The Research And Development Of Workpiece Detection In Robort Sorting System

With the advent of the information age, machion vision recognition has attracted a lot of attention. Workpiece detection, one of the focuses in the machion vision recognition field, is the hotspot and difficulty to researchers at home and abroad for a long time. Recently, an increasing number of automatic workpiece detection systems have been developed and applied in online detection process, to achieve high-speed and accurate detection, which greatly improve the detection speed and reliability of product, and reduce the cost of production. The key technique of workpiece detection is how to get the feature parameters, identify the type and judge the position and posture of the workpiece from the image information. Combined with the current development of digital image processing and pattern recognition techniques, this thesis researches several key algorithms of the workpiece detection, mainly focused on the optimization of the workpiece image preprocessing, workpiece modeling and workpiece identification. The main contributions of the dissertation are as following:1. In order to collect, transfer and display the workpiece image more conveniently, a hardware system of vision machine, based on the CMOS camera, video capture card and computer, is developed.2. Due to different kinds of noises, the image preprocessing is very important to improve the quality of the collected images. On the basis of the analysis of previous workpiece detection technology dissertations, this thesis realizes the key steps of the workpiece image preprocessing. In this aspect, a series of algorithm methods about image denoising, image binarization, rotation correction and size normalization is proposed. By using these methods, the noise is well removed, the contour of the workpiece is retained, and the binary image is clearly obtained. Besides, the rotation angle of the workpiece image is obtained by the minimum bounding rectangle, and the rotation correction is done by the affine transformation. Finally, the size of the workpiece is normalized, which ensures the accuracy and reliability of the feature extraction and template matching.3. The template of the workpiece is established when the image of a new workpiece is collected. First, the modeling image is processed by the image preprocessing method following the principle that the horizontal centroid of the rotated image should be on the left of the image. Second, the features are extracted during the template establishment, which are used to the second recognition. Drawing tool is used to extract the features and the extracted information is stored in the document library.4. In order to optimize the workpiece detection algorithms based on the template matching, a two-level workpiece matching algorithm is developed. First, the point matching recognition is used to compare the collected image and all the template images in the library one by one, during which the gray value of the pixels is compared and the first matching coefficient is recorded. Then, the feature matching recognition is used to three templates with high first matching coefficient, and the second matching coefficient is recorded. The final matching coefficient is added by the first matching coefficient and the second matching coefficient. The highest matching coefficient will be outputted as the result along with the workpiece type, the centroid position and the rotation angle. The experiemental result shows that, this method greatly improves the speed and accuracy of workpiece detection system.5. In order to verify the feasibility and practicability of the automatic workpiece detection algorithms, a software platform is developed by using Visual Studio 2012 and OpenCV. The software has been integrated with a variety of key algorithms and three modules, which can be used to test and research on workpiece detection algorithm.The functional test, fault tolerance test and workpiece detection test are designed to verify the specifications of the system. The results show that, the system is stable and reliable, also meet the predetermined requirements.