蚂蚁算法实现图像分割论文.docVIP

蚂蚁算法实现图像分割 摘 要 模式识别领域图像分割是图像处理的底层步骤，它对于图像识别和图像理解具有举足轻重的作用。一个好的图像分割算法对图形图像的处理会有意想不到的效果，因此探索更好的图像分割算法是图像处理领域中一个重要的研究方向。 通过对蚂蚁、蜜蜂和其他群居昆虫的观察和研究，人们逐步发展了群体智能的概念。行为简单的个体聚合在一起就形成了具有某种智能的群体。群体所拥有的能力超过了个体能力的总合，其原因就是群体间存在着“协同合作”。 本文将通过对蚂蚁观察和基于MARCO DORIGO提出的蚂蚁算法研究，用于对图像分割问题的探索。将灰度图作为蚂蚁群落的栖息地、通过让蚂蚁个体具有的感知其周围有限的范围环境的激素状况和图像的边缘情况的简单能力，依据蚂蚁群觅食的基本原理，实现对图像的分割。 实验中将对不同种类边缘图像的测试，证明该算法能够处理各种类型的图像边缘。通过与传统处理的对比实验，证明其能够在多方面获得更好的效果。 关键词：蚂蚁算法; 群体智能; 图像分割; 边缘检测 The ant system realizes image segmentation Abstract In pattern recognition system, image segmentation is a low-level image-processing. This task is very important, as we know, since the output of an image segmentation algorithm can be fed as input to higher-level processing tasks. Acquiring the better image segmentation algorithm has always been an important subject in this filed. Since people have studied the behavior of the ant colony and other social insert, the notion of social intelligence has been introduced. People find that the colony’s collective behavior exceeds the sum of its individual member’s actions. The main reason of this kind of phenomenon is something we called synergy. This paper discusses the method of using ant system, which is also a social intelligence algorithm, to solve image segmentation problem. Grey level image can be regard as a topographic map where the image is the ant colony habit. The respective gray level intensities and pheromone intensities are considered in each ant perception of his neighborhood. Following the rules of foraging activity of the real ants, we obtained the algorithm for image segmentation problem solved by the ant colony. In our experiments, different types of images have been used proving that this algorithm can commendably obtain their edge-images. Comparing with the result of those classical algorithms, we found that ant system algorithm obtained a better result in many aspects. Keywords：Ant system；Social intelligence；Image segmentation目 录 摘 要 I Abstract II 目 录 1 1 绪 论 1