Edge Detection Algorithm Research On Noise Image

Image edge contains much detail information of the image, edge detection technology is widely used in many engineering, target feature extraction and feature analysis must be based on the accurate image edge. Focusing on the problems of noise interference and edge blur exist in noise images, two algorithms were proposed in this paper:First, wavelet transform has two nature characteristics of the time-frequency local feature and multi-scale feature. The wavelet transform is an effective tool to solve non-stationary problems. The larger the scale is, the more noise be filtered out, but at the same time, the more edge information loss; The smaller the scale is, the more edge information preserved, but the more noise information preserved as well. So there also exist a problem of the scale selection.This paper compared the edge images that under different scales, and then proposed a noise image edge detection algorithm based on wavelet transform. The algorithm first improved the wavelet transform modulus maxi-ma gradient and then applied the improved threshold method to the edge image obtained. Through MATLAB simulation to validate the algorithm, for noise image, the improved algorithm which this paper proposed is superior to the traditional algorithms, the aspect of detection effect is improved as well.Second, Edge detection is becoming the more and more crucial technology in dealing with medical image, through the edge’s positioning of medical image, normal or not can be easy to judge. The result can be a reference for diagnostic, the pros or cons of the result directly affect the treatment of the next step. Due to the influence of noise and complex biological tissue organs, the medical image edge of clear outline and clear organization is hard to get.Mathematical morphology is a new field method which be based on structure elements,and this paper proposed an edge-detection algorithm of medical noise image based on morphological gradient combined with multiple structure elements and multi-scale features.The algorithm first used a combination of open operator and close operator to smooth filtering,and then used the morphological gradient that the paper proposed from the four directions of00、450、900、1350 to obtain image edge, the paper gave the concrete steps and applied to the MRI image. The MATLAB simulation result shows that the algorithm is good on anti-noise performance, the resolution of image edge-detection based on morphological gradient is improved, the algorithm has a certain practicality.