一种矿山地面应急救援井内表面视觉检测方案

Research on Visual Inspection Technology of Inner Surface of Mine Ground Emergency Rescue Well

为了实现对矿井的可视化救援,选用了深度相机作为测量设备,提出了一种改进地面应急救援井内表面的测量方案。针对实验黑暗环境,对SURF算法进行维度上的优化,使得特征提取点明显增多,通过FLANN算法对救援井井壁相邻图像进行特征粗匹配,而后基于RANSAC算法计算H矩阵并剔除图像误匹配对,提纯特征匹配点。结果表明,通过改进的SURF算法进行救援井井壁图像拼接,能获得形变处井壁全景图,从而更加清晰直观观测形变。

In order to realize the visual rescue of the mine,a depth camera is selected as the measuring equipment,and an improved measurement scheme for the inner surface of the ground emergency rescue well is proposed.Aiming at the dark environment of the experiment,the SURF algorithm was optimized in dimensionality,so that the feature extraction points were significantly increased.The FLANN algorithm was used to perform rough feature matching on the adjacent images of the rescue well wall,and then the H matrix was calculated based on the RANSAC algorithm and the mismatched images were eliminated,refine the feature matching points.By using the improved SURF algorithm to splicing images of the wall of the rescue well,a panoramic view of the wall of the deformation is obtained,and the deformation can be observed more clearly and intuitively.