一种基于图像处理的提示盲道检测方法-《计算机技术与发展》

文章信息/Info

Author(s):: ZHAO Lei; LI Zhen-wei; YANG Xiao-li; YANG Bin; JIA Meng-li; School of Medical Technology and Engineering,Henan University of Science and Technology,Luoyang 471000,China

Keywords:: blind sidewalk detection; warning blind sidewalk; Lab space; image segmentation; blind sidewalk boundary model

摘要:: 传统盲道边界模型对提示盲道的预警作用不足,因此提出一种基于颜色信息的提示盲道检测算法。将摄像头采集到的盲道图像从 RGB 颜色空间转换为 Lab 颜色空间,对具有不同干扰颜色的盲道,结合盲道分割常用的颜色空间以及提示盲道的特征,分别提取 a 分量和 b 分量, 综合利用 Otsu 阈值分割、形态学处理分割盲道区域, 得到分割后的二值图像。通过分析二值图像中盲道区域像素点的数量与像素点中心位置的关系,实现提示盲道的检测和分类。实验结果表明,该算法能有效分割出大多数盲道图像,并实现提示盲道的准确检测和分类, 将其应用在导盲设备中, 利用声音或者物理振动的反馈, 提醒盲人提示盲道的存在,帮助盲人寻找更好的行进方向,不仅能够极大地保证出行安全,而且还可以提供更加全面的盲道信息。

Abstract:: Traditional blind sidewalk boundary model has bad warning performance for warning blind sidewalk,so we propose a blind sidewalk detection algorithm based on color information. The image acquired by camera is transformed from RGB to Lab color space.For blind sidewalk with different interference colors, combining the common color space of blind sidewalk segmentation and the charac-teristics of warning blind sidewalk,a component and b component are extracted respectively. The binary image is obtained by using the method of Otsu threshold segmentation and morphological processing. By analyzing the relationship between the number and the central position of pixels in the blind region,warning blind sidewalk is detected. The experiment shows that the proposed algo-rithm can effectively segment most blind sidewalk images and realize the accurate detection and classification of the blind sidewalk. It can be applied to the guide blind equipment,and the feedback of sound or physical vibration can be used to remind the blind people of the existence of blind sidewalk,so as to looking for a better direction of travel. The proposed algorithm can not only greatly ensure travel safety,but also provide more comprehensive blind sidewalk information.