基于颜色特征的多分支路径视觉导航AGV小车
首发时间:2018-02-09
摘要:针对AGV多分支路径引导问题,提出一种基于颜色特征的视觉导航方案。首先,将图像转换为HSI颜色空间。然后,用双边滤波器对图像的H、S分量进行平滑处理,能够在去除噪声的同时较好地保留边缘。研究不同色彩与H、S分量之间的关系后,使用BP神经网络模型,将具有不同颜色特征的多分支路径标识和路径从图像中提取出来。再根据获取的多分支路径标识和路径的状态判断当前道路状态,在不同的道路状态下,采取不同的策略选择目标路径。最后,用最小二乘法拟合目标路径,获取偏转角和偏离距离,根据偏转角和偏离距离对小车的方向进行模糊控制。实验结果表明,该方案拥有较精确的路径选择正确率,为99.27%;以及较低的道路偏离率,其中直线路径为2.13%,而圆弧路径为7.42%。
关键词: AGV 视觉引导 HSI 多分支路径双边滤波器 BP神经网络模糊控制
For information in English, please click here
Cross Path Visual Navigation Automated Guided Vehicle Based on Color Feature
Abstract:A visual navigation scheme based on color feature was proposed for vision-based automated guide vehicle cross path guidance. Firstly, the image is converted to HSI color space, Then, using the bilateral filter to smooth the H and S components of the image, the edge can be retained well while removing the noise. After researching the relationship between different colors and H, S components, the BP neural network model is using to extracted the multi-branch path identifiers and path with different color features from the image. And then, judge the current state of the road according to the obtained the multi-branch path identifier and path, Under different road conditions, a different strategy is taking to choose the target path. The least square method is used to fit the target path and obtain the deflection angle and the deviation distance, and then combine the fuzzy control principle to control the AGV walking direction. The experimental results show that the proposed scheme has accurate path selection accuracy of 99.27% and low road departure ratio of 2.13% on the straight line and 7.42% on the circular path. The average processing time per frame image is approximately 98 ms.
Keywords: AGV visual guidance HIS Cross path Bilateral filter BP neural network fuzzy control
引用
No.****
同行评议
共计0人参与
勘误表
基于颜色特征的多分支路径视觉导航AGV小车
评论
全部评论0/1000