在小波域进行基于模拟退火算法的图像阈值分割

IMAGE THRESHOLD SEGMENTATION BASED ON SIMULATED ANNEALING ALGORITHMS IN WAVELET FIELD

基于小波变换、模拟退火算法及二维最大类间方差(2DMVB)原理,提出1种新的应用于小波域的2DMVB金相图像分割方法。该方法借助"零树编码"的思想,首先在小波变换的低频域进行2DMVB分割,并采用模拟退火算法优化求解过程,然后根据低频域的分割结果,就能获得原始图像的最终分割结果。实验结果表明该方法分割效率和抗噪能力较2DMVB都有提高。

This paper presents a new 2-D maximum between-cluster variance(2DMVB) segmentation method applied in wavelet domain based on wavelet-transform,simulated annealing algorithms and 2DMVB.With the help of "zero-tree".firstly,this method does 2DMVB segmentation in the low frequency domain of wavelet-transform.and optimize the process of seeking threshold by simulated annealing algorithms based on the segmentation result of the low frequency domain we can get the final segmentation result of the original image.Th...