基于蚁群智能和支持向量机的图像分割方法  

Image segmentation based on ant colony intelligence and SVM

作　　者：吴海珍[1] 何伟[1] 蒋加伏[1] 齐琦[1] 

机构地区：[1]长沙理工大学计算机与通信工程学院,湖南长沙410076

出　　处：《计算机工程与设计》2009年第2期408-410,共3页Computer Engineering and Design

基　　金：湖南省自然科学基金项目(06JJ50109)

摘　　要：针对传统阈值法在图像分割时仅考虑像素的灰度信息,对噪声敏感且不易确定全局最优阈值的缺点,提出了基于蚁群智能和支持向量机的图像分割方法。该方法利用蚁群优化算法对图像的二维阈值空间进行全局搜索,并将搜索到的最优点灰度—区域灰度均值对作为阈值来区分目标和背景,然后对支持向量机进行训练和测试,最后用训练好的支持向量机分割图像。实验结果表明,该方法抗噪能力强,分割准确,是一种实用有效的图像分割方法。A method for image segmentation based on ant colony intelligence and SVM is proposed. It aims at the disadvantages of the traditional threshold methods, such as only considering the gray information of pixels; sensitive to noise; difficult to determine the global optimal threshold and so on. First, ant colony optimization algorithm is used to search the 2-D threshold space of the image. Second, the optimal pixel-scale gray is used as the threshold to distinguish the target from the background. Third, SVM is trained and tested. Finally, the image is segmented by trained SVM. The experiment simulation results show that this method is an applied and effective method for image segmentation with high ability of resisting noise and exact effect of segmentation.

关 键 词：图像分割 蚁群智能 优化算法 支持向量机 二维阈值 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]