Research of Adaptive Frame Difference Moving Target Segmentation Based on MRF

Abstract

Moving target segmentation which is an important research subject in many fields, such as artificial intelligence, industrial monitoring and human-computer interaction, the research of moving target segmentation algorithm in video sequence has vital academic significance and practical value. This paper introduced Markov Random Field theory into image sequence target segmentation process which is based on initial segmentation in static scene. In the paper, we constructed Adaptive Markov Random Field algorithm combining with adaptive Otsu method, namely the AMRF model. And also extracted the inter-frame difference mask by this model and inter-frame difference method. Experimental results show that the algorithm is effective and the moving target segmentation effect by this model is desirable.