Method for compressing three-dimensional image video sequence

Abstract

The invention relates to a method for compressing three-dimensional image video sequence, comprising left view L and right view R of three-dimensional image are respectively converted into YUV4:2:0; feature points of the left view L are extracted, and then the left view L is processed by Delaunay triangular mesh subdivision; points matched with the feature points of the left view L are found out in the right view R; hollow triangular mesh of the right view R is set up, and right view R' is filled out; residual information Y' of Y component between the original right view R and the filled out right view R' is then extracted; DCT transformation as well as quantization and entropy coding are carried out, and a JPEG method is used for compressing the left view L. According to the information of the matched feature point set of the left and the right views, every triangle in the left view is treated by affine transformation; the YUV information of the left view and the right view which is newly set up is transformed into RGB information, left and right three-dimensional diagram pairs can be obtained after being uncompressed and then transformed into three-dimensional image according to the display way. The method reduces the data volume of the three-dimensional image, and the three-dimensional image which is decoded and restored has good three-dimensional effect, no uncomfortable feeling and clear image. After being decoded, the image has higher similarity with the original image.

Description

technical field [0001] The invention relates to a method for compressing stereoscopic image video sequences. In particular, it relates to a method of using delaunay triangulation to divide images into grids, compress stereo images, and reduce data volume. Background technique [0002] In the natural world, human beings can observe real three-dimensional scenes through binoculars. How do human beings obtain stereoscopic vision by observing things with binoculars? This is a question that has troubled many scientists for a long time. The function of human eyes has been studied since Descartes, and Newton once guessed about the visual chiasm of the eyes. In 1838, Wheatstone invented the stereoscope, which revealed the reason why human beings obtain stereoscopic vision—parallax gives people a sense of stereoscopic vision. The average distance between the two eyes of a person is about 65mm, and the angle of observing objects is slightly different, resulting in parallax, that is...

AI Technical Summary

Problems solved by technology

The video representation method of YUV4:2:0 format, which is common in video processing, takes advantage of the feature that human eyes are not sensitive to chroma, and reduces the resolution of chroma signals U and V to 1 / 4 of the resolution of Y signal, while In general, the human eye cannot detect the loss of chromaticity information of the actual displayed image quality

Images