Video Moving Object Detection Based On Desicion Fusion

The process of video moving object detection always has some uncertainty in the pixel's decision belonging to foreground or background. In this paper, we attempt to resolve such uncertainty problems by using multiple methods based on decision fusion. First, we convert the decision outputs of the Gaussian Mixture Model (GMM), median filter and codebook three single methods into the membership outputs, respectively. Then, we fuse these three methods by using different decision fusion rules to obtain the detection results. Experimental results show that the fusion-based-detection method has better detection performance.