Researsh Of 3D Surface Topography Measurement Technology Based On Fast And Automatic Registration Of Continuous Sequence

In recent years, with the rapid development of sensor technology,3D measurement technology is widely applied in all walks of life. As an important research direction of 3D measurement, grating projection phase measuring profilometry has borad application prospects due to its high density, fast speed, high precision and low cost.Aiming at some key problems and deficiencies on measurement accuracy and full-range registration of grating projection phase measuring profilometry, this dissertation proposes some key algorithms and discusses some practical methods for automatically, accurately, rapidly implementing the full-range measurement and 3D point cloud registration. The main research contents and results are listed as follows:1. System calibration result affects the precision of the photogrammetric system to a great extent. After analysizing the system calibration tasks and specific contents, ZHANG method is used to calibrate our measuring system that consists of stereoscopic camera and projector. In order to achieve an accurate calibration result, this dissertation adopts a convenient scheme, which uses the completely-flat LCD monitor as calibration target and mades an optimized shooting plan which combines all the essential conditions including translation, tilt and rotation. The experimental results show that this scheme can achieve a high calibration precision and flexibility, and then provides accurate data for the following steps.2. Accuracy, speed and reliability of phase measurment determines the overall performance of measurement system. It’s very important for the selection of algorithm to evaluate Temporal Phase Unwrapping (TPU) algorithms comprehensively and accurately. This dissertation proposes a method to evaluate the reliability of TPU through actan and UINIT functions, and verifies its correctness by comparing and analysizing the results from the proposed method and other methods. Using available precision evaluation method, the accuracy of five typical TPUs are verified by theoretical analyses and experiments. Furthermore, the comprehensive performance of five typical TPUs are compared and summarized from four aspects:data acquisition time, computation cost, reliability and precision.3. In order to improve the accuracy of phase measurement, two phase measurement schemes with low nonlinear error and high precision are proposed after deeply analysizing the defects of Least Squares Fitting (LSF) algorithm. First, for solving the problem that LSF cannot eliminate the nonlinear error from the measuring system, this dissertation proposes a composed scheme, which integrates current research findings of phase measuring profilometry, based on Three-Frequency Heterodyne TPU, and combines the double four-step phase-shifting and LSF (or arithmetic average) method. This dissertation proves its feasibility theoretically and evaluates its performance comprehensively. The experimental results show that this method can effectively improve the measurement precision and correct the nonlinear error at the same time. Second, for solving the problem that LSF cannot stably imporve the precision of different TPU, Phase Averaging Method (PAM) is proposed. PAM increases the number of the repeated sampling at the maximum time point of time sequence in a certain TPU and employs double four-step phase-shifting to correct nonlinear error and uses arithmetic averaging to decrease the random error. Furthermore, two typical applications of PAM are discussed. The experimental results show that PAM is a flexible, fast and stable phase measurement scheme. PAM provides the flexible seclection for the TPU algorithm and length of time sequence and thus can be used in the occasion that needs to overally consider the precision, speed and reliability.4. The depth recovery method based on binocular vision is a flexible method than other ones because it needs less geometric constraints and parameters to be calibrated, but, it consumes a large amount of time in its two-dimensional phase match stage, therefore, a two-dimensional phase feature matching algorithm based on phase space mapping is proposed. According to the corresponding relationship between the obtained two-dimensional phase coding from the stereoscopic camera and the projected two-dimensional phase encoding from the projector, this algorithm maps the two camera’s phase spaces to the projector’s respectively, and then mathches the two camera’s images in the projector’s image space coordinates system. The experimental results show that this algorithm is of higher accuracy and faster speed than other methods. Futhermore, its matching results can be used to calculate the exterior orientation elements between two cameras. This provides a direct theoretical basis and practical algorithm for the further continuous measuring and point cloud registrating.5. Aiming at the problem that available point cloud registration methods need either the initial origination elements or other auxiliary devices and manual participation, this dissertation proposes an all-range stereo registration scheme based on projection space constraints. This scheme, based on the projection phase space constraint, can realize the absolute orientation and 3D point cloud registration among all camera stations and enables two subsystems of measurement system to move independently and crossingly covering the whole surface area of the measured object. The proposed scheme uses ICP algorithm to accurately compute the orientation elements. Experimental results show that this scheme can achieve rapid and precise registration of 3D point clouds. Supported by the mobile platform, this scheme can automatically and fast measure the large scale objects, without accurately calibrating any geometric parameters.Dissertation discusses some key issues including the measure system and its automation, accuracy and speed, in terms of the full-range measurement and 3D point cloud registration, and proposes some practical algorithms and schemes that may have some values for the related researchs.