A fuzzy approach to texture suppression by combining edge length analysis with surround suppression

In object recognition, texture suppression is often required to suppress texture edges while retaining contour edges. Texture edges are usually mutually close and contour edges are relatively isolated. According to this characteristic, surround suppression, which is a mechanism of human visual cells, can achieve texture suppression. In this paper, edge length is considered as another characteristic to discriminate texture edge from contour edge. Combining edge length analysis with surround suppression, a novel fuzzy approach is proposed which defines membership function for edges and computes suppression intensity accordingly. Experiments demonstrate the proposed approach can acquire better results than the approach which only uses surround suppression.