基于线阵CCD相机的汽车底盘图像配准算法

Research on image matching of vehicle’s chassis based on linear CCD camera

在基于线阵CCD相机的汽车底盘图像对比和安全检测中,特征点匹配算法是项目研究重点,但是大多数现存的方法不能满足项目对匹配数量、质量和分布的要求.基于SURF(speeded-up robust features)算法,并引入显著性约束和形变约束过滤掉错误匹配,最终获得数量与质量显著提高的匹配集合.结果实验结果表明,该方法获得的纵向分布完成系数大于80%、误匹配率低于2%,基本到达了安全检测的要求.

The algorithm of feature matching is the focus of research of image contrast and safety detection for the automobile chassis based on linear CCD camera.but most available methods fail to achieve satisfying quantitative,qualitative.Distributed matches simultaneously.This paper introduced saliency constraints and deformation constraints to filter out mismatches based on SURF(speeded-up robust features)algorithm,and finally obtained matching sets with significantly improved quantity and quality.The experimental results showed that the vertical distribution completion coefficient obtained by this method was more than 80%,and the mismatch rate was less than 2%,which basically met the requirements of security detection.