Improved Edge Detection Algorithms Based On Multi-direction Structural Elements In Multi-scale And Morphology

Edge detection is an important preprocessing step in computer vision and multimedia technology. An image can be divided into background and objects. The task of edge detection is to locate and extract the edge pixels between objects and background or the edge pixels between objects by analyzing the image.We first discuss the mathematical morphology edge detection method proposed by Serra in this paper. Mathematical morphology, which studies the morphological characteristics of an image, is a non-linear filtering method based on geometry and set theory and composed of a series of morphological operators. It has advantages of simplifying image and keeping the basic features of image, and it is not sensitive to edge directions and has better anti-noise performance.Next, we present some improved suggestions based on the traditional morphological gradient edge detection algorithm in this paper. The traditional algorithm is sensitive to noise, so we give an improved set of morphological filter to improve the performance of anti-noise. The traditional algorithm based on multi-direction structure elements artificially decides the synthetic coefficient, so we present an improved synthetic coefficient edge detection algorithm based on multi-direction structure elements. The different scale of structure elements impacts the detection results, so we combine our algorithm with the existing morphological edge detection algorithm based on multi-scale structural elements. The inverse operation of membership function in Pal. King fuzzy enhancement algorithm has no solution, so we use a Gaussian membership function and the maximum between-class variance to improve the enhancement algorithm, then combine the algorithm with our algorithm to detect edges. The experiment result shows that the improved edge detection algorithms are accurate and effective than previous ones.Finally, we apply the improved edge detection algorithms to RGB color space to detect the color image edges. The problem is that when the grey-level images of RGB channel combined with equal probability, the result image would has wider edges, so we give an improved fusion coefficient self-adaptive edge detection algorithm based on grey threshold. The experiment results show that our algorithm improves the performance of edge detection and it can also refine edges.