基于Harris多摄像头孔壁图像的全景拼接技术

Panoramic Image Mosaic Technology of Multi-camera Borehole Image Based on Harris Algorithm

提出一种适用于多摄像头孔壁图像畸变校正和全景图像拼接的算法。为获取360°孔壁图像信息,采用四个广角镜头成像设备采集图像,针对孔壁图像的特殊性,提出一种改进的畸变图像校正算法。首先通过平面网格模板法进行摄像机标定,计算畸变参数,为消除孔壁曲率的影响,提出基于参数微调的畸变图像校正算法。并且通过双线性插值算法进行灰度重建。其次为消除孔壁曲率造成的边缘信息的缺失,提出柱面信息还原法将图像展开,还原孔壁的真实像素信息。通过改进的Harris角点检测算法提取图像角点,利用RANSAC算法进行匹配对提纯,计算变换矩阵,采用融合速度较快的渐入渐出融合算法,得到孔壁全景图。

This paper presents an algorithm applicable to distortion correction and panoramic image mosaic of multicamera borehole image.To get information of 360°borehole image,four wide-angle cameras were used to collect images and a new improved distortion image correction algorithm was proposed in view of the particularity of borehole images.Firstly,distortion parameters were calculated by calibrating the camera through the grid template method.A distortion correction algorithm based on parameter adjustment was put forward to eliminate the effect of borehole curvature.The bilinear interpolation method was adopted to reconstruct the undistorted image gray level.In order to eliminate the deficiency of the edge information due to the borehole wall curvature,the cylindrical information reduction method was proposed to expand the image and restore the real pixel information of the borehole wall.The corners in the image were extracted through the improved Harris corner detection algorithm,and RANSAC algo-rithm was adopted to reduce the number of error-matching,and to compute the transformation matrix.Finally,the final borehole panorama image was obtained by using the gradual image fusion algorithm.