一种结合Gabor滤波器与GLCM的盲道检测方法

A Blind Detection Method Combining Gabor Filter and GLCM

基于颜色的盲道检测模型对城市中存在的与背景对比度低的盲道分割效果不太理想,针对这个问题,从盲道上特有的纹理着手,设计了一种基于盲道纹理特征的盲道检测算法。该算法通过经典的纹理特征提取方法Gabor滤波器组与灰度共生矩阵相结合对摄像头采集到的盲道图像进行处理,实现盲道本身存在的纹理特征的提取,然后经过模糊C均值聚类算法实现图像中盲道区域的分割。在城市中背景与盲道区域颜色相近环境中,与一些经典的基于颜色分割的盲道检测算法进行比较,提出的基于纹理的盲道检测算法能够有效实现对盲道区域的分割,分割结果较为准确,可以在经过改善后将其应用在视觉辅助导盲设备中,帮助盲人方便、快速地寻找更好的行进方向,极大地保证出行安全。

The color-based blind lane detection model is less effective in segmenting the blind lanes with low contrast with the background that exist in the city. To solve this problem, a blind lane detection algorithm based on blind lane texture features is designed based on the unique texture of the blind lane. The algorithm achieves the extraction of the texture features present in the blind lane itself by using the classical texture feature extraction method Gabor filter bank combined with the gray scale co-generation matrix to process the blind lane images captured by the camera, and then realizes the segmentation of the blind lane region in the image by the fuzzy C-mean clustering algorithm. The proposed texture-based blind lane detection algorithm can achieve more effective segmentation of blind lane areas in urban environments with similar background and blind lane area colors than some classical color-based blind lane detection algorithms, and the segmentation results are more accurate, which can be used in visual aids to help blind people find a better direction of travel easily and quickly. It can be used in visual aids to help blind people to find a better direction of travel and ensure the safety of travel.