Abstract:

In order to solve the problem that the traditional image edge detection techniques can only emphasize the horizontal and vertical edges, a new image edge extraction method combining the multi-resolution analysis of nonseparable additive wavelet transform and the mathematical morphology has been proposed. A low-pass filter of nonseparable wavelet is constructed and the source image is decomposed into a low-frequency sub-image and high-frequency sub-images. The low-frequency sub-image is filtered to obtain morphological gradient map. The high-frequency sub-images are added to form an enhanced high-frequency edge map by using modulus maxima value method. Inverse transform of nonseparable additive wavelet transform is performed on the gradient map and the edge map and an edge-gradient map is produced. The final edge map is got by binarizing the edge-gradient map. The experimental results show that the proposed method has good visual effect. When compared with the traditional image edge detection techniques, it can extract image edges with integrity, multidirection, shift invariance and high speed.

Key words: image processing, edge detection, nonseparable additive wavelet, morphology