Research On Edge Detection Technology Of Machine Vision

As the most active research field, Machine vision has quite extensive research content and application domain. Edge detection technology is not only the most basic technology and the difficulty of image processing and machine vision, but also one of the basic image processing steps. Through edge detecting we can keep the structure information of the object boundary shapes, which will significantly reduce the data quantity of image processing, so as to simplify the image's analysis process. How to quickly and accurately extract image edges information has been the hotspot research at home and abroad. The camera calibration is to determine the camera's parameter model through its single or many images, and then get all parameters. It has wide application prospect. In order to obtain the corresponding relation from spatial point to the camera image pixels, camera calibration technology is indispensable.This paper uses 0.5 level milliammeters as the research object and developing the automatic calibration system of the pointer instrument as application background. It has a deep research and analysis on the edge detection technology and camera calibration technology. The main research results and innovation points are summarized as follows:1. The research of classic edge detection technologiesThis paper summarizes various classic edge detection operators. With VC++6.0 software platform, it respectively realizes application of multiple classic edge detection operators, simulates in two conditions without noise image and with gaussian noise image processing results simulation, compares and analyses the advantages and disadvantages of various edge detection operators. This kind of traditional methods is mostly based on differential techniques to determine the image edges, which is simple, but the antinoise performance is poorer, so they are just suitable for clearer edge images with larger signal-to-noise ratio.2. Proposed an edge detection algorithm based on mathematical morphologyThrough the designed edge detection algorithm, this paper processes clear images and unsharp edge images and simulates with MATLAB tools. Compared to the results of Sobe algorithm experiment, we can see that this algorithm based on mathematical morphology is superior to the traditional algorithm in edge positioning accuracy, the edge continuity and anti-noise performance aspects.3. A kind of binocular vision calibration system constructed with USB camerasThis paper analyses the camera's imaging model and calibration principle. After conducting several experiments about two cameras'position according to the characteristics of the pointer instrument and the requirements of the system itself, it builds the right instrument binocular vision hardware system. And on this basis, it improves the calibration method of the instrument binocular vision system, corrects the read values combining distance read value method. Finally, it designes some experiments to calibrate USB binocular vision system, and then realizes the machine vision calibration with VC++6.0 programming. The results have the certain errors and we analyses the errors. The proposed calibration method has good realizability and portability.