An improved edge detection algorithm based on mathematical morphology and directional wavelet transform

Traditional methods of the edge detection can´t completely extract the low frequency image edge. It is easy for them to discard some important information in the high frequency sub-images. If there is much noise in the image, these methods can not entirely eliminate the noise. We will get the poor image edge because of the noise. In order to solve these problems, we respectively analyze the theories of the mathematical morphology algorithm and the directional wavelet transform. We propose an improved edge detection algorithm based on the mathematical morphology algorithm and the directional wavelet transform. We firstly use the binary mathematical morphology to de-noise and detect the edge of the low frequency sub-image in the wavelet domain. Then we de-noise the high frequency sub-image in horizontal, vertical and diagonal directions. We adopt the Canny operator to detect the edge of the high frequency image. The low frequency edge image and the high frequency edge image can be respectively got by the above algorithms. Finally we can get a complete and continuous edge by using some fusion rules. The results show that the improved edge detection algorithm can well suppress the noise interference. It also can get a more continuous and complete edge image comparing to other methods.