基于图像预处理分类分割的盲道分割算法  

Blind Sidewalk Segmentation Algorithm Based on Image Preprocessing Classification and Segmentation

作　　者：刘灵 孙晨晨[1] 徐寅林[1] 唐万春[1] 赵华[1] Liu Ling;Sun Chenchen;Xu Yinlin;Tang Wanchun;Zhao Hua(School of Physics and Technology,Nanjing Normal University,Nanjing 210023,China)

机构地区：[1]南京师范大学物理科学与技术学院

出　　处：《南京师范大学学报（工程技术版）》2020年第1期42-48,共7页Journal of Nanjing Normal University(Engineering and Technology Edition)

基　　金：江苏省高等学校自然科学基金项目(18KJB510020);江苏省研究生科研与实践创新计划项目(SJCX18-0362)

摘　　要：高准确分割率的盲道分割算法是实现高性能导盲系统的重要保障.提出一种基于颜色直方图支持向量机方法,将盲道预分类为颜色盲道或纹理盲道,采用对HSV颜色空间的多参数融合的改善OTSU分割方法处理颜色盲道分割,采用基于纹理增强的K均值聚类方法处理纹理盲道分割.通过有效的预处理分类,可以针对性地根据盲道的颜色或纹理特征进行识别,同时由于采用了改善的颜色和纹理分割算法,极大地改善了对不同种类和环境下的盲道的适应能力.测试证明,该方法对于测试库图片平均分割准确率可达到90%以上.The blind sidewalks object segmentation algorithm with high accurate segmentation rate is an important key to a high performance blind guided system.In this paper,a color histogram support vector machine(SVM)method is proposed to classify blind sidewalks into color blind sidewalks or texture blind sidewalks.For the color blind sidewalks,an improved OTSU segmentation method is proposed for multi-parameter fusion of HSV color space.For the texture blind sidewalks,based on texture enhancement,K-means clustering method is proposed.Because of the effective preprocessing classification method,they can be recognized according to the color or texture features of blind sidewalks.Besides the improved color and texture segmentation algorithm has excellent adaptability for different kinds of blind sidewalks under the different environments,and the average segmentation accuracy of the blind sidewalks can reach more than 90%for the test library pictures.

关 键 词：预处理分类 颜色直方图 支持向量机 阈值分割 

分 类 号：TP391.4[自动化与计算机技术—计算机应用技术]