基于纹理特征的盲道区域分割算法  

Blind Road Segmentation Algorithm Based on Texture Feature

作　　者：王民[1] 李媛[1] 张立材[1] 

机构地区：[1]西安建筑科技大学信息与控制工程学院,陕西西安710055

出　　处：《信息通信》2017年第7期23-26,共4页Information & Communications

基　　金：住房城乡建设部科学技术项目计划(2016-R2-045);陕西省教育厅专项基金(2013JK1081);陕西省科学技术研究发展计划项目(CXY1122(2));陕西省自然科学基金青年基金(2013JQ8003)

摘　　要：盲道分割是导盲系统中的重要一部分,核模糊C均值(KFCM)算法使用内核的方法来提高聚类的性能,然而,该算法具有初始化敏感,陷入局部极小和缺乏先验知识对核函数的最优参数等缺陷。若需进一步提高聚类性能,则需要克服上述问题。对此,文章提出了一种基于改进的KFCM算法,将生物地理学优化算法(BBO)与KFCM算法相结合。BBO算法没有聚类特性,主要是采用生物地理学中的迁移算子之间的信息共享解决方案,而KFCM算法受初值影响,对图像噪声很敏感,因此BBO算法可以对KFCM算法进行优化,从而提高聚类性能。实验结果表明,改进的KFCM算法较现有文献相比,可以对盲道区域进行更精确地分割,且聚类效果评价指数更优。Blind channel segmentation is an important part of the blind channel system. Thekemel fuzzy C-means （KFCM） al-gorithm uses kernel method to improve the performance of clustering. However, the algorithm has the defects of initial sensitiv-ity, local minima and lack of prior knowledge to the optimal parameters of kernel function. To improve the clustering perform- ance, it is necessary to overcome the problemsabove. In this paper, an improved KFCM algorithm is proposed, which combines biogeographic optimization algorithm （BBO） with KFCM algorithm. BBO algorithm does not have clustering characteristics, mainly uses the information sharing solution between migration operators in biogeography, and KFCM algorithm is affected by initial value and is sensitive to image noise. Therefore, BBO algorithm can optimize KFCM algorithm, and improving the clus-tering performance. The experimental results show that the improved KFCM algorithm can segment the blind regions more ac-curately than the existing ones, and the clustering effect evaluation index is better.

关 键 词：生物地理学优化算法 核模糊C均值算法 BBO-KFCM算法 图像分割 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]