Research On Surface Distortion Correction Technology In Polarization 3D Imaging

With the rapid development of photoelectric technology and computer vision,the accuracy of three-dimensional imaging becomes higher and higher but the cost of it is reduced gradually,so it is more and more widely used in such fields as medical treatment,architecture,transportation and archaeology.However,commonly used three-dimensional imaging technology under the control of multiple imaging devices and complex optical systems is hard to meet the demand of the low cost and the high abundance of detail.Polarization imaging technology can implement 3D imaging with high richness of detail in a simple imaging system of a single detector combined with polarization device,which is an option to implement cost-effective three-dimensional imaging technology.Through obtaining the polarization information of the target object,polarized 3D imaging uses the relationship between the polarization characteristics of reflected light of the target subject and the surface geometric structure,to compute the surface normal and reconstruct the three-dimensional shape of the target object.This thesis firstly analyzes the characteristics of reflection and refraction of light and Fresnel formula,and on that basis analyzes the relationship between the polarization characteristics of the light scattered on the surface and surface geometric structure to establish polarization 3D imaging model.And then this thesis studies and analyzes surface normal correction,shape from shading and gradient-based three-dimensional reconstruction method to implement polarization 3D imaging.In the study of polarizing 3D imaging,due to the shadow on the surface caused by the shape of the target object or optical source,the reconstruction results will have a large distortion in the shadow area,which will affect the overall three-dimensional imaging effect.Aiming at the distortion problem in the shaded area,this thesis proposes surface distortion correction method based on the shadow image model.Shadow removal based on shadow image model firstly corrects the deviation of the color-lines of the shadow image,in order to eliminate the influence of different materials for the shadow detection.Then this method distinguishes the non-shadow,umbra and penumbra area from the shaded image after the shadow detection,to build the shadow mask.And the shadow of these three kinds of shadow areas in the shadow image will be removed respectively.Finally this method applies the intensity image the shadow of which has been removed to polarization 3D imaging by using SfS method.The experimental results of the shadow removal show that the shadow removal method used in this thesis is capable of implementing highly accurate shadow detection,retention of the image intensity in non-shadow areas and the reduction of image intensity in most shadow areas,which has a certain reliability.It contributes to the realization of the surface distortion correction for polarized three-dimensional imaging.The experimental results of surface distortion correction show that the surface distortion method for polarization 3D imaging in this thesis is feasible and has a certain reliability.It is also possible to implement more accurate 3D surface recovery of the target object,while still retaining detailed surface texture in the three-dimensional surface that is implemented with polarization 3D imaging.Through the analysis,the surface distortion method in polarization 3D imaging of this thesis is feasible to implement polarization 3D imaging with low distortion and high richness of detail,which provides new thinking for more accurate recovery of three-dimensional shape of the target surface.