Research On High-Precision Methods Of Visual Measurement Of The Shape And Size Of The Industrial Parts

In the highly competitive environment of the current global economy, the processing enterprises of all kinds of industrial parts are exploring different ways to improve their product quality. The measurement of the shape and size of produced parts is an important part of products quality assurance. According to statistics, since the beginning of the21st century, the annual expenditure for the inspection of the quality is up to tens of billions of dollars only in the United States. At present, the domestic industrial parts manufacturing industry of processing intensive and high quality requirements, such as aircraft parts manufacturing industry, their parts inspection all depends on coordinate measuring machines. As we all know, this method's operation is complex, which speed can not meet online real-time requirements, and its detection accuracy will be affected by operator proficiency. This shows that the traditional measurement means (such as coordinate measuring machine, theodolite/total station industrial measurement systems, etc.) have already been difficult to meet the requirements of the further inspection requirements of the industrial department, so this is the exactly cause of the research and development of the visual measurement methods. The research and development of visual measurement methods and systems can compensate for the deficiencies of the current detection method to solve the practical problems of the industrial inspection department. Although the visual measurement technology has reached the practical stage in foreign, and a variety of products have been put into use, but the domestic visual measurement technology is still in the preliminary development stage. Therefore the study of independent corresponding visual measurement methods and the development of the utility system with high performance, low cost and easy to operate are necessary and urgent requirements.In this paper, combining the practical needs of high-precision measurement technology and the characteristics of the common parts in the aircraft parts processing industry, the error visual measurement methods of geometry and the shape of the parts are be researched. These studies not only will improve the production efficiency and product quality in aircraft parts machining industry but will also have an important role in promoting the visual measurement techniques to other industrial parts manufacturing industries. The main content in this paper includes:1) The contour's basic graph element segmentation method based on the edge points of the part contour. The main research content in this section is multi-feature extraction joint method based on part contour edge points to improve the accuracy of parts edge contour primitive feature extraction.2) The image acquisition and visual measurement method based on the large-diameter telecentric lens. This method can overcome the deficiencies of the existing two-dimensional image measuring instrument. The main research contents include the selection and design of hardware structure, hardware debugging method research (including the method of telecentric lens distortion evaluation, determining method of the verticality between the camera main optical axis and the inspection platform, and the image calibration method based on telecentric lens), the contour's characteristic parameters of single/multiple parts with additional constraints, and the method of automatically comparison with the design data. This method overcomes the shortcoming of existing two-dimensional image measuring instrument, in which accuracy limited detection range. Our method can capture the whole image of disposable parts and achieve high precision measurement automatically. The experiment proved that the measurement accuracy can be achieved0.01mm, which has meeting the accuracy needs of small and medium-sized plane parts measurement.3) The visual measurement method of flat sheet parts based on single digital camera. That is research and design the monocular visual inspection method based on the large relative amplitude non-metric digital camera for larger flat sheet parts. The research contents include camera parameter calibration based on two-dimensional DLT and bundle adjustment, a single image exterior orientation elements solver based on the information of planar control point, the method of image distortion corrected and vertical corrective, and the part contour features obtained method based on multi-feature extraction method. At last using detection results of the large amplitude non-measurement digital camera for larger flat sheet parts to comparison with the standard data, the experimental result shows that the measurement error is less than 0.1mm, meeting the measurement accuracy requirements of substantially size flat sheet parts.4) The stereo vision measurement method based on the model and generalized point photogrammetry. Combined with the generalized point photogrammetry theory, visual measurement method based on stereo camera is researched and designed for the parts with specific geometric model (such as cylindrical parts). The main researchs include the mathematical model description and contour expression method of parts, the adjustment model establishment method based on generalized point photogrammetry, and three-dimensional visual measurement and inspection processes. And at last experimental results are presented, which shows that for industrial parts in which the corresponding points in the strict sense can not be extracted, using the generalized point photogrammetry adjustment model can achieve high measurement accuracy. Only use the1.3megapixel ordinary industrial camera, the measurement accuracy can reach0.03mm, meetting the requirements of part three-dimensional measurement.