基于改进布谷鸟搜索算法的二维Tsallis熵多阈值快速图像分割

Fast Image Segmentation with Multilevel Threshold of Two-dimensional Tsallis Entropy Based on the Improved Cuckoo Search Algorithm

为了改善二维Tsallis熵在分割复杂图像时存在计算量大、耗时长、实用性差的问题,提出基于改进布谷鸟搜索算法的二维Tsallis熵多阈值图像分割方法.首先,分析了二维Tsallis熵单阈值分割原理并将其推广到多阈值分割,同时推导出了二维Tsallis熵多阈值选取公式.其次,借鉴逐维更新评价策略,同时加入逐维扰动策略来对布谷鸟搜索算法进行改进,并用于求解二维Tsallis熵函数的最优问题.最后,用穷举法、粒子群算法、布谷鸟算法以及改进的布谷鸟算法分别对典型图像进行多阈值分割实验并将分割效果、分割数据和各算法的收敛性能分别进行比较.实验结果表明,所提算法能够快速、准确地对复杂图像进行分割.

Concerning the computationally intensive,long computing time,poor practicability and other issues of complex image segmentation,a image segmentation method with multilevel threshold of two-dimensional Tsallis entropy was proposed based on the improved cuckoo search algorithm. Firstly,the principle of two-dimensional Tsallis entropy was analyzed and the single threshold segmentation was extended to multilevel threshold segmentation. Secondly,by victoria updated assessment and disturbance strategies were used to improve the cuckoo search algorithm,and was used to solve the optimal problem of two-dimensional Tsallis entropy function. Finally,typical image segmentation experiments by using the exhaustive threshold segmentation method,particle swarm optimization algorithm,cuckoo search algorithm and improved cuckoo search algorithm. The effects and data of image Segmentation and the convergence of the algorithm were analyzed and compared respectively. Experimental results showthat the improved algorithm can quickly and efficiently resolve complex image segmentation problems.