Panoramic Three-Dimensional Reconstruction Method Based On Multi-View Encoded Light Field

In a series of applications based on light fields(LF),3D reconstruction is a fundamental and important task.Traditional light fields can only reconstruct a single perspective;however,they cannot reconstruct a panorama and generate accurate 3D information in areas with scarce texture features.To solve these problems,this thesis proposes a panoramic 3D reconstruction method based on a multi-view encoded light field.First,the structured light coding is used to project sine templates to the scene to generate pixel-level phase codewords to enrich the scene features.Then,the light field is collected independently from multiple angles.Furthermore,the depth value is obtained using spatial and angle information in the interior of each light field sampling viewpoint.Finally,3D information fusion is realized between different light field sampling points under the pose guidance,and panoramic reconstruction is realized.Traditional depth value estimation methods have not fully exploited the geometric constraints of encoded light fields,leaving room for improvement in terms of accuracy and efficiency.To solve this problem,this thesis proposes a single-view depth value estimation method based on encoded light field.First,a cost function construction method is proposed according to the spatial relationship of codewords to improve computation accuracy.Then,the Beetle Antennae Search(BAS)algorithm is used for global optimization of the cost function to reduce iteration times and quickly obtain depth values.Compared to traditional traversal methods,the proposed approach can obtains smaller BP and MSE of the depth map under the same number of iterations.On the basis of obtaining single-view depth information,we further proposes a 720°panoramic layout model framework,which simultaneously collects scene information along the horizontal and vertical directions,and fuses information guided by pose to achieve panoramic reconstruction.In addition,this thesis also proposes a statistical filtering-based flying pixel removal scheme to optimize the single-view point cloud accuracy to improve the reconstruction quality.The experimental results show that the method in this thesis can more effectively realize the panoramic reconstruction of objects with less data input,and the reconstruction of the object texture feature-poor area is more accurate and the RMSE is smaller.This research above investigated the application of encoded light fields in panoramic3 D reconstruction,improving the accuracy and completeness of the reconstruction,achieving the purpose of panoramic reconstruction and the effect of high-quality reconstruction,which can be applied to fields such as virtual reality and visual navigation.