论文部分内容阅读
为了自动确定遥感图像分割的最佳阈值,本文提出了一种改进的自适应遗传算法,并利用该算法对二维Otsu图像阈值分割函数进行了全局优化,提高了分割阈值的求解速度。该算法能够根据个体适应度大小和群体的分散程度自动调整遗传控制参数,从而能够在保持群体多样性的同时加快收敛速度,克服基本遗传算法的收敛性差、易早熟问题。实验结果表明,该算法具有良好的收敛速度和稳定性,达到较好的图像分割效果,大大缩短了计算时间。
In order to automatically determine the optimal threshold for remote sensing image segmentation, this paper proposes an improved adaptive genetic algorithm, and uses the algorithm to global optimization of the threshold segmentation function of two-dimensional Otsu image to improve the speed of the segmentation threshold. The algorithm can automatically adjust the genetic control parameters according to the degree of individual fitness and the degree of population dispersal, which can speed up the convergence rate while maintaining the population diversity and overcome the poor convergence and precocious problems of the basic genetic algorithm. Experimental results show that the proposed algorithm has good convergence speed and stability, achieves better image segmentation results and greatly reduces computation time.