基于单帧图像的物体面内位移和速度测量

发布时间：2018-06-27 12:43

本文选题：位移测量 + 速度测量　；参考：《光电子·激光》2016年11期

【摘要】：为了实现单帧图像检测物体两个运动状态的目的,提出了一种利用隔行扫描摄像机采集单帧图像测量物体在远动平面内的位移和速度的新方法。隔行扫描方式的摄像机采集在平面内运动物体的一帧图像,是由相隔一个场周期时间采集的奇场和偶场图像组成,每个单场图像经过傅里叶变换去隔行算法处理,获得与单场对应的帧图像质量相当的去隔行图像。根据物体与背景对光的反射率不同,提取物体的轮廓获得二值化模板,计算质心获得亚像素的定位精度,通过标定获得匹配点的空间坐标,进而计算在一个场周期时间内的位移和平均速度。对面内匀速运动物体的实验中,被测速度的最大绝度误差为0.35mm/s,相对误差为0.78%。实验结果证明本文方法的有效性。
[Abstract]:In order to detect two moving states of objects in a single frame, a new method of measuring the displacement and velocity of objects in remote plane by interlacing scanning camera is proposed. An interlaced camera gathers a frame image of a moving object in a plane, which consists of odd and even field images collected at intervals of one field period. Each single field image is processed by Fourier transform de-interlacing algorithm. A de-interlaced image with a frame image quality corresponding to a single field is obtained. According to the different reflectivity of object and background to light, the contour of the object is extracted to obtain binary template, the centroid of mass is calculated to obtain the positioning accuracy of sub-pixel, and the spatial coordinates of matching points are obtained by calibration. Then the displacement and average velocity in a field period are calculated. The maximum absolute error of measured velocity is 0.35 mm / s and the relative error is 0.78 mm / s. The experimental results show that the proposed method is effective.
【作者单位】：四川大学光电系;平顶山学院电气与机械工程学院;
【基金】：国家科技重大专项(2009ZX02204-008) 国家“863”计划(2007AA01Z333)资助项目
【分类号】：TP391.41

【相似文献】