology in Computer Science (IJETTCS), Volume 3, Issue 3, May-June 2014 [12] A. Elgammal, D. Hanvood, and L. S. Davis, Nonparametric model for background subtraction, in Proc. ECCV, 2000, pp [13] L. Maddalena and A. Petrosino, A self-organizing approach to detection of moving patterns for real-time applications, in Proc. 2nd Int. Symp. Brain, Vision, and Artificial Intelligence, 2007, pp , Lecture Notes Comput. Sci [14] Ms. Pritee Gupta, Dr. Yashpal Singh, Implementation of Dynamic Threshold Method for Human Motion Detection in Video surveillance application, International Journal of Computers and Technology,2014 Vol. 13, No. 8 [15] Wang Weiqiang, Yang Jie and Gao Wen, "Modeling Background and Segmenting Moving Objects from Compressed Video," Circuits and Systems for Video Technology, IEEE Transactions on Volume 18, Issue 5, May 2008 Page(s): [16] M. Mignotte and IKonrad, "Statistical Background Subtraction Using Spatial Cues," Circuits and Systems for Video Technology, IEEE Transactions on Volume 17, Issue 12, Dec Page(s): [17] Chen Yuan, Yu Shengsheng, Sun Weiping and Li Hongxing, "Objects Detecting Based on Adaptive Background Models and Multiple Cues," Computing, Communication, Control, and Management, CCCM '08. ISECS International Colloquium on Volume 1, 3-4 Aug Page(s):