基于改进TSO优化图像熵的图像识别与分割

Image recognition and segmentation based on improving TSO optimizing image entropy

提出一种精英学习改进TSO优化Tsallis熵的遥感图像多阈值分割方法。在传统TSO算法中引入精英对立学习实现种群初始化,利用权重系数非线性更新均衡算法全局搜索与局部开发,引入精英池策略提高收敛精度。以Tsallis熵作为适应度评估个体优劣,利用改进算法搜索阈值最优解。利用4幅遥感图像验证算法有效性,改进算法在峰值信噪比、结构相似度、特征相似度和计算效率上表现更好,能有效提升图像识别精度和分割效率。

A multi threshold segmentation method for remote sensing images using elite-learning improved tuna algorithm optimizing Tsallis entropy was proposed.The elite opposite-learning was introduced to achieve population initialization in traditional tuna algorithms,a nonlinear update of weight coefficients was used to balance the global search and local development of the algorithm and an elite pool strategy was used to improve the convergence accuracy.Tsallis entropy was used as the fitness function to evaluate the quality of individuals,and the improved algorithm was used to iteratively search for optimal solutions of image segmentation threshold.The effectiveness of the algorithm was verified using four remote sensing images.It is confirmed that the improved algorithm performs better in peak signal-to-noise ratio,structural similarity,feature similarity and computational efficiency,and can effectively improve the recognition accuracy and segmentation efficiency of remote sensing images.