Hierarchical background subtraction algorithm using Gabor filter

Detection of a moving object in a non-stationary scene is a very critical and challenging step for visual surveillance applications. Background subtraction (BS) is a widely used algorithm for moving object detection in the presence of static cameras. Its performance purely depends on the choice of features used for background modeling. The traditional BS assumes that the background is static or near-static. This assumption does not hold for practical scenarios and their performance degrades in the presence of non-stationary scenes such as swaying of trees, sprouting of water from fountain, ripples in water, flag fluttering in the wind, camera jitters and noise. In this paper, we have combined both block-based and pixel-based approaches in our hierarchical BS algorithm. Both coarse and fine level background modeling is done using the magnitude feature obtained from the Gabor filter. First coarse level background modeling is done for identifying the blocks which are fully or partially occupied by the foreground objects. Every pixel in the foreground blocks is further classified using the Gabor feature for improving the precision of the detected moving object. The proposed algorithm is a single modal based BS scheme. Quantitative and qualitative results justify our algorithm for efficient moving object detection in the presence of complex dynamic backgrounds.