Sacon: A consensus based model for background subtraction

Abstract

Statistical background modeling is a fundamental and important part for many visual tracking systems and other computer vision applications. This paper presents an effective and adaptive background modeling method for detecting foreground objects in both static and dynamic scenes. The proposed method computes SAmple CONsensus (SACON) of the background samples and estimates a statistical model per pixel. SACON exploits both color and motion information to detect foreground objects. SACON can deal with complex background scenarios including non-stationary scenes (such as moving trees, rain, and fountains), moved/inserted background objects, slowly moving foreground objects, illumination changes etc. Numerous experiments on both indoor and outdoor video sequences show that the method is robust to various types of background scenarios and, compared with several state-of-the-art methods, can achieve very promising performance.