Edge Detection Algorithm Research And Application Of Fused Canny Operator And Morphological Operator

The image edge is the key features of the image in which contains the location of the image, information distribution proportion and contour structure characteristics. Therefore around detection, extraction of the image edge and contour feature, some new thoughts, new technology constantly have been discovered, moreover contour extraction of the image edge has been also widely recognized as one of a classic research project in the field of computer vision and image processing.Edge detection difficulty lies in how to filter out the noise and meanwhile to clarify the edge. In this paper, firstly analyzed and researched the traditional Canny operator and morphological algorithm, and then put forward a new edge detection algorithm. The research working is as follows.First, In the image edge detection process, Canny operator will appear two more difficult to reconcile the problem that Gaussian filter parameters are difficult to set up and detection process cannot be excluded noise region. Therefore, This paper use morphological filtering and sub-8-neighborhood gradient magnitude calculation method to extract the image edge, Experiments confirmed that using this method can not only improved in terms of MSE and PSNR, it is possible to retain more edge pixels, but also get a relatively good subjective visual experience.Secondly, Traditional morphological methods for image edge detection results contain a high concentration of noise is not very good. Therefore, this paper presents an edge contour extraction method based on adaptive morphology. The method first uses 3×3 cross and 5×5 diamond matrix elements to reduce the noise of image. Then it adopts the four different directions of the matrix elements to make the image morphology edge detection and set weights in four directions according to the image profile feature, At last, This method calculate a weighted sum of results that is extracted from different directions through the corresponding weights. Experiments confirmed that improved morphological edge detection method in terms of subjective vision and objective evaluation index get outstanding results, both to locate the real image's contours, but also to retain more edge pixels in the process of removing the noise.Finally, the wavelet transform theory is used to combine the advantages of the two methods. Experimental results show that, whether in the subjective visual perception or in the objective data are better than before the fusion of the results of a single method. In the application of this method in Thangka image edge extraction eventually also won the good extraction effect.