Research Of Image Edge Detection Based On Mathematical Morphology Algorithm

In image processing and computer vision, edge detection occupies a special statue, the edge contains a large number of image information and reflects the main features of the object. Since the actual processing images always contain noise, but most of the classic edge detection algorithms are sensitive to noise, and will strengthen the noise in the edge detection with large amount of calculation. For this situation, it is very important to find an algorithm that can suppress the noise effectively and try its best to maintain the image edge details at the same time.The main purpose of this paper is to study the edge detection algorithm which is based on the idea and the characteristics of mathematical morphology, improving the existing edge detection algorithms, combined with the edge detection algorithms which are based on multi-structure and multi-scale separately, so as to construct an improved edge detection algorithm and apply it to the image scaling.The main research work is organized as follows:Firstly we give a detailed discussion of the basic theory of mathematical morphology and the property of the morphological operations, secondly we make an in-depth study of the existing morphological edge detection algorithm, combined with the current existing edge detection algorithms which are based on multi-structure and multi-scale respectively, we construct an edge detection algorithm based on multi-structure and multi-scale improve from them. It uses Mahalanobis distance to determine the weight of multi-structure edge detection result by image itself, and uses variance to determine the weight of multi-scale edge detection result by image itself respectively, then fuses multi-scale result and multi-structure result together to form the final image. Simulation experiments shows that the algorithm can filter out the noise of the image effectively while retains more image edge, and finally we apply this edge detection algorithm to the image scaling, the experiments shows it can improve the jagged and blurred edge availably, and we also extend it to the scaling of noisy image.