A moving target detection algorithm based on the dynamic background
Download 2.87 Mb.
|
| We can see after a period of time the car at upper into the background, and the car at the right upper out of the background.
B. Determination of threshold In order to increase the adjustability of the threshold and the robustness of the background image on the brightness changes slowly. The determination of threshold as follows: 
And c determined by the experiment, the general admission 3-5; Mi is a region of the background, and generally selects the area at the edge; N is the area of Mi. The algorithm selected the four corners of differential gray image region to be calculated respectively，and makes the mid-value as the final check of the threshold value, and get a better result. C. Extraction of detailed images of moving targets This requires the adoption of connectedness analysis to extract the complete moving target. There are two type of connectedness: four-connected and eight-connected, as shown as Fig. 13 and Fig. 14. 
VI. PERFORMANCE COMPARISON OF THE THREE METHODS’ RESULT
recognize moving objects, and the detected object contour is clear, the extracted object image is complete. But the disadvantage is that when the moving object stop in the scene this method can not put it into the background automatically. Fig. 20 and Fig.21 show the original image and the detected moving object when the car moved into the scene at time t1 and then stop. Fig.22 and Fig.23 show the original image and detected object image at time t2. 
We can find out that from t1 to t2, the car did not move, but this method detected it as moving object. 2) Frame Difference: this method is able to recognize moving object. But the detected object’s contour is dim and the extracted object image is not complete. The change of background has little influence over the result image, and we can always detect out the right moving object. Fig.24 and Fig.25 show the detected object image at time t1 and t2 
We can find out that it recognizes the right object. But the object’s silhouette is not clear and there are holes inside it 3) The moving target detection algorithm based on the dynamic background: it is able to recognize moving target, and the detected object contour is clear, the extracted object image is complete. When the moving object stop in the scene this method can put it into the background automatically. Fig. 26 and Fig. 27 also show the detected object image at time t1 and t2. We can find out that from t1 to t2, the method recognize the right moving object and extracted the clear image of target 
The comparison of operating efficiency We used the three algorithms to detect the same video, and compared them in three aspects: running time (100 frames), memory, smooth output the result. The results as shown in TABLE I, we can find that although the new algorithm is a little larger than other two algorithms, however its output is also smoothly and the memory isn’t too much, so it meets the requirements of real-time. Although the moving target detection algorithm based on the dynamic background can better meet the set performance requirements. However, to design a perfect intelligent visual surveillance system, we should further improve the system robustness and increase target identification functions. ACKNOWLEDGMENT This paper is supported by the National High Technology Research and Development Program of China under Grant No. 2007AA04Z114. REFERENCES [1] Wang Ying-li , Dai Jing-min, “Moving Targets Detection and Tracking Based on Nonlinear Adaptive Filtering,” 2007 International Conference on Computational Intelligence and Security Workshops (CISW 2007), pp. 691-694, December 2007. [2] Yang Shu-Ying , Zhang Cheng , Zhang We-Yu , He Pi-Lian, Unknown Moving Target Detecting and Tracking Based on Computer Vision, Fourth International Conference on Image and Graphics (ICIG 2007), pp 490-495, August 2007 [3] Tan Jiyuan, Wu Chengdong, Zhou Yun, Hou Jun, Wang Qiaoqiao, “Reasearch of Abnormal Target Detection Algorithm in intelligent Surveillance System”, Proceedings of the 2009 International Conference on Advanced Computer Control, IEEE Computer Society, 2009. [4] Fa-quan Zhang , Yong Zhang , Li-ping Lu , Li-ying Jiang , Guang-zhao Cui, “Speedy Detection Algorithm of Underwater Moving Targets Based on Image Sequence,” 2009 International Conference on ComputerEngineering and Technology, pp. 230–233, January 2009 [5] Mayur D. Jain, S. Nalin Pradeep. A video surveillance system under varying environmental conditions. Proceedings of the 24th IASTED international conference on Signal processing, pattern recognition, and applications,ACTAPress, February 2006. [6] By M. A. Zerafat Pisheh , A. Sheikhi, “Detection and Compensation of Image Sequence Jitter Due to an Unstable CCD Camera for Video Tracking of a Moving Target,” Second International Symposium on 3D Data Processing, Visualization and Transmission (3DPVT'04), pp. 258- 261, September 2004 [7] M.-C. Tsai, K.-Y. Chen, M.-Y. Cheng, K. C. Lin, “Implementation of a real-time moving object tracking system using visual servoing” Cambridge University Press , Volume 21 Issue 6, December 2003. Directory: krest-academic-projects -> krest-major-projects -> ECE ECE -> An adaptive audio watermarking based on the singular value decomposition in the wavelet domain ECE -> An Audio/Speech Watermarking Method for Copyright Protection Abstract krest-academic-projects -> Deconvolving Images with Unknown Boundaries Using the Alternating Direction Method of Multipliers Synopsis krest-academic-projects -> Modeling and analysis of an automobile radiator Download 2.87 Mb. Share with your friends:
|