Design And Implementation Of 3D Reconstruction System Based On Pseudo-random Coded Structured Light

With the continuous development of computer vision,vision-based 3D reconstruction technology has attracted extensive attention from academia and industry.It is widely used in 3D inspection,intelligent robot vision,3D printing and other fields.The three-dimensional reconstruction of pseudo-random coded structured light has certain improvement space in terms of precision,real-time and robustness.Therefore,it is of great significance for the research and analysis of three-dimensional reconstruction technology based on pseudo-random coded structured light.In this paper,three-dimensional reconstruction of close-range objects is carried out.The three-dimensional reconstruction system based on pseudo-random coded structured light is analyzed and designed from three aspects: algorithm,software and hardware.The system uses the projector to actively project the pseudo-random coding pattern,and the binocular camera acquires the image and reconstructs the object in three dimensions.The optimal working distance of the system is about 350 mm,the frame rate is about 15 fps,and the fineness range of the three-dimensional reconstruction point spacing.It is 0.2mm to 3mm.The main work of this paper includes the following four aspects:(1)The generation method of pseudo-random coding is studied,and the pseudorandom coding conforming to the requirements of this paper is designed and generated.The pseudo-random code size designed in this paper is 320×147,which is composed of 7 pseudo-random codes with a size of 320×21.In a pseudo-random coding region of size 320 × 21,any 5 × 5 region is unique.The scheme of splicing multiple pseudorandom coding patterns is combined to ensure that the gray value of the image block is unique in the pole line while the pole line search is performed,and the difficulty of the pseudo-random coding is reduced.(2)A high-precision camera calibration plate was produced,and the binocular camera was calibrated multiple times to improve the calibration accuracy.The reflective markings are attached to a glass plate whose surface material is diffusely reflected to form a high-precision calibration plate.Based on the MFC framework,the camera internal reference comparator software was developed to visually compare the difference between the two camera calibration results.(3)A set of hardware and software systems for 3D reconstruction was designed.The hardware system consists of two industrial cameras,LED fill light and projector,which are light,flexible and portable.The software system developed based on MFC framework realizes camera calibration,hardware debugging,real-time display of 3D point cloud and other functions,which facilitates the verification of 3D reconstruction algorithm.(4)The three-dimensional reconstruction system of this paper improves the conventional 3D reconstruction algorithm and has significant improvements in accuracy,real-time and stability.In the process of stereo matching,the methods based on pseudo-random coding pattern,image pyramid and template matching are used to improve the accuracy and efficiency of stereo matching.The image preprocessing method using bilateral filtering improves the stability of the 3D reconstruction algorithm.In the implementation process of the algorithm,in order to improve the realtime performance of 3D reconstruction,multi-threading and GPU acceleration methods are adopted.