Very low resolution face recognition method and system based on semi-coupled decision dictionary learning

Abstract

The invention discloses an extremely low-resolution face recognition algorithm and system based on semi-coupled decision dictionary learning. The method includes the following two stages: a training stage and a testing stage; the training stage includes the following two steps: S1, constructing Initial high- and low-resolution expression dictionaries; S2. Update the initial high- and low-resolution expression dictionaries to obtain standard matrix sets; the testing phase includes the following three steps: S3. Obtain extremely low-resolution face images and obtain corresponding extreme Low-resolution partial expression coefficient matrix; S4, obtaining a new coefficient matrix set to be tested; S5, obtaining a recognition result according to the standard matrix set and the coefficient matrix set to be tested. The present invention improves the expression ability of extremely low-resolution human face images, improves the recognition rate of extremely low-resolution human face images, and has high recognition efficiency and fast recognition speed; Reconstruction also achieved good results.

Description

technical field [0001] The invention relates to the field of automatic face image recognition, in particular to an extremely low-resolution face recognition method and system based on semi-coupled decision dictionary learning. Background technique [0002] In recent years, face recognition technology has been widely used in many fields. However, in real life, since the monitoring equipment is often far away from the monitoring target, the quality of the target face in the picture is poor and difficult to identify; at the same time, changes in lighting conditions, expressions, and postures further increase the difficulty of face recognition , therefore, it is of great practical significance to study face recognition technology at very low resolution. [0003] At present, the traditional methods to solve the problem of face recognition at very low resolution can be roughly divided into two categories: the first category is to down-sample high-resolution images and match the l...

Images