论文部分内容阅读
提出了一种用于倒车辅助系统的基于改进Hough变换的车位线识别方法。算法首先采用局部像素分布特征以剔除非直线的干扰像素,然后使用像素直方图自适应地选择与车位线边缘直线相适宜的梯度方向区间,减少参与Hough变换的像素点数的同时减少背景干扰直线边缘的影响,最后通过判断相邻边缘间的像素灰度关系,进一步确认车位线边缘,从而识别出图像中的车位线。相比于传统hough变换和基于梯度方向区间的改进随机Hough变换(Grad-RHT),本文算法在保持与Grad-RHT运行效率相当的情况下,在各种干扰背景下都取得了更好的车位线边缘直线检测正确率,表现出较好的识别鲁棒性。
A parking line recognition method based on improved Hough transform is proposed for backup assist system. The algorithm first uses the local pixel distribution feature to eliminate the non-linear interference pixels, then uses the pixel histogram to adaptively select the gradient direction interval which is suitable for the line edge of the parking space line, reduces the number of pixels involved in Hough transform and reduces the background interference linear edge Finally, by judging the pixel gray relation between adjacent edges, we can further confirm the edge of the parking line and identify the parking line in the image. Compared with the traditional hough transform and Grad-RHT based on the gradient direction interval, the proposed algorithm achieves a better parking position under various interference conditions while maintaining the same operating efficiency as the Grad-RHT Line edge line detection accuracy, showing good recognition robustness.