Research On The SFS Based 3-D Reconstruction Method Of Industrial Endoscope

As an important technique of nondestructive testing(NDT),the industrial endoscope has been widely used in the field like aviation,automobile and ship,etc.The research on the improvement of industrial endoscope has become a popular research direction nowadays.In this dissertation,the 3D reconstruction problem of industrial endoscope is studied.A three-dimensional measurement approach of industrial endoscope based on Shape From Shading(SFS)is proposed.The new approach can use only one picture to reconstruct the 3-D shape of the scene with scale information by combing the camera matrix of stereo vision,which provides a feasible solution for the development of commercial 3D industrial endoscope systems.The works of this paper are listed as follow:1.The prototype of the endoscope has been well designed and created,which contains two CCD cameras and one point light source.All the components are integrated by a 3-D printed frame.The complexity of the structure has been well reduced which proves its high applicability.2.A novel point light source calibration method based on an accurate overall model of the point light source was proposed.The proposed method uses a high-reflective ball and a Lambertian plane as calibration objects,adapting optimization and stereo vision,the parameters of light source could be precisely obtained.3.The scaled SFS method under point lighting was created.The proposed method,based on the assumption of practical point light source model rather than the ideal parallel lighting model,applied the intrinsic matrix of stereo vision to overcome the drawback of the traditional SFS method during scale measurement and directly obtain the depth map of scene.It has low dependence on the equipment and high measurement efficiency,especially suitable for the application of industrial endoscope.