An Algorithm for the Hole Filling of Motion Foreground Based on Superpixel Segmentation

Abstract

For mainstream sports foreground extraction algorithm in complex scenarios to extract the prospect of movement target image problems is empty, this chapter to the extraction of the prospect areas and the prospects of a nearby empty area between nature has made the thorough analysis, put forward a kind of based on feature fusion of pixel segmentation combined with a significant test of hole filling algorithm. Since the SLIC superpixel segmentation has little difference in the color features of pixels, the moving foreground pixels will be divided into unified superpixel, resulting in the uneven interior of the region. This algorithm cleverly introduces the LBP texture feature, which can be better segmented according to the contour information of the moving foreground image. Finally, with the combination of saliency moving foreground target image filling, a more ideal moving foreground image is obtained.