动车组列车序列图像快速对齐方法

发布时间：2018-09-17 13:02

【摘要】：线阵相机具有高敏感性、高分辨率和高动态性等特点,常用于运动物体成像。在动车组列车运行故障图像动态检测中,由于列车运行速度不是理想匀速通过线阵相机,所拍摄的图像会在列车运行方向上出现拉伸或压缩等变形现象,造成同一列车在不同时间所拍摄到的图像总数不一致,给后续目标自动定位、识别以及故障自动检测带来挑战。为解决图像之间未对齐的问题,提出了一种基于分块式的图像对齐方法。首先对图像进行分块处理、特征提取和匹配、特征量化,然后根据特征点之间的像素距离实现图像分块式校准,最后级联各个校准之后的图像块,完成目标图像与标准图像的对齐。实验结果表明,该方法对线阵相机所拍摄的动车组列车序列图像具有较好的对齐效果,可以精确定位序列图像中的目标。
[Abstract]:Linear array cameras have the characteristics of high sensitivity, high resolution and high dynamic, so they are often used for imaging moving objects. In the dynamic detection of train fault images of EMU trains, because the train speed is not ideal and uniform, the images will be stretched or compressed in the train running direction. The number of images captured by the same train at different times is not consistent, which brings challenges to automatic location, recognition and fault detection of subsequent targets. In order to solve the problem of unaligned images, a method of image alignment based on block is proposed. Firstly, the image is divided into blocks, feature extraction and matching, feature quantization, then the image block calibration is realized according to the pixel distance between feature points. Finally, the image blocks after each calibration are concatenated. The target image is aligned with the standard image. The experimental results show that the proposed method has a good alignment effect on the train sequence images taken by a linear array camera and can accurately locate the targets in the sequence images.
【作者单位】：北京航空航天大学仪器科学与光电工程学院;
【基金】：国家重大科学仪器设备开发专项(2012YQ140032) 科学研究与研究生培养共建项目—成果转化与产业化项目—列车弓网运行状况在线动态检测系统
【分类号】：TP391.41

【参考文献】