Research On Object Depth Measurement System In Three-dimension Vision Technology

Object depth measurement system has great application value in the fields of machinery, electron and civil engineering and etc, while three-dimension vision technology is one of the important ways to realize it. This paper takes practical problem of object depth measurement as an entry point to research, combining with three-dimension vision solution, puts more efforts into the study of the important role played by approximation of functions in the calibration problem of conventional camera, meanwhile it demonstrates the feasibility of this scheme by doing comparative analysis between theory and simulation experiment and finally completes system design to successfully implement the function of this system.Binocular three-dimension vision model is adopted in this scheme design. Interpolation method and sub pixel method are using in acquisition process of characteristic points of sample. Laser projection together with matching method of characteristic points based on a sort order to simplify the problem in matching process of characteristic points, moreover, through constraint validation to ensure the objectivity and authenticity of characteristic points.Traditional calibration method was using parameters of photography geometric model which required solving parameters of both internal and external of each camera. Experimental results showed that stability and accuracy of this method was not satisfactory in the influence of many other factors. With respect to application environment and equipment condition, three-dimension coordinate function directly goes into approximation of function, hiding calibration of camera parameters, taking intelligent calculation as approximation of function, and theoretical and experimental simulation have been done. This method which shields direct solution of parameters hiding influencing factors from either unfavorable environment or camera of low accuracy and provides direct mapping from two-dimension coordinate to three-dimension coordinate so as to achieve doing comparative analysis of three-dimension vision measurement by stable and effective acquisition process of characteristic points. Final experimental data and results show that measurement results are all within controllable accuracy.