Panoramic depth camera system capable of acquiring 360-degree scene structure

Abstract

The invention discloses a panoramic depth camera system capable of acquiring a 360-degree scene structure. The system is characterized in that eight fisheye lens cameras are placed in the range of a circle, objects in the periphery 360-degree space can be acquired at a time, and the base length is appropriately selected according to the scene range. By adopting the technical scheme, the system achieves the advantages that being different from an active detective type depth camera, the scheme still can normally work outdoors; according to a late algorithm, the camera can provide the 360-degreethree-dimensional scene structure and can be applied to the fields of seamless panoramic photo composition, three-dimensional measurement, three-dimensional modeling and the like; due to multiple projection constraint, the noise resisting capacity of visual three-dimensional reconstruction can be improved, and the system robustness can be improved; and the three-dimensional objects of the 360-degree space can be shot simultaneously without multiple times of scanning and composition, and therefore the shooting efficiency of the system is improved.

Description

technical field [0001] The invention relates to the field of virtual reality, in particular to a panoramic depth camera system capable of collecting 360-degree scene structures. Background technique [0002] Now the virtual reality (Virtual Reality) market is hot. Among them, panoramic video has become one of the fastest landing products in the VR market due to its low production cost and short cycle time. The production process of panoramic video is usually first to use multiple cameras (greater than or equal to 2) to shoot the scene at the same time, and then to post-splicing the photos or videos taken by multiple cameras to obtain a 360x180-degree panoramic picture. When watching through the VR helmet, the client can turn the head angle arbitrarily to obtain the visual content of the entire scene. [0003] In the current panorama shooting schemes on the market, since the three-dimensional structure of the captured scene cannot be obtained, in the algorithm model of pano...

AI Technical Summary

Problems solved by technology

[0007] 1. If an active detection depth camera is used, the working distance of the depth camera is fixed, and the working angle of view is small. To complete the modeling of the 360-degree space, multiple scans are required, and because of the working principle of the depth camera, if it is outdoors, it will be affected by various Interference doesn't work well, so mostly used indoors

[0008] 2. If it is a binocular camera (such as figure 1 As shown), the existing solutions generally place two ordinary field of view lenses side by side, so the working angle of view is small

And if two cameras are used to measure the three-dimensional space, the space constraints will be insufficient, so it will be easily affected by noise interference, resulting in insufficient three-dimensional reconstruction accuracy

Images