Digital Three - Dimensional Integrated Imaging Reconstruction Based On Lens Array

Integral Imaging (Ⅱ) is one of the mainstream research directions in the field of 3D reconstruction and display techniques. It is a kind of technologies based on lens array which to obtain and reconstruct the goal scene for 3D.Compared with other 3D display techniques, II has obvious advantages. It can provide a 360-degree view and continuous viewpoints within the scope of the perspective. Importantly, there is no need for taking any auxiliary equipment (special glasses, etc) to watch the scene. In this paper, the research and analysis has been carried out for the Ⅱ system based on lens array. Captured images through the lens array, the 3D reconstruction for the goal scene has been completed ultimately and some problems are solved as following:Firstly, The traditional calibration method is applied to calibrate the sub images, but it can not get accurate center positions of the sub images. In order to improve the accuracy and correct the distortion fast, the combination of laser calibration method and characteristic line method is put forward to obtain the sub-images center positions and estimate the parameters of lens distortion.Next, for the feature points matching of sub images, the principles of Harris and SIFT algorithm are stated. And the experiment on extracting and matching the feature points is carried out with the algorithm of the combination of Harris and NCC, and SIFT, but the effect is not good because of poor matching points and hugely time-consuming. For this problem, ASIFT algorithm with the combination of the bilateral filter template is introduced for the feature points extraction and matching, which makes the improved ASIFT can suppress the noise. Compared with Harris and SIFT algorithms, BF+ASIFT can get more feature points without hugely time-consuming.Finally, for the 3D image reconstruction problem, the principle of cutting chip is applied.Before the 3D reconstruction, the method of least square method to optimize the depth value is put forward, which makes depth value achieve an optimal result, and improves the display quality to some extent. In the process of display, the use of bilateral filter makes the image look more smooth and delicate.