Zernike矩边缘检测与多项式拟合的CT图像三维测量算法

针对工业CT切片z方向分辨率不足问题,研究一种Zernike矩边缘检测与多项式拟合结合、自动预测顶端的高精度三维测量算法。首先通过Zernike矩对每张切片图像提取亚像素级轮廓,并用多项式拟合该轮廓;然后对拟合曲线进行等相角间隔采样,并建立同相角的轮廓点匹配,采用层间插值算法得到中间层的轮廓点并外插预估工件顶端点;最后采用台体法计算工件体积。对体积相等的圆柱、圆锥和半球3个标准试块进行CT扫描、测量并分析了测量误差的来源。实验结果表明,对实际工件体积测量误差降低到1%以下,能够满足工程应用中的自动、快速、高精度体积测量要求。

3D measurement algorithm of CT images based on Zernike moment edge detection and polynomial fitting

Aiming at the problem of low resolution of industrial CT slices in Z-Direction,a high-accuracy 3D measuring algorithm combining Zernike moment and polynomial fitting is presented,which can automatically calculate the apex.Firstly,sub-pixel-level contour of each slice is detected with Zernike moment,and the contour is fitted using polynomial fitting method.Then,the contour is sampled at the same angle interval,and the edge points at the same phase angle of each slice are matched;and then the apex of the work-piece is calculated using extrapolation method.Lastly,the volume of the work-piece is calculated using frustum method.Scanning experiments and volumetric measurement experiments were carried out for three equal-volume standard work pieces: cylinder,cone and hemisphere,and the measurement errors were analyzed.The result indicates that the measurement error of actual work piece using the proposed method is reduced to below 1%,which proves that the proposed algorithm can meet the requirements of automation,high speed and high accuracy in engineering applications.