Method and system for example-based face hallucination
First Claim
1. A method for example-based face hallucination on a face hallucination system, comprising the steps of:
- preparing a training database with a plurality of training images and inputting a low resolution (LR) face image to be hallucinated;
determining a projection matrix for projecting said plurality of training images of said training database onto a manifold domain for clearly differentiating projected images of said plurality of training images using a manifold learning technique;
projecting said plurality of training images of said training database and said LR face image onto said manifold domain using said projection matrix, where yL representing a projected image of said LR face image and ytrain representing projected training images of said training;
images of said training database;
selecting a training set best matching yL from N projected training images ytrain, where N is not greater than a total number of said plurality of training images;
learning a set of basis images by using basis decomposition on said training set and yL, said set of basis images including a set of high resolution prototype faces of in said training set and a low resolution (LR) prototype face of yL; and
reconstructing a high resolution face image of said LR face image by using said basis images.
1 Assignment
0 Petitions
Accused Products
Abstract
A method for example-based face hallucination uses manifold learning to project a plurality of training images in a training database and an input low resolution (LR) face image into a same manifold domain, then iteratively refines the reconstruction basis by selecting a training set having k projected training images which best match the parts of the projected LR face image, where k≦N and N is the number of projected training images. Through the best-match training set, a set of prototype faces are learned, and the set of prototype faces are used as the reconstruction basis to reconstruct a high resolution face image for the input LR face image.
11 Citations
13 Claims
-
1. A method for example-based face hallucination on a face hallucination system, comprising the steps of:
-
preparing a training database with a plurality of training images and inputting a low resolution (LR) face image to be hallucinated; determining a projection matrix for projecting said plurality of training images of said training database onto a manifold domain for clearly differentiating projected images of said plurality of training images using a manifold learning technique; projecting said plurality of training images of said training database and said LR face image onto said manifold domain using said projection matrix, where yL representing a projected image of said LR face image and ytrain representing projected training images of said training;
images of said training database;selecting a training set best matching yL from N projected training images ytrain, where N is not greater than a total number of said plurality of training images; learning a set of basis images by using basis decomposition on said training set and yL, said set of basis images including a set of high resolution prototype faces of in said training set and a low resolution (LR) prototype face of yL; and reconstructing a high resolution face image of said LR face image by using said basis images. - View Dependent Claims (2, 3, 4, 5, 6, 7)
-
-
8. A system for example-based face hallucination, comprising:
-
a training database, for collecting and storing a plurality of training images; a projection module, for receiving said plurality of training images, using a manifold learning technique to obtain a projection matrix, and projecting said plurality of training images and an input low resolution (LR) face image onto a manifold domain to obtain N projected training images ytrain and a projected LR face image yL, said projection matrix being determined by said manifold learning technique for clearly differentiating said N projected training images ttrain; a matching module, for selecting a training set of k face images best matching yL from said N projected training images ytrain, k≦
N;a basis decomposition module, for using basis decomposition on said training set and yL to learn an LR prototype face of yL and a set of high resolution prototype faces of said training set so that a difference between said LR prototype face of yL and said set of high resolution prototype faces matches a threshold requirement; and a face hallucination module, for reconstructing a high resolution face image for said LR face image by using said LR prototype face of yL and said set of high resolution prototype faces as a set of basis image. - View Dependent Claims (9, 10, 11, 12, 13)
-
Specification