A background subtraction method is a computationally inexpensive way to identify moving objects in the scene without any prior information about object and it also provides a sufficient light of information to accomplish critical task in traffic monitoring, object tracking, pattern recognition, human gait and gesture detection. However, for real time systems, the background scene is seriously affected due to changes in lightening condition, shadow cast by moving object, swaying tree, rippling water and much more, which hurdles to produce a reliable motion mask. In this concern, we focus toward the selection of the background pixel by mapping the time variance and absolute difference image in order to cope with abrupt illumination and preserve the spatial consistency. Further the local statistical properties and variance of background image are employed to reduce the local noise impulse within background candidate. Experimental results show that it can work well under static and dynamic background condition.
Background subtraction is one of the most reliable approach to localize the moving object under static camera arrangement. As seen, the moving object detection is a preliminary task in many vision applications such as video analysis, object tracking and activity analysis. However, the quasi-stationary pixels, aperture effect, ghost trail and varying illumination are still an annoying factors in the extraction procedures of the actual moving object in video. To alleviate the above problems, a video segmentation method is proposed that utilizes background subtraction and the 3-class fuzzy c-means clustering algorithm for extracting the relevant moving pixels. The proposed algorithm modifies learning parameters of adaptive filters to adapt the changes in the background. Afterwards, it incorporates the Markovian framework in which the initial motion field provides a prior information to regularize the segmentation process. The method achieves better visual and quantitative performance than other well-known background subtraction methods reported in this paper.
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.