暗夜下基于激光辅助的路面障碍物视觉识别方法

为了解决现有视觉识别易受天气、时间等因素干扰的问题,通过交互矩阵和投影原理提出了一种暗夜下单线激光辅助的单目视觉识别方法。首先对相机进行标定,并利用交互矩阵建立并描述相机运动与图像特征变化之间的关系。然后利用投影原理进行一般障碍物特征提取和轮廓提取,最后通过试验对算法进行验证。利用激光光线的错位作为障碍物的识别标志,试验结果表明该方法克服了视觉识别对特殊环境适应性差、识别效果不好的问题,得到了比较准确的障碍物轮廓和距离,有效扩大了视觉识别的工作时间范围。

Visual Recognition of Road Obstacles in Dark Based on Laser-assisted

In order to solve the problem that the existing visual recognition is susceptible to weather, time and other factors, a single-line laser-assisted monocular vision recognition method at night is proposed by using interactive matrix and general obstacle feature extraction and contour extraction methods. Firstly, the camera was calibrated, and the relationship between the camera motion and the change of image features was established and described by using the interactive matrix. Then the general obstacle feature extraction and contour extraction were carried out. Finally, the algorithm is verified by experiments. The dislocation of laser light can be used as an obstacle recognition mark. The experimental results showed that this method overcomes the problems of poor adaptability to special environment and poor recognition effect of visual recognition., obtains more accurate obstacle contours and distances, and enlarges the working time range of visual recognition effectively.