Research On Images Edge Detection Method Base On Local Energy

Edge detection is an important part of image processing.It is widely used in image segmentation,industrial monitoring,intelligent traffic video surveillance and many other fields.As its accuracy have a great impact on the subsequent image processing,so research on improving the accuracy has been a main trend.Common edge detection methods are based on differentiation,which determine the edges through detecting the maxima of the first derivative or the zero-crossing of the second derivative.However,the performances of the edge detection methods based on differentiation are extremely dependent on the smoothness of actual edges.Considering that the integration does not depend much on the smoothness of data,this paper proposes a new edge detection method based on local energy,which is aroused by integral calculation.Firstly,we gave a detailed description of the definition of local energy and explained the changing process of local energy values.According to the description of local energy,the local energy value of the edge point is much larger than the local energy value of the pixel point in the smooth region.Therefore,based on the constructed local energy function,the location of the image edge can be judged by analyzing the local energy value,so that the edge point can be found effectively.Secondly,we make a theoretical analysis of the proposed edge detection method and discuss the process of theoretical derivation in detail.Then,we briefly introduce the selection of the integration direction and the detection process in the proposed method,especially the boundary tracking method used in the detection process.Finally,the performance evaluation of our method is divided into two aspects.On the one hand,the edge detection result of our method is compared with the traditional edge detection methods in the visual to verify the effectiveness of our method.On the other hand,the three kinds of objective evaluation methods based on connectivity component,Pratt's figure of merit and Baddeley error measure are used in this thesis based on the images in BSBD.Among them,the continuity of the edge detection results is measured by the evaluation method based on the connectivity components,and the accuracy of edge detection results is measured by the Pratt's figure of merit and Baddeley error metric.Experimental results show that the performance of our method is better than others both in the effectiveness and in theaccuracy of the edge detection results.