Transparent Surface Polarization Imaging Measurement System And Three-dimension Reconstruction

Three-dimension reconstruction, as an important research direction of computergraphics, is widely used in industrial design and manufacture, construction, biology andother fields. Since the traditional contact measurement is slow, and the simple non-contactmeasurement is limited by the resolution of instruments, the three-dimensional shape can notbe obtained well. Second, to objects such as mirror, transparent materials and underwaterobjects, because of the reflection refraction effects on optical instruments, the traditionalmethods turned to be pale. So a measurement based on polarization imaging theory isproposed in this thesis which can accurately, rapidly solve this problem.Feature points and fitting algorithm is the core of digital image3D modeling. Apparentobject measurement may be affected by background, but its polarization state is not changed.Therefore, choosing polarization state as the character of object, using polarization imagingprinciple, an equation between polarization parameters and surface vector can be set up.In this thesis, several common fitting algorithms are discussed, from which the Zernikepolynomial algorithm is chosen to gain the3D topography simulation expression.Based on the polarization imaging3D modeling study, a set of polarization imagemeasurement software system with real-time display, data processing, polarization imagedisplay,3D shape display is developed. Currently, the software system robustness is strong,friendly and interactive interface, fully functional.