Research On Port Machinery Measurement Method Based On Close-range Photogrammetry

The development of world trade is inseparable from the port,and the normal operation of the port is inseparable from the port crane,and the port crane may have safety problems due to continuous heavy work for a long time.Therefore,the dimensions related to its safety assessment are periodically measured.It’s the key to ensure its normal operation.The port crane has a large working intensity and a large size.The traditional measuring method is complicated,difficult and time-consuming.Therefore,a combination of photogrammetry and computer vision is selected,and a single non-measuring camera is used to collect the image of the object to be measured and process it to obtain the object-side space coordinates of the object to be measured.The method is fast in field operation,does not require direct contact with the object to be measured,has low cost,and has high measurement accuracy.The existing photogrammetry methods are not well adapted to the environment in which port cranes are located.This article takes port cranes as an object,improves key links in the measurement process,and introduces computer vision and other technologies to propose a measurement method for port cranes.The specific research content is as follows:(1)A camera calibration method suitable for port machinery is studied.The traditional calibration method requires a calibration field that is substantially the same size as the actual measurement site.Because the port environment is too complicated,it is costly and difficult to establish an actual calibration environment suitable for port cranes.Aiming at this problem,a real-time calibration method suitable for the port environment is proposed.On the basis of Zhang Zhengyou’s calibration method,the camera’s focal length is calibrated on-site in real time.(2)The problem of corner extraction in image processing is studied.The photographs include not only the object to be measured,but also complex backgrounds and other objects.Therefore,in response to the need to extract the object to be measured from the captured image,Faster R-CNN in computer vision technology is introduced into the photogrammetry process,important parameters of the deep learning neural network are adjusted,the training of the neural network is completed,and fusion of image detection and corner detection completes the extraction of key corners on the object to be measured.(3)A photogrammetric matching point matching algorithm suitable for port machinery was studied.There are two traditional methods of matching points of the same name.The matching of points of the same name based on image information and the matching of points of the same name based on geometric constraints have higher error rates in the former and too much calculation in the latter.To address the problem of matching points of the same name for port cranes,an improved method of matching points of the same name based on geometric constraints is proposed.First,the improved Huff transform is used to extract the straight lines in the image,and then the position relationship between the straight lines is used to extract the intersection points between the straight lines.Finally,use the epipolar relationship between the images to search for points with the same name,and complete the matching of points with the same name.(4)The flatness and symmetry related to the safety assessment of port machinery are studied.The concepts of flatness and symmetry are introduced,and mathematical models of flatness and symmetry are established.Find the mark points on the spot,and then use the total station to collect the coordinate information of the mark points to complete the calculation of flatness and symmetry.And compare its measurement results with the total station measurement results.