| Structured light imaging method is widely used in 3D surface reconstruction of objects,and the accuracy of 3D reconstruction has an important impact on subsequent measurement,detection,recognition and other tasks based on reconstruction results.However,in practical applications,the complex shape characteristics,such as convex structure,mirror structure and uneven reflectance distribution inevitably interfere with the imaging process of structured light,resulting in image quality degradation and 3D reconstruction accuracy reduction.In the face of complex surface shape,the structured light imaging method still faces the following problems:First,the block of projected light by the convex structure will induce shadows in the image,and the missing phase coding information in the shadow region of the image is difficult to accurately repair;Secondly,the high intensity reflected light caused by the mirror structure will induce saturation in the image.In order to avoid saturation,the reflected light will be weakened and the image quality will be reduced.In addition,the uneven distribution of surface reflectance will lead to the uneven distribution of reflected light intensity,which will reduce the image contrast.Finally,the uneven distribution of reflected light intensity enhances the intensity diffusion of the image and leads to the fringe distortion.In view of the above problems,the main research contents and innovations of this dissertation are as follows:1.A structured light imaging method for the shape covered by shadow is proposed,which can realize the classification of shadow regions in fringe images,and use different methods to repair the missing phase information according to their types,and effectively eliminate the shadow induced 3D reconstruction defects.To be specific,firstly,according to the neighborhood information of the shadow region,it is divided into two categories:effective shadow region and invalid shadow region.Then,for the effective shadow region,a neighborhood information fusion phase estimation method is proposed to repair the phase information.The method uses Kalman filter algorithm and kernel density estimation method to integrate the phase gradient and phase distribution of the shadow region neighborhood,which improves the precision of phase repair.For the invalid shadow area,the phase of the background area corresponding to the shadow area is directly used to fill in order to avoid introducing additional phase errors.Experimental results show that the proposed method can accurately repair the phase information of the shadow region and improve the 3D reconstruction accuracy of the shape covered by shadow.2.A structured light imaging method for the mirror-like shape is proposed,which can suppress mirror reflection from the source,avoid image saturation caused by high intensity reflected light,and realize high precision 3D reconstruction of the mirror-like shape.Specifically,the receiver of the reflected light is converted from a real camera to a virtual camera by establishing an unfixed view structured light imaging system.Then,the mirror-like bidirectional reflectance distribution is adjusted by utilizing the flexible position of the virtual camera,and the extrinsic matrix transformation equation is established to correct the system calibration parameters according to the position of the virtual camera.In addition,an optimal bidirectional reflectance distribution adaptive adjustment algorithm is designed based on the system,which effectively improves the reconstruction efficiency.The experimental results show that the proposed method can effectively avoid image saturation caused by mirror reflection,and improve the 3D reconstruction precision of mirror-like shape.3.A structured light imaging method for the shape with uneven reflectance distribution is proposed,which can achieve accurate matching between the shape reflectance distribution and the coded value of the projection pattern,and reduce the interference of the large-scale reflectance variation.Specifically,firstly,a projection reflection imaging model and a crosstalk model are established based on the structured light imaging system.Then,based on this,a nonlinear projection intensity adjustment method is designed to calculate the pattern coding value matching the reflectivity distribution of the current surface shape,and the corresponding projection pattern is generated according to the calculated results.In addition,in order to improve the efficiency of 3D reconstruction,a cyclic utilization of fringe image technology is proposed to reduce the number of projection patterns required.The experimental results show that the proposed method can effectively improve the 3D reconstruction accuracy of the shape with uneven reflectance distribution,and enhance the robustness of the structured light imaging method to the wide range of reflectance changes.4.A intensity diffusion compensation method for fringe image is proposed,which can quantify the intensity diffusion caused by the uneven distribution of reflected light intensity in fringe image and compensate the diffused intensity to correct the fringe distortion.Specifically,based on the motion law of photocarriers and the coding characteristics of fringe patterns,the process of intensity diffusion caused by the uneven distribution of reflected light intensity is studied.Then,the mechanism of fringe distortion caused by diffused intensity is analyzed.On this basis,a fringe image intensity diffusion model is constructed to quantify the intensity.Finally,the fringe distortion is corrected by compensating the diffusion intensity to improve the image quality.The experimental results show that intensity diffusion is one of the factors causing fringe distortion.The proposed method accurately quantifies and compensates the intensity of diffusion in fringe images,and improves the accuracy of 3D reconstruction of structured light based on fringe patterns.In summary,based on the above methods,this dissertation establishes a shapefeature-oriented structured light imaging method,improves the accuracy and efficiency of 3D reconstruction based on structured light imaging,solves a series of problems faced by 3D reconstruction of complex shape based on structured light imaging,and proves the effectiveness and necessity of the proposed method. |
