基于改进二进制小波变换的图像边缘检测算法

Edge Detection Algorithm for Image Based on Improved Dyadic Wavelet Transform

针对传统二进制小波变换在图像边缘检测应用中的不足,提出了基于改进二进制小波变换的图像边缘检测算法.该算法首先按水平、垂直和对角方向对图像进行改进的多尺度二进制小波变换,提取三个方向的小波系数.然后采用相邻尺度小波系数相乘的方法去除图像的噪声,再对去噪后的小波系数乘积极大值点进行检测.最后将这3个方向上的极大点进行融合,形成图像的边缘.

In view of the shortage of traditional dyadic wavelet transform on the application of edge detection for image, the edge detection algorithm is presented based on improved dyadic wavelet transform for image in the paper. The processes of the algorithm are as follows： firstly improved multiscale dyadic wavelet transform is utilized with horizontal, vertical and diagonal directions respectively to obtain wavelet coefficients of the three orientations. Secondly the method of coefficients multiplication of wavelets in adjacent scales is applied to remove noise and to take out the maxima of the wavelet coefficients product after de-noising. Lastly merge the maximum points of the three directions to conform the image edges.