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

Image super-resolution reconstruction method based on learning

A technology of super-resolution reconstruction and image reconstruction, applied in the field of image and video, can solve the problems of inability to realize real-time super-resolution reconstruction, low reconstruction efficiency, etc., to reduce the number of dictionary atoms, low cost, and improve computing. The effect of efficiency

Inactive Publication Date: 2017-09-05
XIAN UNIV OF SCI & TECH
View PDF3 Cites 4 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

Although the super-resolution reconstruction algorithm based on learning is effective, but due to the need to train a large amount of prior information, the reconstruction efficiency is relatively low, and the purpose of real-time super-resolution reconstruction cannot be achieved.

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
  • Image super-resolution reconstruction method based on learning
  • Image super-resolution reconstruction method based on learning
  • Image super-resolution reconstruction method based on learning

Examples

Experimental program
Comparison scheme
Effect test

Embodiment 1

[0045] See figure 1 , figure 1 This is a schematic diagram of a learning-based image super-resolution reconstruction method provided by an embodiment of the present invention. The method includes the following steps:

[0046] Step 1. According to the degraded model blur and N times downsampling, process the first resolution image to form the second resolution image;

[0047] Step 2. Train the first resolution image and the second resolution image to form an image learning dictionary;

[0048] Step 3. Perform block division on the image to be reconstructed to form multiple image blocks to be reconstructed;

[0049] Step 4. Using an edge extraction algorithm, determine a fixed threshold of image information according to the second resolution image;

[0050] Step 5. If the information amount of the image block to be reconstructed is greater than the fixed threshold, perform image reconstruction on the image block to be reconstructed according to the image learning dictionary to form a fir...

Embodiment 2

[0080] This embodiment describes the technical solution of the present invention in detail on the basis of the foregoing embodiment. Specifically, the method includes:

[0081] Step 1: Using a large number of samples of high-resolution images (ie, the first resolution image), the high-resolution images are subjected to blur processing and N-fold down-sampling according to the modified degradation model to obtain the corresponding low-resolution images (ie, the first resolution image). Two-resolution image) sample.

[0082] Step 2: For the low-resolution image obtained in step 1, the image features are extracted through the feature extraction algorithm to obtain the high-resolution feature information X of the space target s (I.e. first resolution feature information) and low resolution feature information Y s (Ie the second resolution feature information).

[0083] Step 3: Use the K-SVD algorithm to jointly train the feature information to obtain a high-resolution image learning dic...

Embodiment approach

[0101] S1: Perform block segmentation on a large number of low-resolution image samples to obtain image blocks;

[0102] S2: Use an edge extraction algorithm to extract edge information of low-resolution image blocks, and count the information volume of each low-resolution image block and the distribution of information volume of all low-resolution image blocks;

[0103] S3: Select the value with the highest amount of information in the low-resolution image block, obtain the pixel value of the low-resolution image block as F1, f=F1 / 4, then f*40% <= threshold <=f*60%, take several representative thresholds in this range, for example, you can take the following representative thresholds: f*40%, f*45%, f*50%, f*55%, f*60%, calculate each The learning-based sparse representation of each threshold point corresponds to the reconstruction time of the image super-resolution reconstruction algorithm and the resolution of the reconstructed image, which can be judged according to subjective ev...

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 an image super-resolution reconstruction method based on learning. A first resolution image and a second resolution image are trained so as to form an image learning dictionary; block division is performed on an image to be reconstructed so as to form multiple image blocks to be reconstructed; an image information content fixed threshold is determined according to the second resolution image; if the information content of the image blocks to be reconstructed is greater than the fixed threshold, image reconstruction is performed according to the image learning dictionary so as to form a first reconstruction subarea; or image reconstruction is performed according to a double cubic interpolation algorithm so as to form a second reconstruction subarea; and imaging splicing is performed on the first reconstruction subarea and the second reconstruction subarea so as to obtain a super-resolution reconstruction image. According to the image super-resolution reconstruction algorithm based on learning, real-time image resolution reconstruction is realized, and data dimension reduction is performed in the reconstruction process by using the KPCA method; and the imaging system hardware structure does not need to be changed so that the method has the advantages of low cost and high economic benefit.

Description

Technical field [0001] The invention relates to the field of image and video, in particular to a learning-based image super-resolution reconstruction method. Background technique [0002] Image super resolution reconstruction (super resolution, SR) refers to the use of a computer to process a low resolution image (LR) or image sequence to restore a high resolution image (HR) Image processing technology. HR means that the image has a high pixel density and can provide more details, which often play a key role in the application. [0003] With the development of computer technology, people’s demand for high-resolution images in the military, satellite remote sensing imaging, and medical fields is becoming more and more urgent. However, traditional equipment is subject to system blur, atmospheric motion, noise, and imaging during the imaging process. Various factors, such as the environment, result in low resolution of acquired images, which makes it difficult to meet specific needs...

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): G06T3/40G06T5/50
CPCG06T3/4053G06T5/50G06T2207/20021G06T2207/20024
Inventor 周筱媛
Owner XIAN UNIV OF SCI & TECH
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