Research On Parallel Calibration Technology For CCD In Dimension Measurement

CCD camera calibration technique is one of the most important aspects in image measurement. With the measurement precision improvement of image measurement, the CCD camera calibration technology also has set higher requirements.The task of CCD camera calibration is to identify the geometric parameters and optical parameters of the camera model, as well as the camera position in the world coordinate system. The accuracy of image measurement is directly determined by the calibration accuracy. At present, there have been a series of methods of camera calibration, so that how to utilize an appropriate, straightforward and exact camera calibration method for the idiographic application is an actual pivot for the pursuers. In this paper, the parallel calibration has been studied based on some of existing methods for the dimension measurement of machine parts. There have been some improvements for the camera model according as the specifically problem, which lead to a increase of calibration accuracy and robust.This paper presents a method of dimension measurement in the world coordinate system based on getting the orientations of the edge points from a single image of the machine part. There are four steps for this method. First of all, it must use the parallel camera calibration method presented in this paper; the second is denoising and detecting the edges of the parts; the third is calculating the word coordinates of the edge points. Finally the last step is fitting a least square straight line for one of the edges, and then calculating the dimensions from the points on the other edge to the fitting line.This article consists of six parts shown as follow:Ⅰ. IntroductionThis part has introduced the development state of the camera calibration technology all over the world, and then presented the parallel camera calibration method of this paper.Ⅱ. The imaging model of camera First of all, this part has explained the four reference frame in the camera model, and then introduced the linear camera model according to the principle of pinhole imaging. Secondly it has introduced the lens distortions including radial distortion, decentering distortion and thin prism distortion. Finally it has introduced the nonlinear camera model considering the lens distortions. The intention of this part is explaining the physical meanings of the parameters in the camera calibration process.Ⅲ. The parallel camera calibration method based on radial alignment constraintAt first, this part has introduced the meaning of radial alignment constraint. Secondly it has researched the sub-pixel corner detection technology for extract the feature points, and then described the calibration process in detail. At the end of this part, it has validated the feasibility of this method by experiment as well as the precision evaluation of this method.Ⅳ.The improved parallel camera calibration method based on radial alignment constraintAt first, this part has reasonably improved the rotation matrix expressed by the sine and cosine of the euler angles according to the specific constraint of the parallel camera calibration method. Secondly, it has introduced the camera model of this improved calibration method. The third is describing the calibrating process of the improved calibration method in detail. At the end of this part, it has verify better precision of the improved calibration method than it of the non-improved method by experiment, and it has studied the relationship between the precision of the calibration parameters and the variance of gaussian noise in the image.Ⅴ.The application on dimension measurement of parallel camera calibration methodAt the beginning of this part, it has introduced the principle of the dimension measurement method presented in this paper. Secondly, it has described the process of this measurement method in detail including median filtering, sub-pixel edge detect technique base on gaussian curve, calculation of the word coordinates, and the principle of linear least squares fitting. At last it has compared the measurement accuracy between the improved camera model and the non-improved camera model by experiments, and it indicates that the former is better. Furthermore, it has analyzed the factors of the measurement error.Ⅵ. ConclusionThis part has given the conclusion of this article and analyzed the shortcomings of the paper. Furthermore, it has noted the entry points for the next in-depth study.