Detection and Segmentation of Moving Objects Based on Support Vector Machine

In order to improve the accuracy of multi-moving objects detection in surveillant video, this paper presents a new method of detection and segmentation for moving objects based on SVM (support vector machine). To further enhance the accuracy of segmentation using support vector machine, we modify the kernel function based on its nature, and some experiments have been done to compare with other kernel functions commonly used. The experimental results show that the classifier with the kernel function of RBF + Gaussian RBF has the better classification performance. We also compare our algorithm with frame difference and background subtraction method. Experiments show that our algorithm is effective and robust for the coming, gradient and leaving of moving objects in video, and it is immune to the illumination changes in scene and the speed changes of moving objects movement, besides, no significant non-connectivity exists in the detected moving objects. Moreover, no thresholds, which are often hard to select in most segmentation methods, are involved in our algorithm.