Research On Surface Reconstruction Of Highly Anti-reflective And Deep Objects Based On Binocular Structured Light

With the improvement of social science and technology,3D reconstruction technology has become more and more common in production and life,and more and more close to our lives.The three-dimensional reconstruction technology of ordinary objects is relatively mature,and the quality and accuracy of the point clouds obtained by reconstruction are relatively high.However,there are still some problems with the three-dimensional reconstruction of highly anti-objects and dark objects.Due to the overexposure and underexposure of the collected image and dark object,the reconstructed point cloud is missing.In addition,due to the different positions of the images collected by the binocular camera,the reflected areas on the collected images are different,which also exacerbates the lack of point clouds.This article has conducted in-depth research on the above two problems and proposed corresponding solutions.First of all,based on the needs of the subject,this paper builds a binocular structured light system hardware platform and completes the calibration of the system camera.At the same time,a corresponding software platform is developed.Point cloud 3D reconstruction,point cloud processing and other functions and designed a human-computer interaction interface,through this software and hardware platform can complete all the work of 3D reconstruction.Secondly,this paper analyzes and studies the overexposed and underexposed areas.Due to the different reflective characteristics of the overexposed and underexposed areas,the lack of image information on the images collected by the camera affects the subsequent decoding and matching work,which leads to the The point cloud produced holes or even missing.In this paper,the design of structured light images is completed for the hardware system parameters in actual operation and the reflective characteristics of the object surface.For highly reflective objects and dark objects,different structured light images are used to reduce the underexposure and overexposure of the object surface Influence.Pure black and pure white images are introduced to remove shadow areas,reducing the amount of calculation and the introduction of related errors.Aiming at the influence of the surface characteristics of high-inverted objects and dark objects on the collected images,this paper uses multi-exposure image fusion technology,based on the characteristics of Gray code and four-step phase shift images,respectively designed different image fusion strategies,the fusion of the resulting image Able to provide complete image information.Aiming at the problem of different areas of overexposure and underexposure on the images collected by the left and right cameras,this paper proposes an image division processing method based on the reflective position of the image.After the camera is calibrated in the system,its relative position is fixed,so for a single camera,the area where the reflection occurs is relatively fixed.Therefore,in this paper,the reflective area in the image collected under a short exposure time is used as the guide area to perform image correlation processing.Different fusion weight calculation methods are used for the reflective area and the non-reflective area.The final image has complete image information and obvious image characteristics.Finally,in this paper,through the built binocular structured light system,the image fusion experiment and the three-dimensional reconstruction experiment were carried out,and the point cloud obtained by the reconstruction was analyzed for completeness,size measurement and flatness.The experimental results show that the method in this paper can obtain the complete point cloud information of high anti-reflective objects and dark objects.The size measurement errors are all around 0.07 mm,and the root mean square error value of the point cloud surface is less than 0.1.The flatness verifies that the method in this paper is effective and feasible.