Study On Performance Optimization Of Portable Visual Metrology System

With the increasing requirements of digital assembly in assembling and manufacturing of aerospace, large ships, high-speed trains, more strict requests are made for field measurement. However, traditional large-scale measurement instruments can hardly satisfy the demand. As a result, large-scale combined measurement method gradually appears. Portable visual metrology system plays an important role in large-scale combined precise measurement engineering for its non-contact, high precision, wide measurement range, high efficiency and high degree of automation.Based on the principle and key technology of portable visual metrology, performance optimization of measurement system is investigated, including system reliability, measurement accuracy, easy operating and convenience. Feature point matching is the key to system reliability, it is mainly investigated and a match method based on coding network is proposed. In order to improve the accuracy and to get rid of the limitation of traditional self-calibration model, a measurement method based on angle measurement model is proposed. Besides, based on finite element analysis and semi-parametric model optimization, systematic error compensation method for traditional self-calibration model is investigated. Easy operating and convenience are concerned, thus system software development method and database management are introduced in detail. Finally, performance optimization methods above are integrated into the portable visual metrology system, and the effect of system performance optimization is verified by practical measurement engineering. The main work of this paper is as follows:1. To reduce the influence of high mismatch rate on system reliability in the case of feature points distributing intensively and complex measurement enviorenment, an indirect epipolar line matching method based on coding network and an intersecting on space matching method are proposed. The former method obtains the initially optimized internal and external parameters by establishing coding network, which improves the fundamental matrix solving precision indirectly and then improve the epipolar line correct matching rate. The later method recovers the image points into the three-dimensional space with the initially optimized camera internal and external parameters, and determines the match relationship of corresponding image points with the constraint relationship of intersecting on space.2. A portable visual metrology method based on angle measurement model is proposed. Firstly, the limitation of traditional self-calibration model and its influence on the measurement accuracy are analyzed. Secondly, feasibility of angle measurement is investigated. Calibration principle of precise angle measurement platform is introduced in detail, and the establishing method and determination of image point angle are introduced. Finally, error source of angle measurement and its influence on spatial point localization accuracy are analyzed.3. Determination and compensation method of traditional self-calibration measurement model systematic error are investigated in detail to improve the accuracy of traditional self-calibration model. In this paper, two systematic error determination methods are proposed. One method references the finite element analysis, which determines the distortion model systematic error of partition corner combining with the angle-measurement calibration platform and determines the distortion model systematic error of each image point with bilinear interpolation. The other method directly compensates the systematic error using the collinear equation to construct semi-parametric model and optimizes all the unknown quantities simultaneously with two-step kernel estimation algorithm.4. Combined with practical application, performance of the optimized portable visual metrology system is verified. Software is the kernel of portable visual metrology system, and the development goal and method are introduced in detail. Besides, database management method aiming at improving data processing speed and logic is introduced. Finally, application of our system in large-scale system engineering is displayed combining with the measurement requirement.