基于小波尺度积的图像边缘检测算法

Approach of image edge detection based on wavelet scale multiplication

图像边缘检测技术是图像处理中最重要的内容之一,针对噪声大、灰度对比度低的图像,利用小波尺度乘积模极大值对其进行边缘检测。在进行二进小波变换后,先把相邻尺度的小波系数直接相乘,抑制噪声和增强图像边缘信息,再运用模极大值检测图像边缘,利用隶属函数确定阈值。该算法可以免除传统小波模极大值边缘检测后繁杂的边缘链接,简单易实现,实验证明其对抑制噪声和增强弱边缘有良好的效果。

Edge detection is one of the most important techniques in image processing. A new image edge detection approach is presented by wavelet scale multiplication, which has the advantage of magnifying the edge structures and suppressing the noise. The dyadic wavelet transforms at tow adjacent scales are multiplied as product function and image edge is detected by wavelet model maximum. It need not link edge points, realize simply and is usually applied to noisy, low contrast image.