Abstract:

 For color images, the state-of-the-art edge detection methods may result in loss of edges, imprecise edge positions, pseudo-edges or discontinuities. In order to solve these problems, an improved Canny edge detection algorithm is proposed. The algorithm combines photometric invariant differentials with color structure tensor to produce new image edge-gradient amplitudes. The canny detector based on new gradient amplitudes can detect edges of color images more precisely. Experiments demonstrate that the proposed method can well detect and classify variant color edges. The quality factor of edge detection evaluation shows that the color edge detection error rate has been reduced and the edge detection performance has been effectively improved

Key words: color invariance, Canny algorithm, color structure tensor, edge detection