基于局部相位信息的图像边缘检测方法研究

ue to the important role of phase information in image structure, this paper proposes an image edge detection algorithm which is based on local relative-phase information. Firstly, the dual- tree complex wavelet transform is utilized to decompose the image into multiscale and multi-directional subbands. Then, the phases of coefficients in each directional subband is calculated and the corresponding relative-phase information is further obtained. According to the linear relationship of relative-phase and edge angle, the edges in directional subbands are extracted. Finally, an inverse dual-tree complex wavelet transform is conducted to produce the edge image and the binary edge image is obtained via dual-threshold method. Experimental results shows that, comparing with Canny edge detector the proposed algorithm has better noise robustness and extracts more accurate edges.