Moving object detection by fuzzy set theory

According to the problems of numerous noises and incomplete detection of moving object detection in complex natural environment, an improved solution method based on fuzzy set theory is put forward. First, pyramid-style multi-resolution model is used for background difference, and the preliminary foreground mask is obtained. Then the model of color, time, space and locality for the current image are represented by fuzzy sets to form a cluster of fuzzy vector set. At last, the features of these four vectors are fused together by fuzzy mathematic theory to get the membership of foreground for each pixel and detect the object. The method not only can detect relative complete moving object effectively, but overcome the influence of small moving background in natural environment at the same time. The experiments are carried out and the results show that the foreground recognition rate by the proposed method is 0.7174 and the error rate is 0.011 8. It can adapt to the influence of the dynamic background in natural environment.