Research On Workpiece Ranging Based On Binocular Stereo Vision

With the rapid development of science technology and market economy in our country, a large number of die-casting production enterprise emerge. In the face of competition in the industry, it is a tendency to use industrial robots in intelligent production and processing across entire production line. In the process of die-casting work piece production, as the work piece temperature would be up to 200 degrees in the pipeline. It is impossible for workers sampling the work piece directly. Traditional method to solve this problem is set a fixed position across manipulator and then grabbing the target work piece. For the reasons such as the shape of the workpiece, workpiece position uncertainty. It often failed to grab the target work piece while using traditional robot Based on binocular stereo vision model, the article put forward a new method to range artifact distance whose position is unknown. Aim at improving the precision that industrial robot locate target work piece, finally the efficiency of scraping was improved. The main content of this article as follows:(1) Design of binocular stereo vision system. In order to satisfy the needs of project, the author builds the hardware platform and designs the process of software based on binocular stereo vision system platform.(2) Camera calibration and image correction. After compare advantages and disadvantages of various domestic camera calibration methods. Combined with the objective of the hardware platform, the author chooses the Zhang Z.Y Calibration Method to calibrate the camera. Comparing the advantages and disadvantages of the Hartley algorithm and Bouguet algorithm, the author chooses the Bouguet algorithm to realize the image correction.(3) Image preprocessing and ROI area target extraction. In order to reduce the influence from the external environment such as noise, light intensity, the author using some image preprocessing methods including image contrast enhancement, image noise processing etc. As stereo matching algorithm could only match an integrated image, it could not extract matching pair of target work piece. To solve the problem, the author put forward a new method to self-select ROI area, which could realize single matching on target work piece. The simulation experiment shows that the proposed method could segment target artifact quickly and reduce the subsequent stereo matching work area, finally the method improves the matching speed and matching precision.(4) Four stereo matching algorithms and artifacts ranging. Based on self- select ROI area, using stereo matching algorithms(Harris、SIFT、SURF、FREAK) to extract image feature points and stereo matching respectively. The author select the most suitable matching algorithm for project by comparing four matching algorithms in the side of matching time, the number of matching successful and matching stability. Experiment target distance from the camera and Experiment distance between two target workpieces compare with actual target measuring distance. Finally the experiment obtained the work ranging error in the range of 2mm.The simulation result tells.that the proposed method has a better result than traditional target machine workpiece positioning methods in terms of accuracy and speed, basically satisfied the requirement of the project.