Research On The Precision Of 3D Reconstruction Based On Matrix Analysis Method

Three-dimensional reconstruction is a new technique that can get the three-dimensional structure through two or more images. To get the three-dimensional model and the scale information is the ultimate purpose of the three-dimensional reconstruction. In recent years, the technology of 3D reconstruction has developed rapidly because of the advantage in data acquisition. The application of 3D reconstruction also has a breakthrough in the field of close-range photogrammetry. However, 3D reconstruction technology was initially only be used in the field of computer vision, which needs less demanding of accuracy. Thus the research on the accuracy of the 3D reconstruction is less than the other fields. For the application of 3D reconstruction in close range photogrammetry, precision problems must be fully demonstrated. In this paper, a method on evaluating the accuracy of 3D reconstruction is present, and the specific research and innovation are as follows:1) The traditional photogrammetry and 3D reconstruction process based on multi-view geometry were described. Through the analysis of the method of precision evaluation of traditional photogrammetry and the 3D reconstruction model, it is concluded that the existing theory is not applicable to assess the accuracy of the results of 3D reconstruction;2) The shortcomings of the existing constraints in the process of 3D reconstruction was briefly introduced, and a matrix analysis method was present, which can get the quantitative accuracy of the results of 3D reconstruction more rapidly; and the effect of fundamental matrix and projection matrix to the final results of 3D reconstruction is analyzed;3) Finally, the image data got by close-range photogrammetry was used to verify the method of precision evaluation. The results show that the method can estimate the accuracy of 3D reconstruction and give some guidance in the actual work.