基于外轮廓与霍夫变换的印刷品图像配准算法

Print image registration algorithm based on outer contour and Hough transform

为解决传统印刷品图像配准算法精度和速度不能兼得的问题,本文提出了一种基于外轮廓与霍夫变换的印刷品图像配准算法。该算法先对图像进行灰度化、去噪、对比度增强、图像分割完成图像预处理;根据图像分割结果获得印刷区域的外轮廓,计算外轮廓形心坐标,得到偏移量,再根据轮廓计算出旋转量;根据偏移量和旋转量对待配准图像进行刚性变换和双线性插值完成图像粗配准;在感兴趣区域(Region of interest,ROI)中限制霍夫变换的动态参数,计算出ROI中的直线,根据直线交点得到3个坐标,根据3个坐标对待配准图像进行仿射变换和双线性插值,完成图像精配准。实验验证表明,本算法配准结果的平均NNC系数达到0.903264,平均配准时间为256.91。

In order to solve the problem that the accuracy and speed of traditional printed image registration algorithms cannot be achieved,this paper proposes a printed image registration algorithm based on outer contour and Hough transform.Perform grayscale,denoising,contrast enhancement,and image segmentation to complete image preprocessing.Obtain the outer contour of the printing area according to the image segmentation result,calculate the centroid coordinates of the outer contour,obtain the offset,and then calculate the rotation amount according to the contour,and perform rigid transformation and bilinear interpolation on the image to be registered according to the offset and rotation amount.Complete the rough registration of the image;limit the dynamic parameters of the Hough transform in the ROI area to calculate the straight line in the ROI,obtain three coordinates according to the intersection of the straight line,and perform affine transformation and bilinear interpolation on the image to be registered according to the three coordinates.Image fine-registration.The experimental verification shows that the average NNC coefficient of the registration result of this algorithm reaches 0.903264,and the average registration time is 256.91.