Research On Visual Positioning Technology Of Slender String Based On 3D Point Cloud

The automation level of China's oil drilling platform is relatively low,especially the string transfer operation.The traditional technology is difficult to provide an effective automation solution for it,and the manual operation mode is still used at present.With the development of vision technology,the positioning technology based on machine vision can realize the non-contact and non-destructive real-time detection and positioning of specific targets,which provides a new solution for the automation of string transfer of drilling platform.According to the operation of the existing drilling platform,this paper designs an automatic positioning system based on vision to realize the independent detection of the string transfer link.According to the working environment of the drilling platform and the characteristics of the string to be tested,the imaging scheme based on active stereo vision is selected.The 3D point cloud of the working scene is obtained by the laser structured light industrial 3D camera installed in the non working area around the platform,and the target detection and positioning of the string are completed by the 3D point cloud processing algorithm.Due to the limitations of existing point cloud clustering target detection methods,such as complex background removal,difficult parameter setting,poor robustness and efficiency,this paper proposes an improved density clustering segmentation method based on the principal curvature feature score by using the geometric characteristics of the string target,which achieves stable detection and high integrity of the extracted target point cloud.Based on the segmentation results,the two-step projection method is used to complete the pose estimation of the string object.The direction adaptive axis direction detection method based on Hough transform and the fast circle detection method based on region subdivision are used.Compared with the existing methods,the stability and accuracy are improved.Finally,the stability and relative positioning accuracy of the proposed method are verified by the positioning experiment of outdoor string.The visual positioning scheme and corresponding visual positioning method proposed in this paper can complete the stable positioning of the slender string object,and can meet most of the detection requirements of the string transfer link of the drilling platform.And it has certain applicability to the scene of autonomous string transportation.