肇事车辆刹车痕迹图像特征匹配技术仿真

The Vehicle Braking Track Simulation Image Feature Matching Technology

肇事车辆刹车痕迹图像匹配的准确性,对车祸过程的复原至关重要。痕迹特征提取过程中,肇事车辆刹车痕迹受到车祸突发的影响,在距离和强度上都不同。随着车辆运动轨迹的不同,关键区域的刹车痕迹会被其它无关痕迹覆盖。造成关键区域的痕迹特征存在较大的相似性干扰。传统融合背景差法和边缘检测法的肇事车辆刹车痕迹匹配技术,无法对痕迹由于干扰形成的痕迹图像重叠区域进行准确的分割,不能实现肇事车辆刹车痕迹的有效匹配。提出了一种视觉优化的肇事车辆刹车痕迹匹配技术,通过成像光路的图像匹配方法,塑造关于双目视觉的摄像机模型,按照摄像机模型的成像光路获取摄像机视场交汇的几何模型,通过集合关系预测两幅肇事车辆刹车痕迹图像中存在的最大肇事车辆刹车痕迹图像重叠区域,采用特征匹配算法在该区域完成肇事车辆刹车痕迹的匹配分析。实验结果表明,所提方法能有效地提高肇事车辆刹车痕迹匹配精度。

The image matching accuracy of accident vehicle braking mark, is very important to the recovery of traffic accident process. The paper proposed an accident vehicle braking mark matching technology based on visual optimization. Through the image matching method of imaging optical path, a camera model for binocular vision was constructed. According to the optical path of camera model, a geometric model of the junction of the camera view can be obtained. Through the aggregation relationship, the maximum image overlap area of the accident vehicle braking marks existed in two accident vehicle braking mark images was predicted. Using feature matching algorithm, the matching analysis of the accident vehicle braking mark in the area was completed. The experimental results show that the proposed method can effectively improve the accuracy of accident vehicle braking mark matching.