基于多尺度小波变换的图像边缘检测

Image Edge Detection Based on Multiscale Wavelet Transform

图像边缘是图像中的重要信息,为了检测图像中的边缘信息,提出了一种基于多尺度小波变换的图像边缘检测算法.该算法充分利用了图像边缘在多尺度下的信息,首先选用二次B样条小波对原始图像进行多尺度小波分解,提取出图像中的高频信息,包括真实的图像边缘和噪声,然后根据图像边缘和噪声在不同尺度下具有不同的传递性,抑制噪声分量,保留图像边缘分量.实验结果表明,该算法获得了较好的图像边缘检测效果.

Image edge is the important information to an image, and an algorithm based on multiscale wavelettransform is proposed to detect image edge. In tins algorithm, image edge information in multiscale can be fully utilized. Firstly, muhiscale wavelet transform with quadric B - spline wavelet is carried out to extract high frequency information ,including true image edge and noise;Secondly, according to the different transitivity of true imageedge and noise in different scales, the noise can be suppressed and true image edge can be kept. The experimentalresult shows that the image edge can be obtained effectively by this algorithm.