Applied Research Of Machine Vision Technology In The Measurement Of Sheet Metal Forming

In recent years, along with the rapid development of computer technology,artificial intelligence technology, parallel processing, neural network and otherrelated technology, machine measurement technology has being improvedcontinuously. The application research of machine vision technology has become themain research direction of machine vision. During the process of sheet drawing,resilience is one of the most fundamental defects. At present, there is no perfect wayto control the springback of shallow drawing parts as motor vehicle coving parts,especially. Therefore, the key concern on forming numerical simulation and molddesign are how to measure the springback value and position where springback takeplace precisely and to establish sheet metal forming parameters precisely. In thispaper, we propose applied research of machine vision technology in the measurementof sheet metal forming. The main contents are as follows:On the bases of considering fully the projection width of projector and the fieldrange of camera, we choose the hardware equipment of measurement system, designthe experiment platform and construct the whole measurement system.Studies on technology of binocular stereo vision measurement method withsurface structured light. We complete the camera calibration by using the stereocalibration of Zhang’s, to obtain the internal and external parameters of the camera.We complete binocular stereo rectification by using the technology of stereorectification, and make all polar at the same level, which on the left and the rightcamera. we realize phase unfold of raster image by taking the technology whichbasing on the combining of Gray code and four steps phase-shifting of binocularstereo vision, to obtain the accurate absolute phase value. On the basis of imagestereo rectification, together with the polar constrained geometry, we achieve thequick match of corresponding match points of the left and the right images.Studies on technology of3D model reconstruction. We realize the coarsealignment of measurement patch by using of way of extracting identify points. Wesolve the accurate splicing of patch by using of ICP algorithm. Take the bilateral filtering technology to finish point-cloud data filtering, to solve the problems suchas forecasting and filtering the normal vector of point-cloud’ vertex, filtering theposition of point-cloud’ vertex. We realize to reconstruct the triangular mesh mold byusing point-cloud data.Studies on technology of springback measurement of coving parts. Matching andaligning the point-cloud data to the design mold of coving parts. Then, we calculateprojective point in theoretical model. In turn, we calculate normal deviation relativeto the theoretical model. The measurement results can be demonstrated in visual bymeans of color clouds. Finally, we can provide the distribution of springback on anycross section of the model.