一种基于颜色纹理与SVM的盲道分割算法  

Blind Road Segmentation Algorithm Based on Color Texture and SVM

作　　者：陈都 王雷[1] 方天宇 CHEN Du;WANG Lei;FANG Tian-yu(School of Electronic and Optical Engineering,Nanjing University of Science and Technology,Nanjing 210094,China)

机构地区：[1]南京理工大学电子工程与光电技术学院,南京210094

出　　处：《软件导刊》2018年第11期77-80,85,共5页Software Guide

摘　　要：为了帮助盲人更好地利用盲道,需要将盲道从复杂的前方环境图像中提取出来,提出一种基于颜色纹理和SVM的盲道分割算法,首先利用SVM对样本进行特征训练,再利用训练后的SVM数据模型对输入的图像进行判别,从而将盲道部分提取出来。通过对比选取了HSV颜色空间的颜色特征和3个频率、2个方向角的Gabor滤波器组样本纹理特征,再将其输入到SVM分类器中训练。结果表明,相较于现有算法,该盲道分割算法具有更加稳定、普遍性高、系统处理时间短等优点。In order to make better use of sidewalk for the blind,blind roads need to be divided from complex surrounding in pictures.In this paper,a blind road segmentation algorithm based on color texture and SVM is proposed,which firstly uses SVM to train the samples with different features and then distinguishes the input picture with the data trained by SVM to get the blind road.In comparison,we get the the color features of samples in HSV color space and the texture features of samples through the Gabor filter which has three frequencies and two direction angles and then put those features into SVM classifier to be trained.Experiment results shows that this algorithm is more stable,of higher universality and less process time compared with existing algorithms.

关 键 词：电子导盲设备 盲道分割 HSV GABOR滤波器 SVM 

分 类 号：TP312[自动化与计算机技术—计算机软件与理论]