Looking for breakthrough ideas for innovation challenges? Try Patsnap Eureka!

Optical flow information compression method and device based on auto-encoder

A self-encoder and optical flow information technology, which is applied in the field of codec and information compression, to achieve the effect of optimizing storage space, reducing storage cost, and reducing storage space

Active Publication Date: 2020-06-05
SHANDONG INSPUR SCI RES INST CO LTD
View PDF8 Cites 4 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

However, this technical solution cannot use the self-encoder to realize the compression of optical flow information and improve the compression efficiency

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
  • Optical flow information compression method and device based on auto-encoder
  • Optical flow information compression method and device based on auto-encoder

Examples

Experimental program
Comparison scheme
Effect test

Embodiment 1

[0044] The optical flow information compression method based on an autoencoder of the present invention is to extract the optical flow features through a trained neural network model based on an autoencoder (Encoder), generate a feature map (Feature Map), and quantify (Quantize) Reduce the storage space of the data, and then use entropy coding to further encode and compress the quantized data; when decoding, the saved entropy coded data is entropy decoded and dequantized, and then passed through a decoder with the opposite structure (Decoder) Decoding, so that the feature map is restored to two-channel optical flow information; as attached figure 1 As shown, the details are as follows:

[0045] S1. Build a neural network model based on an autoencoder: set the number of convolutional layers required for encoding, the size of the convolutional kernel, the method of padding, and the number of strides; the design principle of the convolutional layer is usually the size of the conv...

Embodiment 2

[0055] The present invention mainly comprises encoder and decoder two parts, as attached figure 2 As shown, the specific workflow is as follows:

[0056] (1), the optical flow is input to the convolutional layer 1, convolutional layer 2 and convolutional layer 3 of the encoder for encoding to obtain a feature map;

[0057] Among them, the number of kernels (kernal) of the convolutional layer 1: 256, the size of the kernel number (kernal size): 9×9, the convolution step size (stride): 4, and the padding "SAME";

[0058] The number of kernels (kernal) of the convolutional layer 2: 192, the size of the kernel number (kernal size): 7×7, the convolution step size (stride): 2, the padding (padding) "SAME";

[0059] The number of kernels (kernal) of the convolutional layer 3: 128, the size of the kernel number (kernal size): 3×3, the convolution step size (stride): 2, padding (padding) "SAME";

[0060] (2), quantize and entropy encode the feature map;

[0061] (3), save the encod...

Embodiment 3

[0068] The optical flow information compression device based on the self-encoder of the present invention, the device includes,

[0069] The neural network model building unit is used to build the neural network model based on the autoencoder, and set the number of convolutional layers required for encoding, the size of the convolution kernel, the number of strides and the method of padding;

[0070] The training unit is used to use the training set to train the neural network model based on the autoencoder, set the label of each optical flow graph information to itself, construct the loss function through mse and bpp, and use the Adam optimizer to optimize. After the second iteration, the trained neural network model based on the autoencoder is obtained;

[0071] The feature map acquisition unit is used to encode the optical flow map information to obtain the feature map: input the optical flow map information to the Encoder part of the trained autoencoder-based neural networ...

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 an optical flow information compression method and device based on an auto-encoder, and belongs to the field of information compression and encoding-decoding. The technical problem to be solved by the invention is how to utilize an auto-encoder to realize compression of optical flow information and improve the compression efficiency. The method comprises the following steps: optical flow features are extracted through a trained neural network model based on the auto-encoder to generate a feature map, the storage space of data is reduced through quantization, and then entropy encoding is used for further encoding and compressing the quantized data; and during decoding, entropy decoding and inverse quantization are carried out on the stored entropy coded data, and then decoding is carried out through decoders with opposite structures to make the feature map recovered into optical flow information of two channels. The device comprises a neural network model building unit, a training unit, a feature map acquisition unit, a feature map quantization unit, an entropy coding unit, a serialized file storage unit, an entropy decoding unit and a decoding unit.

Description

technical field [0001] The invention relates to the fields of information compression and encoding and decoding, in particular to an optical flow information compression method and device based on an autoencoder. Background technique [0002] In the era of digital media, a large amount of image and video data is generated and stored from daily life, social networking, public security monitoring, industrial production and other fields, which requires a lot of storage space. At present, the mainstream video compression format h264 still has room for improvement in compression rate, and block-based motion estimation will also produce chromatic aberration. H265, which has not yet been popularized, is not favored due to its low compression efficiency. [0003] The optical flow method is an important method for moving image analysis at present. Its concept was first proposed by JamesJ.Gibson in the 1940s, and it refers to the speed of the pattern in the time-varying image. Becaus...

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): H04N19/124H04N19/91G06N3/04
CPCH04N19/124H04N19/91G06N3/045G06N3/088
Inventor 段强李锐金长新
Owner SHANDONG INSPUR SCI RES INST CO LTD
Who we serve
  • R&D Engineer
  • R&D Manager
  • IP Professional
Why Patsnap Eureka
  • Industry Leading Data Capabilities
  • Powerful AI technology
  • Patent DNA Extraction
Social media
Patsnap Eureka Blog
Learn More
PatSnap group products

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

© 2024 PatSnap. All rights reserved.Legal|Privacy policy|Modern Slavery Act Transparency Statement|Sitemap|About US| Contact US: help@patsnap.com