基于萤火虫算法的二维熵多阈值快速图像分割

Fast image segmentation with multilevel threshold of two-dimensional entropy based on firefly algorithm

提出了基于萤火虫算法的二维熵多阈值快速图像分割方法以改善分割复杂图像和多目标图像时存在计算量大、计算时间长的问题。首先,分析了二维熵阈值分割原理,将二维熵单阈值分割扩展到二维熵多阈值分割。然后,引入萤火虫算法的思想,研究了萤火虫算法的仿生原理和寻优过程;提出了基于萤火虫算法的二维熵多阈值快速图像分割方法。最后,使用该方法对典型图像进行阈值分割实验,并与二维熵穷举分割法、粒子群算法(PSO)二维熵多阈值分割法进行比较。实验结果表明:该方法在单阈值分割、双阈值分割和三阈值分割时分别比二维熵穷举分割法快3.91倍,1040.32倍和8128.85倍;另外,在阈值选取的准确性和计算时间方面均优于PSO二维熵多阈值分割法。结果显示,基于萤火虫算法的二维熵多阈值快速图像分割方法能快速有效地解决复杂图像和多目标图像的分割问题。

A fast image segmentation method with multilevel threshold of two-dimensional entropy was proposed based on the firefly algorithm to overcome the large amount of calculation and long computing time.Firmly,the principle of two-dimensional entropy threshold segmentation was analyzed,and the single threshold segmentation of two-dimensional entropy was extended to multilevel threshold segmentation.Then,the bionic mechanism and searching optimization process of the firefly algorithm were analyzed,and the multilevel threshold segmentation method of two-dimensional entropy combined with firefly algorithm was proposed.Finally,typical image segmentation experiments by using the proposed method were performed and the results were compared with those of two-dimensional entropy exhaustive segmentation method and the multilevel threshold segmentation method of two-dimensional entropy based on Particle Swarm Optimization（PSO）.Experimental results show that the speeds of the proposed method in single threshold segmentation,dual-threshold segmentation and the three threshold segmentation are 3.91,1 040.32 and 8 128.85 times faster than those of the two-dimensional entropy exhaustive segmentation method respectively.Moreover,the threshold selection accuracy and running speed of the proposed method are both better than those of the multilevel threshold segmentation method of two-dimensional entropy based on PSO.Therefore,the fast image segmentation method with multilevel threshold of two dimensional entropy based on firefly algorithm can quickly and efficiently resolve complex and multi-target image segmentation problems.