Study On Techniques Of The Single Camera Probe Imaging 3D Measurement And The Feature Points' Angular-Displacement-Error Compensation

Visual spatial coordinate measuring technique is a new coordinate measuring technology, and its essence is to use specific algorithms, which are based on optical imaging principle and imaging analysis, to get the spatial coordinates of the measured points. This paper researched the probe imaging 3D coordinate measurement technology which is on basis of one single camera and more than 4 feature points. Compared with the traditional measurement technologies (such as CMM), the new technology was adequate to meet the demand of large-scale on-line 3D measurement of large-scale manufacturing industries due to the advantages of light weight, flexible layout, and large range. At present, some of foreign companies have researched and sold mature high-precision 3D coordinate measuring equipments; however China still lacks the products with independent intellectual property rights. Although some progress has been made, but measuring range and accuracy have not reached the level of similar foreign products. Therefore, the paper further in-depth researched the single camera probe imaging 3D coordinate measurement technology and proposed effective measures to compensate the error were of great significance.In this paper, the measurement system was composed of an industrial CCD camera, a handheld probe, a high-performance computer and some related accessories. It installed an image acquisition card in the computer, and installed some infrared light-emitting diodes, which were measured precisely, as the feature points. In order to filter environmental visible light, it installed an infrared high-pass filter on the head of camera lens. And then, the software of the system got the measured point coordinate through analyzing the positional relation between feature points and imaging points.Firstly, this paper analyzed the hardware constitution and components selection requirements of the measurement system, and used them as the basis for constructing a set of verification experiments measuring system, which included a CCD camera, a hand-held probe, a computer and so on. It established photographic imaging model and the coordinate systems of the measurement system based on pinhole imaging model, and analyzed the transform relations between coordinate systems. It confirmed the layout scheme of feature points and trial-produced a prototype probe based on some important conclusions of the Perspective-n-point Problem, and then measured the precise coordinates of the feature point centers by using microscope and NC machine tool table. It reached the error factors and calibration methods of the photographic imaging system, and confirmed the calibration model and correction method of the image distortion.Secondly, this paper researched the core algorithms of this technology based on coplanar 5 feature points probe, included 5 steps, such as threshold determination, regional recognition of imaging, imaging center coordinates extraction, feature points and imaging points matching, and measured point coordinate solving. On basis of the imaging characteristics of the feature point element, it proposed the dynamic threshold adaptive algorithm based on the p-parameter method, analyzed several available extraction algorithms of imaging center coordinates, and researched the feature points 3D coordinates solving algorithm based on the SVD-TLS method and the measured point coordinate solving algorithm. After that, it extended these algorithms from coplanar 5 feature points to 6 or more.Thirdly, in order to verify the actual measurement results of the algorithms, the paper programmed a lot of testing procedures and verified the algorithms through a large number of experiments. It proposed the mixed programming method based on MS-VC++6.0, Sapera LT class library, OpenCV class library, and Matlab C++ class library, and proved that this method effectively shortened the process development cycle, improved the efficiency and precision of the procedures, and met the requirements of the real-time measurement by experiments. The paper programmed the camera calibration and image distortion correction procedures by using OpenCV class library, and then compared the results with Matlab toolbox based on the same calibration equipments and camera parameters, and then proved that the procedures which were programmed by using OpenCV had the same precision as Matlab toolbox, but more convenient and faster than Matlab toolbox. It got the priori parameters of the feature point element through a large number of experiments, and verified the feasibility of the threshold determination and imaging regional recognition algorithms by using these parameters, and then proved that the dynamic threshold adaptive algorithm which was based on the p-parameter method could meet the requirement of real-time threshold determination, adapt the environmental light, determine a reasonable value, and resist the impulse noise interference. After contrasting imaging center coordinates extraction algorithms (include the bi-linear interpolation gray square centroid method, the Gaussian function fitting method and other 6 methods) through experiments, it proved that the bi-linear interpolation gray square centroid method had the best stability under the hardware environment of this paper. The paper contrasted the result of measurement point coordinate solving algorithms between coplanar 5 feature points probe to 7, proved the correctness of the algorithms above and verified the feasibility of improving the measurement accuracy by increasing information redundancy of the feature points.Finally, in order to improve the accuracy of the system, this paper analyzed the error factors of the measurement system, researched the systematic error caused by feature points'angular-displacement, and proposed the error compensation method. At first, based on the LED construction and imaging principle, it analyzed refraction and reflection factors. Then established the refraction compensation function by theoretical analysis and fitted the reflection compensation function by experimental tests. After that, it researched the comprehensive compensation function depended on the two functions above. At last, on basis of the predictor-corrector method, this paper got the probe center coordinate through real-time modifying feature points'angular-displacement-error, and verified the feasibility of this method through the accuracy measurement experiments.