中文摘要：

图像的边缘检测是对图像进行进一步处理和识别的基础，虽然图像边缘产生的原因不同，但反映在图像的组成基元上，它们都是图像上灰度的不连续点或灰度剧烈变化的地方，这就意味着图像的边缘就是信号的高频部分。因此所有的边缘检测方法都是检测信号的高频部分。但在实际图像中，由于噪声的存在，边似部分和细节部分，近似部分是原图像对高频部分进行滤波所得的近似表示。经缘检测成为一个难题。提出一种利用小波把图像分解成近滤波后近似部分去除了高频分量，因此能够检测到原图像所检测不到的边缘。

英文摘要：

Image edge detection is the basic step in processing and identifying the image, though there are many different reasons for the cause of image edge, reflect in the basic element of the image. This means that the edge of image is the high frequency part of signal. So all the methods of image edge detection is to detect the high frequency part of signal. But in practical image, edge detection becomes a problem because of the using wavelet to transform the image into approximate express that undertake the filter of the high frequency part by original image. The ap- proximate part get rid of high frequency through the filter, so it can detect the edge which the original image can＇t detect.