Video predictive encoding method based on neural networks

A neural network and predictive coding technology, applied in the field of video predictive coding based on neural network, can solve the problem of increasing coding complexity, and achieve the effect of improving coding efficiency, reducing complexity, and optimizing decision-making process.

Active Publication Date: 2018-11-30
UNIV OF ELECTRONICS SCI & TECH OF CHINA
View PDF5 Cites 11 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

[0008] The purpose of the present invention is: in order to solve the problem that the existing utilization rate distortion optimization recursively searches for e

Method used

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
View more

Image

Smart Image Click on the blue labels to locate them in the text.
Viewing Examples
Smart Image
  • Video predictive encoding method based on neural networks
  • Video predictive encoding method based on neural networks
  • Video predictive encoding method based on neural networks

Examples

Experimental program
Comparison scheme
Effect test

Embodiment 1

[0041] Such as Figure 4 with Figure 5 As shown, the present embodiment provides a neural network-based video predictive coding method, including the following steps:

[0042] S1. Input a coding tree unit with a size of 64×64, and use a Bayesian classifier to make a rough judgment on it to determine whether the SKIP mode is adopted. If so, determine that the current coding tree unit is not divided down, and directly obtain the coding tree unit The coding unit size decision of , otherwise, execute S2;

[0043] The judgment method of the Bayesian classifier is as follows:

[0044] Consider whether to use the SKIP mode as a two-category problem, and the two categories are marked as y 1 and y 2 , P(y j ) is the prior probability, and the conditional probability of the class is P(x|y j ), j is a mark of two categories, which can be 1 or 2, which means that SKIP is not executed or executed, P(y j |x) is the posterior probability, the calculation formula is:

[0045]

[0...

Embodiment 2

[0072] This embodiment is further optimized on the basis of Embodiment 1, specifically:

[0073] The three neural networks in the S2 are trained using the training data set, and the training method is as follows:

[0074] Step 1. Perform data augmentation preprocessing on the images in the training data set;

[0075] Step 2, performing 0-1 regularization on the preprocessed image data;

[0076] Step 3. The regularized image is input into the first neural network, the regularized image is divided into 4 equal parts and then input into the second neural network, and the regularized image is divided into 16 equal parts and then input into the third neural network. The neural network is trained;

[0077] In the step 1, the preprocessing of data augmentation on the images in the training data set specifically includes four kinds of image transformations, and the four kinds of image transformations are specifically:

[0078] a. Flip the image horizontally and vertically;

[0079...

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
Login to view more

PUM

No PUM Login to view more

Abstract

The invention discloses a video predictive encoding method based on neural networks, and relates to the technical field of the video compressed encoding. The method comprises the following steps: S1,inputting an encoding tree unit with a size of 64*64, and performing rough judgment on the encoding tree unit through a Bayes classifier to judge whether to adopt a SKIP mode, if yes, it is determinedthat partitioning is not performed on the current encoding tree unit, and an encoding unit size decision of the encoding tree unit is directly obtained, otherwise executing the S2; and S2, performingencoding unit blocking decision on the depth of the encoding tree unit in parallel through three neural networks to obtain a blocking result of the encoding unit; S3, obtaining an encoding unit sizedecision through the encoding unit blocking result obtained in the step S2; and S4, performing predictive encoding according to the encoding unit size decision obtained in the step S1 or S3 so as to obtain an encoding result. In the premise of ensuring the encoding performance, the encoding complexity can be greatly reduced, and the encoding efficiency is improved.

Description

technical field [0001] The present invention relates to the technical field of video compression coding, and more specifically relates to a neural network-based video prediction coding method. Background technique [0002] Video coding generally also refers to video compression, which uses methods such as prediction, transformation, quantization, and entropy coding to reduce redundancy in video data as much as possible, and use as little data as possible to represent video. However, under the bandwidth limitation of the existing network, the distortion degree after video encoding is relatively large, and the final video viewing result is not good. [0003] Traditional video coding is based on the HEVC standard, using a hierarchical quadtree structure, introducing coding tree units (CTUs), coding units (CUs) and prediction units (PUs), and adjusting the size of coding units through quadtree traversal. and the mode of the prediction unit are selected. The HM encoder adopts a ...

Claims

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
Login to view more

Application Information

Patent Timeline
no application Login to view more
IPC IPC(8): H04N19/176H04N19/96H04N19/119H04N19/85H04N19/50G06N3/04G06K9/62
CPCH04N19/119H04N19/176H04N19/50H04N19/85H04N19/96G06N3/045G06F18/24155
Inventor 赵丽丽张梦王文一张汝民
Owner UNIV OF ELECTRONICS SCI & TECH OF CHINA
Who we serve
  • R&D Engineer
  • R&D Manager
  • IP Professional
Why Eureka
  • Industry Leading Data Capabilities
  • Powerful AI technology
  • Patent DNA Extraction
Social media
Try Eureka
PatSnap group products

Browse by: Latest US Patents, China's latest patents, Technical Efficacy Thesaurus, Application Domain, Technology Topic.

© 2024 PatSnap. All rights reserved.Legal|Privacy policy|Modern Slavery Act Transparency Statement|Sitemap