Study Of Image Edge Detection Method Based On Morphological Theory

Vision is a sensory produced by object images stimulating the retina and intuitions obtained in the cerebral cortex. Machine vision is also a visual device to the machine, which has been the visual fuction as human and aims to improve the machine automation and intelligence.It need to extract image information from objective things outside, and make the image processed correspondingly and understood and so that it can be applied in the actual test, measurement and control.Edge is one of the most basic and most important characteristics in visual image, and it contains much useful information such as location and contour. Most of image information exists in the edge because edge is between targets and targets, targets and background, region and region and primitive and primitive. Edge detection is a discontinuity or mutation detected technology based on gray or texture between object and background. It is the basis of analysis field in image segmentation, pattern recognition, machine vision and regional shape extraction,. Edge detection algorithm will affect directly the accuracy of contour extraction and system performance. Therefore, edge detection is more critical part in visual image and it has high significance in image location and target extraction.Traditional edge detection methods, mathematical morphology, improved fuzzy morphology and wavelet analysis based on morphology are proposed in this paper. Traditional edge detection methods owe their own characteristics, but they exist all kinds of defects simultaneously. All kinds of mathematical morphological edge detection have their own shortcomings and deficiencies. However, the improved morphological gradient operators have a certain resistance to noise, and the capacities of anti-noise are limited. Mathematical morphology is based on set theory, and can be used to handle nonlinear signal in our daily life because fuzzy morphology not only retains all the advantages of mathematical morphology, but also it is built on the basis of fuzzy set. Therefore, an improved edge detection algorithm based on fuzzy morphology is proposed in this paper. This algorithm has good edge detection effect and strong robustness of anti-noise, the extracted edge is more complete and edge contour is also quite clear and the edge image never appear missed detection and false edge. So this algorithm based on image edge detection has a good effect and prior to any other traditional edge detection methods. Because of the limited anti-noise capacity of mathematical morphology, mathematical morphology and wavelet are combined for image edge detection in wavelet analysis method based on morphology. The image is decomposed and fused respectively by wavelet transform so that the noise of original image can be filtered out more effectively and some details of the image features can be enhanced. Wavelet transform can filter out a lot of noise in the image, and morphological gradient operator can locate the image edge more accurately. Not only more complete edge information can be extracted from the image, but also it has a strong robustness of anti-noise. Therefore, this algorithm has better edge detection effect.