Design And Implementation Of 3D Reconstruction System Based On Binocular Vision And Structured Light

With the continuous development of machine vision technology,binocular stereo vision technology is favored by more and more scholars,and the surface defect detection and texture detection of curved surface products is an important development direction of binocular stereo vision.Surface defect detection and texture detection of curved surface products are one of the difficult problems in product detection.Most of the existing detection equipment,in the detection process of curved surface products,adopts the two-dimensional detection method.Through the parallel light source,the products are polished,and the single camera collects the image mode.According to the different product defects and the reflective degree of the texture,the method can only detect the obvious defects And the products with large texture differences have effect,which can not meet the high-precision detection requirements of complex products.In order to achieve the accurate detection of surface products,3D reconstruction of surface products is an indispensable part.In this paper,aiming at the surface defect detection of curved surface products,a three-dimensional reconstruction system based on binocular vision structured light is developed by using the lighting method of different frequency structured light.First of all,according to the research status of binocular vision 3D reconstruction technology at home and abroad,this paper makes an investigation.According to the types of products detected by binocular vision,the detection requirements and accuracy requirements of products are analyzed,as well as the overall scheme design of the system.Complete the optical platform design and hardware selection,including camera,lens,projector,etc.In addition,the camera calibration part is introduced.The chessboard calibration method is used to calculate the internal and external parameters of the camera.A variety of calibration boards are used to verify the calibration error,and the conversion relationship between the image coordinate system and the world coordinate system is obtained.Secondly,the image acquisition part of this paper is saved to the projector by using the falling grating image,which projects the different frequencies of the positive and negative structured light through the projector,then uses the USB interface to control the projector,and takes pictures of the camera instructions in the external trigger way.In this paper,basedon the existing three-dimensional reconstruction of binocular vision,a faster and more accurate algorithm of binocular matching is studied.A phase unwrapping algorithm based on Multi frequency heterodyne is proposed,and the algorithm is improved to adapt to more changeable application scenarios.Then,this paper analyzes and designs the requirements of image acquisition system,software module and package module respectively,improves various algorithms combined with product detection requirements,and completes the application software design and development of 3D reconstruction system,so as to complete the research and development of 3D reconstruction system of curved surface products.Finally,this paper tests the developed 3D reconstruction system,through the analysis of camera acquisition time,system operation time and detection accuracy,the final experimental results show that the designed 3D reconstruction system based on binocular vision structured light meets the needs of project detection and the 3D reconstruction effect of complex surfaces.