一种抗阳光干扰的视觉机器人路径识别算法

A path recognition algorithm with anti-sunlight interference for vision robot

视觉机器人能够准确识别路径是其循迹导航的关键。针对全局动态阈值、双峰动态阈值路径识别算法无法应对阳光干扰的问题,提出了一种抗阳光干扰的路径识别算法。该算法基于图像的照射-反射模型,首先通过局部伽马矫正改善图像光照质量,再将其转换到对数域中获取图像梯度信息,以进一步削弱低频入射光量,达到抗阳光干扰的目的。然后合成4个检测方向的改进Sobel算子,并取无穷范数为最终梯度,以提高路径识别的准确性。最后通过局部动态窗口阈值法来匹配最佳阈值,以进一步提高抗阳光干扰能力。通过求取路径三角形的有向面积,以有向面积的正负、大小来判断路径走向与弯曲程度,并以此作为视觉机器人行进转向的控制依据。搭建机器人操作系统(Robot Operating System,ROS)平台进行实验,实验结果表明该算法能够有效改善视觉机器人的抗阳光干扰能力,提高循迹的可靠性与鲁棒性。

It is the key of tracking navigation that the vision robot can recognize the path accurately.In order to solve the problem that the global dynamic threshold and bimodal dynamic threshold path recognition algorithm can not cope with sunlight interference,an anti-sunlight interference path recognition algorithm was proposed.Based on the irradiation-reflection model of the image,the algorithm first improved the illumination quality of the image by local gamma correction,and then converted it to the logarithm domain to obtain image gradient information to further weaken the amount of low-frequency incident light to achieve the purpose of anti-sunlight interference.Then the improved Sobel operators in four directions were synthesized,and the infinite norm was taken as the final gradient to increase the accuracy of path recognition.Finally,the local dynamic window threshold method was used to match the optimal threshold,so as to further improve the anti-sunlight interference ability.By calculating the directed area of the path triangle,the path direction and bending degree were judged by the positive or negative and size of the directed area,which was used as the control basis for vision robot steering.The Robot Operating System(ROS)platform was built for experiments,and the experimental results show that the algorithm can effectively improve the anti-sunlight interference ability of the vision robot and improve the reliability and robustness of tracking.