A Light Field Data Compression Method Based on Homography Transformation

Abstract

The invention discloses a method for compressing light field data based on homography transformation, comprising: inputting the original light field data set Φ(S 1 ,S 2 ,...,S i ), the sub-aperture image S i Divided into central sub-aperture image C i and the corresponding neighborhood subaperture image A i‑j ; According to the central sub-aperture image C i and the corresponding neighborhood subaperture image A i‑j Find the homography matrix H i‑j , based on the homography transformation, the neighborhood sub-aperture image A i‑j projected onto the central subaperture image C i The transformed neighborhood sub-aperture image A′ is obtained on the imaging plane i‑j , the transformed neighborhood sub-aperture image A′ i‑j with the central subaperture image C i Subtract to get the residual map R i‑j ; For all central sub-aperture images C in the original light field data set i and the residual image R i‑j scan separately to generate the central video sequence V C and the residual video sequence V R ; Center video sequence V C and the residual video sequence V R compressed separately. The present invention effectively reduces redundant information in the air domain, thereby effectively improving coding efficiency.

Description

technical field [0001] The invention relates to the fields of computer vision and digital image processing, in particular to a method for compressing light field data based on homography transformation. Background technique [0002] Recently, the plenoptic camera based on the microlens structure has aroused widespread attention in academia and industry. The plenoptic data captured by it can record the position information and angle information of light at the same time. It is widely used in 3D reconstruction, stereoscopic display imaging and virtual reality. It shows great application prospect. However, compared with images captured by traditional cameras, plenoptic images have ultra-high resolution, and at the same time, there is huge information redundancy in the special microlens pixel distribution form. The surge in data volume has brought enormous pressure to the transmission and storage of light field data, so an effective light field data compression scheme is urgent...

AI Technical Summary

Problems solved by technology

[0003] For the original light field image captured by the light field camera, the sub-aperture image generated by its rendering can be compressed. The compression scheme can be roughly divided into three categories: the first category is to use different scanning forms to generate pseudo-aperture images from sub-aperture images. The video sequence is then compressed based on the video encoder. This type of scheme has certain uncertainty because it cannot provide theoretical compression efficiency analysis; the second type of scheme is to convert sub-aperture images from space to time according to different perspective relationships Domain conversion, and then use multi-view encoding to complete the compression of light field data. In this scheme, multi-view encoding needs to configure camera parameters, and at the same time, the number of views of the sub-aperture image and the number of views of the multi-view encoder make it difficult to implement. ; The last type of scheme is to use the parallax relationship of adjacent sub-aperture images, and introduce a parallax compensation coding tool in the video encoder to reduce spatial redundancy information. This type of scheme fails to effectively utilize the macropixels in the light field image. The similarity relationship, which limits the further improvement of coding efficiency

Images