基于和声搜索优化的木材死节缺陷图像分割

Wood dead knot defects image segmentation based on harmony search optimization

针对木材表面死节缺陷,提出一种基于和声搜索优化(HSO,Harmony Search Optimization)的KAPUR死节缺陷图像分割算法。将RGB彩色图像转换成灰度图像,分别对R通道、G通道、B通道灰度图进行处理。将KAPUR中的前景和背景熵之和作为目标函数,将图像直方图内可行搜索空间中的随机样本编码为候选解,候选解通过进化迭代得到最优解,最后通过阈值分割得到目标。试验结果表明,算法能分割出木材表面死节缺陷,SD、Dice指数、ER、NR平均值分别为94.30%、97.07%、6.05%、0.00%。

Aiming at dead knot defects on wood surface,a KAPUR image segmentation algorithm based on Harmony Search Optimization(HSO)is proposed.The RGB color image is transformed into gray image,and the gray images of R channel,G channel and B channel are processed respectively.The sum of foreground and background entropy in KAPUR is taken as objective function.Random samples in the feasible search space of image histogram are coded as candidate solutions.The candidate solutions are obtained by evolutionary iteration.Finally,the target is obtained by threshold segmentation.The experimental results show that the algorithm can segment dead knot defects on wood surface.The average values of SD,Dice,ER and NR are 94.30%,97.07%,6.05%,0.00%,respectively.