基于小波变换的多尺度图像边缘检测算法

Multi-scale Edge Detection Method Based on Wavelet

针对边缘检测算子的抑噪能力和定位精度之间矛盾,提出了一种基于B样条小波变换的边缘检测方法。首先对图像进行小波多尺度分解,在每种尺度下分别提取图像边缘;而后利用边缘信息的多尺度特性,融合多尺度边缘得到了单像素宽边缘。通过计算机仿真对该方法进行验证,实验结果表明该方法不仅能准确检测出图像边缘,而且能有效地抑制噪声,效果优于传统的边缘检测算法。

To the contravention between traditional edge detectors＇ ability of restrain noise and orientation precision, a detection algorithm based on B-spline wavelet is proposed. Firstly, image edge is detected at multi-scale, and then the edge with only a pixel width is obtained by fusing different scale edges with the muti-scale characteristic of edge information. Computer simulation results show that the proposed method can not only detect image edge precisely, but also effectively restrain noise. Comparing with several traditional algorithms, the experiment results show the feasibility and effectiveness.