Professional stereoscopic video visual comfort classification method based on attention and recurrent neural network

Abstract

The invention relates to a professional stereoscopic video visual comfort classification method based on attention and a recurrent neural network. The method comprises the following steps: 1, performing scene segmentation on a training video set and a to-be-predicted video set, and performing preprocessing to obtain a disparity map; 2, performing frame-level processing to obtain frame-level features; 3, performing lens-level processing to obtain a hidden state set; 4, performing double-flow fusion: fusing the hidden state set output in the previous step by using an attention network to obtaina final hidden state; 5, outputting a classification probability of the final hidden state through a classification network, and classifying the professional three-dimensional video to be suitable forchildren to watch or only suitable for adults to watch; 6, inputting the left view and the corresponding disparity map of the three-dimensional video in the to-be-tested video set into the trained model for classification. Whether professional stereoscopic videos are suitable for children to watch or not can be effectively distinguished.

Description

technical field [0001] The invention relates to the fields of image and video processing and computer vision, in particular to a method for classifying professional stereoscopic video visual comfort based on attention and cyclic neural networks. Background technique [0002] Stereoscopic video, also known as 3D video, is different from 2D video. Its most important feature is that it has depth information, so that the presentation of the landscape in the video is no longer limited to the screen. The vigorous development of stereoscopic technology has enabled people to obtain a better viewing experience, but it has also brought some troubles. For example, watching uncomfortable stereoscopic videos for a long time will cause dizziness, dry eyes, nausea and other feelings. These adverse reactions It will discourage viewers from watching, and even affect the physical health of viewers. Therefore, how to evaluate the visual comfort quality of stereoscopic images has become a conc...

AI Technical Summary

Problems solved by technology

[0004] The purpose of the present invention is to propose a professional stereoscopic video visual comfort classification method based on attention and cyclic neural network, which solves the problem that the current stereoscopic video comfort evaluation algorithm does not consider children as audience objects, and can effectively distinguish whether the professional stereoscopic video is suitable for children

Images