Research Of Workpiece Positioning Based On Template Matching

The machine vision technology is introduced in the industrial control system become one of the research topic of the current relatively representative. Because the machine vision system has the quickness, accuracy, objectivity, in the modern automation production, gradually become the research focus in the people, based on the application of machine vision is also growing rapidly.Machine vision as an important means of access to environmental information instead of human eyes, and can improve the flexibility and autonomous ability of industrial robots for the workpiece. From the collected images to extract the characteristic information of the target artifact real-time and accurately identify the target artifact space position information machine vision system is one of the key technology in the field of automation application. In this paper, based on the edge of the geometric features of template matching technique for the workpiece positioning is studied. The main research content is as follows:1. the image distortion correction, and the realization of coordinate transformationIn this paper, the imaging model of camera and zhang zhengyou calibration method was studied, zhang zhengyou calibration method is introduced in three dimensional space target projection to the theoretical derivation of the imaging plane, and the role of internal parameters and distortion vector in the model; Also zhang zhengyou plane calibration method was deduced in detail in the image coordinate and world coordinate transformation formula. This article also zhang zhengyou calibration algorithm for programming, achieved through the experiment of image distortion correction and the coordinate transformation matrix is calculated.2. The design and implementation of the template matching algorithm based on the edge of the geometrical characterTemplate matching algorithm based on feature points gradient information was in detail in this paper. This article not only was deduced using gradient equation of curved surface fitting method, and the calculation formula of pixel coordinates is deduced, the most important thing is also derived by the least squares adjustment to realize high precision matching formula. Not only that, the article also puts forward the target search area local maximum values of extraction method. Even search images appear in the multiple target sample, the algorithm can identify and locate it all out. Due to the complexity of algorithm theory, this paper also introduces the algorithm of detailed implementation details, such as a template to create, matching process, etc., and draw program flow chart and experiment effect of each step is shown through the experiment and data.3. The algorithm of the actual performance testVisual system hardware environment, software environment and the visual system of the overall program flow chart are introduced in this paper. article respectively to the condition of single objective robustness, real-time and accuracy were tested. This article also algorithm for multi-objective situation of real-time and accuracy were tested. Through testing, we know that the actual performance of the algorithm, at the same time we also analyzed the result of the test.Finally, this paper summarizes the main research results, and points out the deficiency and put forward the improvement direction of this article.