基于SURF和最佳缝合线的车道图像序列拼接研究

Lane Image Sequence Stitching Based on SURF and Optimal Seam

为解决车辆压线行驶或发生交通事故时占用多个车道而无法采集到完整的车辆图像问题,基于SURF(speed up robust features)算法和最佳缝合线的思想,提出一种车道图像序列拼接方法。首先根据车辆运动轨迹提出寻找最优对应图像算法,用于找出近似同步拍摄图像;然后用SURF检测图像中的关键点对图像进行粗配准,用RANSAC方法去除误配准点实现精配准;最后利用最佳缝合线方法沿图像的重叠区域进行分割拼接。实验表明:利用车道图像序列拼接方法能找出近似同步拍摄图像,准确计算出图像的重叠区域,避免缝合线经过运动区域,解决了拼接过程中产生的重影、裂缝等问题,合成了质量较高的多车道图像,以此确保获得完整的车辆图像。

In view of the fact that the lane camera couldn’t capture the complete vehicle image when vehicles were running on line or occupying more than one lane in traffic accidents,a lane image sequence mosaic method was proposed based on SURF( speed up robust features) algorithm and the optimal seam,in order to solve the problem of incomplete vehicle information in the process of image acquisition.Firstly,an algorithm for finding the optimal corresponding image was proposed according to the trajectory of the vehicle,which was used to find the approximate synchronous image.Secondly,the SURF was used to detect the key points in the image and the image was coarsely registered,and then the RANSAC method was used to remove the misregistration points to achieve the accurate registration.Finally,the optimal seam was used to segment and stitch along the overlapping area of the image. The experiment shows that the proposed method can find the approximate synchronous captured image,accurately calculate the overlapping area of the image,avoid the suture passing through the motion area,and solve the problems of ghosting and splicing crack caused by the stitching,and a high quality multi-lane image is synthesized to ensure the complete vehicle image.