基于SIFT的路面裂缝配准及拼接算法

An Algorithm of Pavement Crack Image Registration and Mosaic Based on SIFT Algorithm

针对贯穿双车道的长裂缝,路面裂缝采集过程中往往不能得到完整的裂缝信息,可能造成对路面损毁程度的错误评估。为解决路面图像检测过程中采集裂缝信息不完整的问题,拟采用SIFT算法对在不同时刻对同一路面采集到的同一裂缝信息进行特征点提取。针对裂缝边缘处的特征点灰度较背景像素灰度有阶跃变化的特点,在裂缝图像特征点筛选过程中引入阈值分割理论,使裂缝信息从背景中分离,并且标记裂缝像素点,从而将图像中背景的特征点过滤。再将目标图像和参考图像中的特征点进行配准,并采用加权平均融合法将目标图像和参考图像拼接成一幅完整的裂缝图像。改良后的算法可以提高裂缝特征匹配的精度、速度、可靠性和准确性,以此保证对路面损毁状况的正确估计。

The information about the longer pavement completely, which may lead to the error evaluation of problem about the incomplete information of pavement cracks through two-lane often cannot be collected the pavement damage degree. In order to solve the cracks in detection process, the SIFT algorithm was used for the feature point extracting of the same crack on the same road detected at different times. For the gray-scale of feature points at the edge crack have the characteristics of a step change over gray-scale of background pixels, the threshold segmentation theory was introduced in the process of feature points selection in the crack images. This method can separate the crack pixels from the background and mark the pixels, and filter the feature points of the background in the image. Then register the feature points in the target image and the reference image, and join the target image and the reference image together into a complete picture of crack image by using the weighted average fusion method. The modified algorithm can improve the precision,rapidity, reliability and accuracy of the image matching, and ensure accurate assessment of the road damaged situation.