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

Compressed sensing measurement matrix optimization method and system based on automatic encoder network

An automatic encoder and measurement matrix technology, applied in neural learning methods, biological neural network models, instruments, etc., can solve problems such as the single efficiency of measurement matrix, and achieve the effect of improving overall reconstruction quality, high image quality, and image quality improvement.

Active Publication Date: 2018-03-09
INST OF COMPUTING TECH CHINESE ACAD OF SCI
View PDF5 Cites 20 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

[0007] The purpose of the present invention is to solve the single and low efficiency problems of the above-mentioned existing measurement matrix, and propose a method for constructing a measurement matrix based on an autoencoder network

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
  • Compressed sensing measurement matrix optimization method and system based on automatic encoder network
  • Compressed sensing measurement matrix optimization method and system based on automatic encoder network
  • Compressed sensing measurement matrix optimization method and system based on automatic encoder network

Examples

Experimental program
Comparison scheme
Effect test

Embodiment Construction

[0035] In order to make the above-mentioned features and effects of the present invention more clear and understandable, the following specific examples are given together with the accompanying drawings for detailed description as follows.

[0036] The present invention makes a new design for the measurement matrix used in the compressed sensing sampling process, and uses the automatic encoder network training, wherein the pictures involved in the training data can come from any gallery, and this embodiment only uses images from Kulkarni K, Lohit S, Turaga P, et al.ReconNet:Non-Iterative Reconstruction of Images fromCompressively Sensed Measurements[J].2016:449-458The 91 pictures used are for the purpose of comparing with the existing technology to demonstrate the technical effect of the present invention .

[0037] The overall network training is divided into 2 steps:

[0038] 1) The autoencoder network implements sampling and preliminary reconstruction, starting from figur...

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 relates to a compressed sensing measurement matrix optimization method and system based on automatic encoder network. The method comprises the steps of obtaining original images as training data and dividing the training data into a plurality of image blocks through segment cutting processing; conducting sampling on the image blocks based on the preset sampling rate and the automaticencoder network, and generating initial reconstruction images; calculating the residual error value between the initial reconstruction images and the original images based on the deep residual errornetwork; combining the residual error value with the initial reconstruction image and generating a reconstruction result, and establishing a loss function based on the reconstruction images and the image blocks, and conducting training on the parameter in the automatic encoder network through the loss function, and using the automatic encoder network parameter after training completion as the compressed sensing measurement matrix. The invention is advantageous in that through the transformation of data dimensions via the automatic encoder, the process from collecting to reconstructing of images can be simulated and realized, wherein the parameter in the collecting process is the measurement matrix, which has good reconstruction quality.

Description

technical field [0001] The invention relates to the field of measurement matrix design used in compressed sensing sampling, in particular to a method and system for optimizing a compressed sensing measurement matrix based on an automatic encoder network. Background technique [0002] Compressed sensing theory, as a relatively new sampling theory at present, can reconstruct and restore the original signal from much less sampled data than the traditional sampling theory, which reduces the difficulty of traditional signal acquisition. [0003] According to the compressed sensing theory: y=Φx, in the sampling process of compressed sensing, the signal needs to be randomly projected on the measurement matrix Φ to obtain the measured value y, and in order to make the measured value contain as much information as possible of the original signal, It is necessary for the measurement matrix to meet certain conditions, that is, the construction of the measurement matrix. Current researc...

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
Patent Type & Authority Applications(China)
IPC IPC(8): G06T11/00G06T5/50G06N3/08
CPCG06N3/084G06T5/50G06T11/00G06T2207/20021G06T2207/20221
Inventor 代锋马宜科赵强张勇东李宏亮田蔚
Owner INST OF COMPUTING TECH CHINESE ACAD OF SCI
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