Thangka image edge detection algorithm based on morphology and RCF

Thangka image includes rich content and complex texture. Because the contour of Thangka image contains abundant image data information, edge detection is very important in the analysis of Thangka image. The edge extracted by mathematical morphology method is smooth and continuous, but fuzzy and unclear phenomena exist in the detection of complex edges. CNN can be used to extract much high-level, multi-scale information [1]. Therefore, this paper proposed edge detection method to extract the original image edge by using the optimized mathematical morphology algorithm and the trained RCF network model [2]. Then, according to the decomposition and reconstruction principle of wavelet transform, the edge images obtained by the two methods are fused. Experiments show that compared with the traditional method, the fused image edges are more clear and continuous, the contour information removes the invalid detail texture, which is more in line with human visual cognition, and is more conducive to the follow-up study of Tangka image.