A Novel Binary Image Filtering Algorithm Based on Information Entropy

The results of moving object detection in image sequences are often represented as binary image sequences containing both masks of moving objects and false positives, which make it necessary to quickly eliminate false positives and preserve the interested moving objects as completely as possible online. Traditional approaches such as morphological filter and area threshold filter, sometimes, are unfit for the job because of unsatisfactory results and heavy computational load. In light of the fact that foreground pixels density caused by moving objects is higher than that of caused by disturbances, we propose a novel binary image filtering method based on discrete information entropy (BIFBE) for the task. Experimental results show that BIFBE is more effective and has less computational load than traditional methods