Image Super-resolution Reconstruction Method Based on Maximum Linear Block Neighborhood Embedding

A neighborhood embedding and super-resolution technology, applied in the field of image processing, can solve problems such as unsuitable for engineering applications, difficulty in registration of multiple fuzzy and noisy images, and blurred images

Active Publication Date: 2018-11-16
XIDIAN UNIV
View PDF4 Cites 0 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

[0010] 1. It does not have sufficient discrimination ability, which will lead to the problem of blurred images;
[0011] Second, it is often not easy to obtain a sufficient number of low-resolution images in practical applications, and the registration operation between multiple fuzzy and noisy images is very difficult, so it is not suitable for engineering applications in actual scenes;
[0012] 3. Factors such as inappropriate training samples or feature extraction will lead to obvious artificial traces and noise

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 Maximum Linear Block Neighborhood Embedding
  • Image Super-resolution Reconstruction Method Based on Maximum Linear Block Neighborhood Embedding
  • Image Super-resolution Reconstruction Method Based on Maximum Linear Block Neighborhood Embedding

Examples

Experimental program
Comparison scheme
Effect test

Embodiment 1

[0083] The present invention is innovating and researching in the field of super-resolution technology, specifically an image super-resolution reconstruction method based on maximum linear block neighborhood embedding, see figure 1 , the super-resolution reconstruction process of the present invention includes the following steps:

[0084] Step 1: According to the characteristics of the image to be reconstructed, construct a training sample set, extract mid- and high-frequency features from a set of training image blocks, and use this feature to perform hierarchical splitting clustering on the training set where the training samples are located. Clustering, the nonlinear manifold is approximately divided into multiple largest linear blocks, so that it is suitable for the use of linear embedding, so that the clustering results are more accurate. After clustering, according to the medium and high frequency characteristics, multiple largest linear blocks of different classes are c...

Embodiment 2

[0096] The image super-resolution reconstruction method based on the maximum linear block neighborhood embedding is the same as that in Embodiment 1. In step 1, the process of clustering and obtaining the maximum linear block is as follows:

[0097] (1a) When using the hierarchical splitting clustering method for clustering, the block with the largest degree of nonlinearity is selected from the training sample set X each time for splitting until all blocks reach the linearity of the set threshold, thus obtaining multiple clusters Class, called "maximum linear block", performs disjoint division on the training sample set X to obtain a set of disjoint maximum linear blocks X (i) , approximating multiple linear manifolds namely:

[0098]

[0099]

[0100]

[0101] where P is the number of local blocks, N i is the number of data points in the i-th local block, N is the total number of data points in all training sample sets X, and clustering is completed;

[0102] (1b) ...

Embodiment 3

[0125] The image super-resolution reconstruction method based on the maximum linear block neighborhood embedding is the same as that in Embodiment 1-2, and the process of classifying the low-resolution test image in step 2 is as follows:

[0126] (2a) Classify the input low-resolution image blocks, build a block histogram according to the pixel gradient information, and first calculate the gradient g(i, j) and angle θ(i, j) at a pixel point (i, j) j), where,

[0127] gradient:

[0128] Angle: θ(i,j)=arctan(g y (i,j) / g x (i,j)),0<θ(i,j)<2π;

[0129] (2b) For each test image patch, calculate its angle histogram h( ):

[0130] if g(i,j)>T g

[0131] where δ θ Represents the interval width in the histogram, in order to reduce the noise problem, use the threshold T g Used to limit the gradient amplitude;

[0132] (2c) Perform low-pass filtering on the obtained histogram to find the maximum value l 1 and the next largest value l 2 , judge whether the test image block i...

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 image super-resolution reconstruction method based on maximum linear block neighborhood embedding, the main steps of which include: constructing a training sample set, clustering by using a hierarchical splitting clustering method, and approximately dividing the nonlinear manifold into multiple The largest linear block, after clustering, uses medium and high frequency features to construct the largest linear block; classifies low-resolution test images into edge blocks and non-edge blocks, and uses two different neighborhood selection methods to reconstruct the results more accurately ; Neighborhood selection; Neighborhood embedding; Image reconstruction, deblurring the initial reconstructed image to obtain a complete and clear high-resolution reconstructed image. The present invention approximates a plurality of maximum linear block structures from the nonlinear manifold of training samples through a clustering method, combines feature representation and neighborhood selection to realize local linear neighborhood embedding, reconstructs more accurate high-frequency information, and greatly reduces The time complexity is reduced, and the super-resolution reconstruction of natural images can restore clearer edge details.

Description

[0001] technology neighborhood [0002] The invention belongs to the field of image processing technology, and further relates to an image super-resolution reconstruction method, specifically an image super-resolution reconstruction method based on maximum linear block neighborhood embedding, which can be used for super-resolution reconstruction of natural image data. It is easier to classify, identify and further apply the reconstructed images. Background technique [0003] With the rapid development of electronic information technology, digital image acquisition technology is widely used in many practical applications such as computer vision, remote sensing and medical imaging, video surveillance and so on. In various neighborhoods, with the development of technical equipment and people's needs, there are increasingly higher requirements for the resolution of digital images. However, due to certain limitations in the physical imaging system and imaging environment, high-reso...

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 Patents(China)
IPC IPC(8): G06T3/40G06K9/62
CPCG06T3/4053G06F18/231
Inventor 杨淑媛焦李成刘正康刘红英侯彪刘芳马文萍马晶晶缑水平曹向海张继仁
Owner XIDIAN UNIV
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