Face super-resolution reconstruction method based on feature transformation based on nearest feature line
A recent feature line and feature conversion technology, applied in the field of image processing, can solve the problems of training sample library size limitation, inability to meet noise robustness, and unsatisfactory reconstruction effect
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Publication Date
- 2020-05-19
Smart Images

Figure 1 
Figure 2 
Figure 3
Abstract
Description
technical field
[0001] The invention relates to the technical field of image processing, in particular to a face super-resolution reconstruction method based on feature transformation based on nearest feature lines. Background technique
[0002] Face super-resolution reconstruction is a method based on the observed low-resolution face image, using high-resolution image training library samples and low-resolution face image training library samples, to reconstruct the low-resolution face image to be reconstructed The most similar high-resolution face image; it can reproduce the local details of the face, achieve the purpose of enhancing the accuracy of face recognition, help improve the detection rate of public security organs, and protect the lives and property of the people.
[0003] The face super-resolution algorithm based on local feature transformation introduces the idea of partial face into the feature transformation super-resolution method, which improves the image...
Examples
Embodiment Construction
[0051] The technical solution of the present invention can adopt the form of software to realize automatic process operation. The technical solution of the present invention will be further elaborated below in conjunction with the embodiments and accompanying drawings, as follows: figure 1 , figure 2 and image 3 As shown, a face super-resolution reconstruction method based on feature transformation based on the nearest feature line, specifically includes the following steps:
[0052] Step 1, input low-resolution face image x to be reconstructed, low-resolution image training sample set and high-resolution image training sample set N represents the number of training sample face patterns in the low-resolution image training sample set X and the high-resolution image training sample set Y.
[0053] In this embodiment, the FEI face database in the face super-resolution field is selected as the training sample library for the algorithm reconstruction experiment; the FEI fac...