基于遗传算法的三维重构图像阈值分割

Image Threshold Segmentation of 3D Reconstruction Based on Genetic Algorithm

图像阈值分割是三维重构中图像处理的一个重要内容.最佳熵阈值的图像分割具有很多优点,但同时也需要大 量的运算时间,从而限制了其实际应用.将遗传算法应用于最佳熵阈值的确定中,提出一种新的图像阈值分割方法,以灰度 图像的直方图熵作为评价标准,把图像阈值分割问题定义为一个优化问题,利用遗传算法寻优的高效性,搜索到能使分割质 量达到最优的分割参数———图像分割阈值.实验结果表明,采用遗传算法不仅可以实现正确的图像分割,而且使得分割速度 大大提高.

Image threshold segmentation is of important significance for image processing of 3D reconstruction. The optimal entropy thresholding algorithm has many advantages in the use of the image segmentation, but it needs a lot of computational time, thus limits its practical application. This paper applies the genetic algorithm in determining the optimal entropy threshold, and presents a new approach of image threshold segmentation. Entropy of histogram of gray-level images is taken as the segmentation criteria and the problem of image threshold segmentation is formulated as an optimal problem. The efficiency of genetic algorithm searching ensures the best segmentation parameter——image segmentation threshold. The experimental results show that the genetic algorithm not only achieves the right image segmentation but also greatly improves the segmenting speed.