基于混合杜鹃搜索算法的图像二维熵阈值方法

An Image Segmentation Approach Using Hybrid Cuckoo Search Algorithm and 2-D Entropy

二维最大熵阈值法利用了图像的空间位置信息具有较好的分割结果,然而大量运算降低了它的效率。提出一种混合杜鹃搜索算法和局部搜索的二维最大熵阈值分割方法。仿真结果表明与基于遗传算法、粒子群算法、差分进化算法优化的二维最大熵阈值方法相比,提出的方法可以快速的获得图像二维最优分割阈值并且能够避免局部最优阈值的情况,显著的降低了基本二维最大熵阈值的执行时间,是一种快速且性能鲁棒的图像阈值分割方法,能够满足图像分割的实时性要求。

Thresholding based on 2 - D maximum entropy utilizes space information of images and has good segmentation performance, however, heavy computation reduces its efficiency. In this paper, a segmentation approach was proposed based on 2 - D maximum entropy, which employs the hybrid cuckoo search algorithm combined with a local search to obtain optimal thresholds of 2 - D maximum entropy thresholding. Experimental results display that compared with 2 - D maximum entropy thresholdings optimized with genetic algorithm, particle swarm optimization al- gorithm and differential evolution algorithm, the proposed method can achieve the optimal thresholds quickly, avoid traps of local optimal thresholds and significantly decrease the seeking time of optimal thresholds. The proposed method is a fast and robust image segmentation method, which is suitable for real - time image segmentation.