改进的狼群优化二维Otsu阈值分割算法

Improved Wolf Group Optimization Two-dimensional Otsu Threshold Segmentation Algorithm

二维Otsu阈值分割是以每个灰度像素点的灰度值和周围灰度的均值作为分割标准,其通过概率统计的方法构建二维直方图,以离散度作为分割好坏的评判标准。针对二维Otsu阈值分割算法实时性差的缺点,将改进的狼群优化算法运用在分割过程中,加快了分割阈值的收敛速度,也提高了分割精度。通过对狼群优化算法游走行为中增加扰动算子可扩大全局搜索的范围,提升寻优的速度和精度。考虑到狼群围攻容易陷入局部最优的情况,研究引入了模拟退火算子,使该算法在寻优搜索过程中能够跳出局部最优。

The two-dimensional Otsu threshold segmentation is based on the average value of the gray value of each gray pixel and the surrounding gray.The two-dimensional histogram is constructed by the method of probability statistics,and the dispersion is used as the evaluation criterion of the segmentation.Aiming at the disadvantage of the poor real-time performance of the two-dimensional Otsu threshold segmentation algorithm,the improved wolf pack optimization algorithm is used in the segmentation process,which accelerates the convergence speed of the segmentation threshold and improves the segmentation accuracy.By adding the perturbation operator to the wolf optimization algorithm’s walking behavior,the scope of global search is expanded,and the speed and accuracy of optimization are accelerated.For the situation where the siege of the wolves is easy to fall into the local optimum,a simulated annealing operator is introduced to make the algorithm jump out of the local optimum during the optimization search process.