基于改进PSO的三维Tsallis熵图像分割  

Three Dimensional Tsallis Entropy Image Segmentation Based on Improved PSO

作　　者：吴浩然 WU Hao-ran(School of Mathematics,Hefei University of Technology,Hefei 230601,China)

机构地区：[1]合肥工业大学数学学院,安徽合肥230601

出　　处：《计算机技术与发展》2023年第3期41-48,共8页Computer Technology and Development

基　　金：国家自然科学基金(61872407)。

摘　　要：针对二维Tsallis熵图像分割不精确以及优化图像阈值分割函数的元启发式优化算法容易陷入局部最优这两个问题,提出了一种新的三维Tsallis熵阈值分割法以及一种新的改进粒子群优化算法。通过引入均值、中值、梯度三种因素,构建出三维直方图,并结合Tsallis熵理论提出了一种三维Tsallis熵阈值分割法。为了避免粒子群优化算法陷入局部最优,通过引入综合学习策略并改进粒子群优化算法的迭代方式,提出了综合学习改进粒子群优化算法。将提出的三维Tsallis熵阈值分割法与综合学习改进粒子群优化算法结合进行图像分割。与其他元启发式算法相比,综合学习改进粒子群优化算法能在低维环境下有效避免局部最优。实验结果表明相比于二维Tsallis熵阈值分割法,三维Tsallis熵阈值分割法分割效果更好,且具有更好的抗噪性能。由此可以表明综合学习改进粒子群优化算法结合三维Tsallis熵进行图像分割可以取得更好的结果。Aiming at the two problems that the two-dimensional Tsallis entropy image segmentation is inaccurate and the meta heuristic optimization algorithm for optimizing the image threshold segmentation function is easy to fall into local optimization, a new three-dimensional Tsallis entropy threshold segmentation method and a new improved particle swarm optimization algorithm are proposed. By introducing three factors: average, median and gradient, a three-dimensional histogram is constructed, and a three-dimensional Tsallis entropy threshold segmentation method is proposed combined with Tsallis entropy theory. In order to avoid the particle swarm optimization algorithm falling into local optimization, comprehensive learning improved particle swarm optimization is proposed by introducing comprehensive learning strategy and improving the iterative method of particle swarm optimization algorithm. The proposed three-dimensional Tsallis entropy threshold segmentation method is combined with the comprehensive learning improved particle swarm optimization algorithm for image segmentation. Compared with other meta heuristic algorithms, the comprehensive learning improved particle swarm optimization algorithm can effectively avoid local optimization in low dimensional environment. The experimental results show that compared with the two-dimensional Tsallis entropy threshold segmentation method, the three-dimensional Tsallis entropy threshold segmentation method has better segmentation effect and better anti noise performance. It is showed that the improved particle swarm optimization algorithm combined with three-dimensional Tsallis entropy can achieve better results.

关 键 词：粒子群优化算法 TSALLIS熵 图像分割 综合学习策略 三维直方图 

分 类 号：TP391.4[自动化与计算机技术—计算机应用技术]