Large Size And High Temperature Forging Binocular Stereo Vision Measurement Method

As the key parts of important equipments, large forgings play an important role in the process of industry revitalization. With the development of electric power, shipbuilding industry, chemical industry and metallurgical industry etc, the demand of large forgings is increased. Real-time sizes measurement and feedback of large forgings has great theoretical and economic significance for improving efficiency of forgings manufacturing and material utilization. The paper combines the urgent need of high-quality large forgings by state's major industries and research status of large forgings measurement technique. With the support of National Key Basic Research and Development Scheme-973, a comprehensive and exploratory research on large forgings dimension measurement method is developed so that it can provide support for improving the performance of forging manipulator.Large forgings are big and their manufacturing process associated with high temperature, high pressure and strong vibration. The measurement method combines binocular stereo vision and spectrum choice is put forward based on analysis of domestic and international researches on large size measurement present. The problems of forging image acquisition in high temperature condition are solved. The parameters of camera are calibrated, then the point clouds are reconstructed and forging sizes are estimated. The paper's study has several points as follows:The energy radiation characteristic of the forging in high temperature is studied and then a scheme is established. The scheme is that a band-pass filter is used for filtering strong light and an auxiliary light source is used for increasing energy of useful light wave band. As a result, forging image as good as it in house temperature is available in forging temperature.The camera parameters are calibrated by the plane board method like Zhang's which is flexible and high precision. Then, accuracy of parameters is proved by three-dimensional coordinate reconstruction.A pixel in color image is a vector contains three parts R, G, and B. Traditional gradient operators are always insufficient in the process of color image edge detection. Making full use of information of the RGB color space, edges of the three images are detected with the method of combining the Iterative Median Filtering and the Sobel Gradient Operator while keeping the edge information integrated. Next, the three images are integrated and the isolated points are removed. Finally, a non-maximum inhibition is implemented in order to simplify the process of feature extraction.The feature point-based gradual stereo match method is put forward. The feature points are detected by Harris operator which is high efficiency and good location. Our approach is to use gray correlation to find an initial set of matches, then use relaxation, robust recover polar geometry to discard false matches in this set and find more matches.Finally, the three-dimensional reconstruction mathematical model of point clouds is established based on the relation of image points and space points. Point cloud on surface of part is reconstruction. The experiments give a believable measurement result. It is a foundation of high-precision measurement. Further more, it has the guiding significance for further research on stereo vision measurement system of Large Forgings.