小波模极大值法与数学形态学边缘检测细化结果

图像边缘检测的关键是尽可能多的检测到边缘并且抑制噪声的同时,尽可能的满足单线的边缘定位精度；为此选取了一种融合小波模极大值和数学形态学的边缘检测方法来获取图像边缘；首先在对图像进行小波分解,分别利用模极大值法和多尺度多结构数学形态学方法来处理小波分解的高频分量和低频分量,利用差影法对二者的结果进行融合；然后利用大律法得到二值化图像,并用形态学边缘细化算法细化图像边缘得到最后结果；实验结果显示,融合的方法可以得到比较完善的边缘,经过二值化和边缘细化后,获得的单线宽边缘更加清晰,定位精度更高。

Wavelet Modulus Maxima Method and Mathematical Morphology Edge Thinning Results

The key to image edge detection is to detect edges as much as possible and suppress noise effectively. Finally we want to meet the single-line edge positioning accuracy. For the purpose, a new edge detection method based on wavelet transform modulus maxima and mathematical morphology is selected to get the edge of image. First, the original image is decomposed by wavelet decomposition. After that, wavelet modulus maxima method and multi-scale and multi structure mathematical morphological method were used to deal with high-frequency components and low frequency components separately. Then, the two edges processed before were fused by sub-subtraction method and morphological edge thinning algorithm is used to get the thinning edge of the image. The experiments show that the fusion method can get more perfect edge, the single-line edge is more clear, and the positioning accuracy is improved.