一种彩色图像分割的障碍物识别方法  

A Method of Color Image Segmentation Used in Obstacle Recognition

作　　者：侯之旭 张建勋[1] 

机构地区：[1]重庆理工大学计算机科学与工程学院,重庆400054

出　　处：《重庆理工大学学报（自然科学）》2016年第3期94-98,111,共6页Journal of Chongqing University of Technology：Natural Science

基　　金：企业信息化与物联网测控技术四川省高校重点实验室项目(2015WZJ02)

摘　　要：移动机器人障碍物检测是机器人避障和移动路径规划的基础,为了提高视觉机器人障碍物检测能力,尤其是障碍物识别能力,针对传统的灰度图像分割作为障碍物目标识别方法在某些环境下效果不佳的问题,提出一种基于彩色图像分割的障碍物检测方法。首先充分利用HSI颜色模型中各个颜色分量的特征,再结合Otsu自适应阈值方法进行对应空间的图像阈值分割。考虑到机器人工作环境下外界光照条件变化较大的特点,提出了对亮度空间I进行非线性指数变换处理的做法,以解决图像分割和障碍物识别易受光照变化影响的问题。实验结果证明了该方法的可行性和有效性,可为障碍物检测提供清晰准确的障碍物识别结果。Mobile robot obstacle detection is the basis of robot obstacle avoidance and mobile path planning. In order to improve the detection ability of the obstacle detection in the visual robot,especially the obstacle recognition ability, aiming at the problem of the traditional gray image segmentation as an obstacle to the target recognition method in some environment,a method for the detection of obstacle detection based on color image segmentation was proposed. Firstly, the characteristics of each color component in the HSI color model were fully utilized,and the image threshold segmentation based on the Otsu adaptive threshold method was applied. In addition,considering the change of the environment of the robot working environment,the changes of theillumination conditions are relatively large, we proposed a method to deal with the nonlinear exponential transformation of the luminance space I in order to improve the image segmentation and obstacle recognition,and the problem is easily affected by illumination changes. Experimental results demonstrate the feasibility and effectiveness of the proposed method and it can provide clear and accurate obstacle recognition result for obstacle detection.

关 键 词：灰度图像分割 HSI颜色空间 OTSU算法 指数变换 开运算 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]