The Research Of Casting Producting Line Aluminum Stack Positioning Technology Based On Binocular Vision

Demand for aluminum is second only to steel non-ferrous materials, an important position in the national economy. In recent years, with a yearly production of aluminum, high-speed aluminum ingot continuous casting production lines market demand is increasingly urgent. In ingot casting production lines, aluminum ingot stack requires packaging, marking and metering station stops, due to the weight of aluminum ingots per stack height(weighing about 1 ton), inertia is large and irregular,leading to chain drive production line in these stations ingot stack difficult to locate the current position based on the control signal detection ingot stack stop strategy is difficult to meet the needs of high-speed casting production, on the other hand, in some stations, targeting not only aluminum ingots stack measurement position information, also we need to measure the spatial position and orientation of aluminum ingot stack information. For these reasons, research ingot stack stationary state secondary positioning method has important application value.In this thesis ingot stack stationary state secondary targeting needs, proposed using binocular stereo vision ranging positioning method.Firstly, determine the binocular vision measurement program, using Zhang camera calibration method to calibrate the camera, in order to solve the internal camera parameters: According to the characteristics of the fixed camera coordinate axis point, the camera outside Senate alone for about camera calibration, respectively,using a linear equation solving the remaining parameters of the camera, the completion of the baseline length between two cameras, the angle between the optical axis and the relative position of the coordinate system calibration.Secondly, collected aluminum pile image feature extraction and matching algorithms for selected SIFT algorithm performance better than the SURF-based algorithm, based on highly reflective aluminum pile surface texture of complex features, the proposed package stack steel ingots as with the main basis for extracting feature points, on the basis of SURF feature extraction is proposed using KD-tree-tree indexes for feature extraction stack steel ingots on the little bit of the distance constraints, end up feature points. Harris corner feature point matching method based on matching, based on prior knowledge of proposed stereo matching algorithm based on Harris corner extraction and improved SURF algorithm combines features.Thirdly, to address the actual measurement distance measuring optical axis parallel to the problem may not be based on the angle between the translation conversion to convert between two cameras and two cameras optical axis coordinate system, ranging formula for parallel stereo vision model were correction. Finally, the experiments demonstrate the effectiveness of the proposed method, the ingot stack to meet positioning requirements.