Foley-Sammon face recognition method fusing matrixes and vector feature extraction

Abstract

The invention discloses a Foley-Sammon face recognition method fusing matrixes and vector feature extraction. According to the method, first, a two-dimensional Foley-Sammon identification vector matrix in the horizontal direction of a face image matrix is calculated, a two-dimensional linear identification vector matrix in the vertical direction of the face image matrix is calculated, and then the two identification vector matrixes are utilized to realize bidirectional compression of face images; each compressed face image is pulled into vectors according to rows and columns, and then the vectors of each compressed face image are fused into one piece of vector data; identification information of the vector data obtained after fusion is extracted; and last, a nearest neighbor classifier is used to perform classification processing on a result to realize accurate recognition of the face images. Through the method, the Foley-Sammon method is adopted in the horizontal direction and the two-dimensional linear identification method is adopted in the vertical direction to extract the identification information, complementary identification feature information of the face images can be obtained, main feature information of the face images is reserved as much as possible, extraction of the identification information of the face images is more sufficient, and therefore the recognition rate of the face images is increased.

Description

technical field [0001] The invention relates to the fields of pattern recognition and artificial intelligence, in particular to a Foley-Sammon face recognition method which combines matrix and vector feature extraction. Background technique [0002] Face recognition is very active in the field of computer pattern recognition research and has been applied in the commercial and security fields. From static matching of photos in controlled formats (e.g. passports, credit cards, driver’s licenses) and mugshots to real-time matching of surveillance video images. There are three main problems in face recognition technology: image segmentation, feature extraction and recognition. Among them, extracting the features of face images is the key to complete the task of face recognition. [0003] At present, there are mainly vector-based feature extraction methods for extracting the features of face images, such as: principal component analysis, linear discriminant analysis method, Fol...

AI Technical Summary

Problems solved by technology

The disadvantage of the vector-based feature extraction method is that the image matrix must first be converted into a high-dimensional vector, resulting in a large amount of time-consuming face recognition processing and small sample problems.

The vector is generally a high-dimensional vector. When processing these high-dimensional vectors with the Foley-Sammon feature extraction method, the amount of calculation is large, and there is still a small sample problem.

Images