Unlock instant, AI-driven research and patent intelligence for your innovation.

Face image super-resolution reconstruction method based on non-linear compressed sensing

A non-linear compression, face image technology, applied in the field of image processing, can solve the problems of not considering the difference of the details of the face, the quality of the reconstructed face image is general, and the image reconstruction time is long, so as to reduce the amount of calculation and avoid the The effect of loop iteration process and algorithm complexity reduction

Active Publication Date: 2015-09-30
XIDIAN UNIV
View PDF2 Cites 8 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

The above methods do not consider the differences in face details, and the reconstruction effect is poor
At the same time, these methods require a large number of training samples to ensure the effect and quality of the reconstruction, the amount of calculation is huge, the image reconstruction time is long, resulting in low efficiency, and under the condition of a high amplification factor, the quality of the reconstructed face image generally

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
  • Face image super-resolution reconstruction method based on non-linear compressed sensing
  • Face image super-resolution reconstruction method based on non-linear compressed sensing
  • Face image super-resolution reconstruction method based on non-linear compressed sensing

Examples

Experimental program
Comparison scheme
Effect test

Embodiment 1

[0029] The present invention is a face image super-resolution reconstruction method based on nonlinear compressed sensing, see figure 1 , the present invention comprises the following steps to face image super-resolution reconstruction:

[0030] Step 1: Input the training face image pair, and use the low-resolution training face image to learn to construct a low-resolution image block dictionary with a size of N And use high-resolution training face images to learn to construct a dictionary with a size of N and low-resolution image blocks The corresponding high-resolution image patch dictionary The input training face image pair, see figure 2 , figure 2 There are 5 rows of images in , and each row of images has six face images, and various face images are included in these 30 face images. In this example, some face images are shown, some are frontal photos, some are profile photos; some are looking up, some are looking down; some are white, some are black, and some ar...

Embodiment 2

[0042] The face image super-resolution reconstruction method based on nonlinear compressed sensing is the same as that in Embodiment 1.

[0043] The process of calculating the sparse matrix of the test image block described in step 5 includes the following steps:

[0044] 5a) Low-resolution test image patch The corresponding high-resolution test image block It is nonlinear K-sparse in the nonlinear space Ω, using a set of sparse basis in the nonlinear space Ω Perform sparse representation, namely: Φ ( y t i ) = V i β i = Σ k = 1 K β i k v i k , in v ...

Embodiment 3

[0050] The face image super-resolution reconstruction method based on nonlinear compressed sensing is the same as that in Embodiment 1-2.

[0051] The process of reconstructing the high-resolution image block described in step 6 includes the following steps:

[0052] 6a) The non-linear sampling of the original high-resolution test image is expressed as

[0053] 6b) According to the principle of Pre-image, any signal is represented by a set of sparse orthogonal bases in space, and the expressed coefficients are the inner product of the signal and the sparse orthogonal base. For high-resolution test image blocks Do the refactoring: y t i = Σ l = 1 n k - 1 ( Σ j = 1 ...

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 face image super-resolution reconstruction method based on non-linear compressed sensing and mainly solves the problems that the operating speed of a conventional method is low and an image reconstructed under higher amplification factors is comparatively blurred. The method comprises main steps as follows: preprocessing a group of training face images, and constructing a pair of low-resolution image block and high-resolution image block dictionaries; inputting a low-resolution face image as a test image and blocking the image; searching m adjacent blocks of the input face image blocks to obtain corresponding high-resolution adjacent blocks, and performing kernel principle component analysis training to obtain a sparse coefficient; constructing a non-linear compressed sensing super-resolution reconstruction model, calculating a sparse matrix with a least-squares method, and obtaining a reconstructed high-resolution image with a Pre-image method. The face image super-resolution reconstruction method based on the non-linear compressed sensing is low in complexity and short in operating time, can effectively improve the reconstructed image quality under the condition of high amplification factors and is suitable for super-resolution reconstruction of various face images, and the image reconstruction efficiency and quality are high.

Description

technical field [0001] The invention belongs to the technical field of image processing, and mainly relates to image super-resolution reconstruction, in particular to a face image super-resolution reconstruction method based on nonlinear compressed sensing, which can be used for super-resolution reconstruction of various face images. Background technique [0002] Face image processing has always been one of the research hotspots in the fields of pattern recognition, computer vision and multimedia information processing. Face image super-resolution reconstruction has important practical application value, especially in face recognition, video surveillance, security department and other fields. Existing facial image super-resolution reconstruction techniques can be divided into two categories: reconstruction-based methods and learning-based methods. The reconstruction-based algorithm establishes an imaging model for the image acquisition process, and restores high-resolution ...

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): G06T5/50
CPCG06T5/50G06T2207/20081G06T2207/20212
Inventor 杨淑媛焦李成张继仁熊涛刘红英马晶晶缑水平刘芳侯彪刘正康崔顺
Owner XIDIAN UNIV