基于提升小波和形态学融合的图像边缘检测

Edge Detection Based on Fusion of Lifting Wavelet and Morphology

针对传统边缘检测方法边缘定位不精确的缺点,结合提升小波变换和数学形态学的优势,提出了一种融合提升小波和多尺度形态学的边缘检测方法。首先,对原始图像进行提升小波变换;再用小波变换和多尺度形态学算子分别对低频图像进行边缘检测,根据异或原则融合成低频边缘;然后,用小波变换检测高频图像边缘;最后,通过提升小波反变换得到边缘图像。实验结果表明,与传统或其他的形态学边缘检测方法相比,该算法在保持图像边缘清晰的同时,具有很强的边缘定位能力。

Considering the disadvantage of the inaccurate edge location in the traditional edge detection methods,an edge detection method is proposed which based on the fusion of lifting wavelet transform and multi-scale morphology according to the advantage of lifting wavelet transform and mathematical morphology. Firstly,lifting wavelet transform is applied on the original image, and then successively wavelet transform and multi-scale morphological op- erators are used to extract edge of the low-frequency image, so the low-frequency image is fused according to the xor rule. Then wavelet transform is ap- plied to detect the high-frequency edge. Finally, edge image is got through the lifting wavelet inverse transform. The experimental results show that, com- pared with traditional or other morphological edge detection methods, the proposed algorithm has a strong edge positioning capabilities while it keeps the sharp-edged image.