The Method And Realization Of Lens Distortion Correction For High-Precision Image Size Measurement

With the rapid development of modern parts processing industry, traditional measure technology has been unable to meet the demand of higher and higher precision measurement. The traditional manual measurement cost a lot of labor, high cost of quality and management, and its precision is conditioned by factitious factors. So, it has been unable to meet the demand of current production. The rising computer vision gradually occupies the main position in the detection field. This thanks to its high-precision measurement and the advantages of the automatic detection. The requirement of high-precision measurement is both opportunity and challenge for computer vision. In the high-precision measurement, image is not only sensitive to the influence of noise, but also affected by geometric distortion which also decline the accuracy. Small distortion can also decline the accuracy greatly. Only the real image can reflect the reality of the object truly. How to be better adapt to the requirements of modern industrial, lens distortion correction of high precision image measurement has important significance.In order to achieve the high-precision image size measurement, this topic make the sizes measurement of automobile engine valve as the research object, and research the lens distortion correction method based on computer vision. Through this method, the topic can realize the improvement of measure accuracy and efficiency. In the measure system of this topic referred to, image of valve is extracted by high-speed CCD. After that, the image will be processed by the median filtering and its edge feature will be get by sub-pixel edge detection. The optimization of calibration methods will be improved by combining the principle of camera radial distortion, the relevant calibration method and the practical application. The exact results of size parameters can be get by distortion correction which is through the sub-pixel edge to achieve. Finally, the high precision results will be output. The main research contents are as follows:1. Design the size detection algorithm according to the actual valve sizes detection requirements. 2. Extract the edge detection of related area and improve image quality. Through filtering the image noise by median filter, the image quality can be improved. After that, the sub-pixel edge of location area can be extracted through prewitt modified algorithm combined with the qualitative method. So, the high precision edge characteristics can be get.3. Calibrate the image distortion and output the result. Based on the sub-pixel edge, this topic introduce the radial distortion parameters, and establish the mapping relation which connect world coordinates and image coordinates. As a result, the function of the error compensation curve can be achieve. By using the coefficient matrix and the radial distortion parameters, the error correction of lens distortion can be realized.Experiment result shows that this method can effectively improve the measure accuracy, and it also can reduce the cost of equipment. The accuracy can reach to2μm within. Now the method has been realized in computer program and applied to production.