A Two-Stage Recognition Method Based on Non-Negative Representation Coefficients

Abstract

The invention proposes a two-stage recognition method based on non-negative representation coefficients. Including: use all training samples to linearly represent the test samples and the constraints indicate that the coefficients are non-negative, and calculate the coefficient vector; arrange the elements in the coefficient vector in descending order, extract larger coefficients and their corresponding training samples; classify the training samples obtained in the previous step , calculate the reconstructed image of each type of sample; use all reconstructed images to linearly represent the test sample and the constraint indicates that the coefficient is non-negative, calculate the coefficient vector, and classify the test sample according to the residual. The invention selects the training samples according to certain screening conditions and uses the training samples to linearly represent the test samples, thereby improving the recognition rate.

Description

technical field [0001] The invention belongs to the technical field of machine learning, and in particular relates to a two-stage recognition method based on non-negative representation coefficients. Background technique [0002] Feature extraction is very important in the field of pattern recognition. Principal Component Analysis (PCA) and Linear Discriminant Analysis (LDA) are widely used as classic linear feature extraction methods. Both PCA and LDA methods ultimately need to rely on a classifier to classify the test samples. Among them, the more widely used classifiers such as the nearest neighbor (Nearest Neighbor, NN), the purpose of this classifier is to classify the test samples to the training samples closest to it. category. [0003] The newly proposed sparse representation is a new method in the field of face recognition. The basic idea is that given enough training samples, any test sample can be represented by a linear combination of training samples, and the ...

AI Technical Summary

Problems solved by technology

The sparse representation classification (Sparse representation classification, SRC) method proposed by John Wright et al. has been widely used, but because this method needs to calculate l 1 norm, so the calculation process is complex and time-consuming

Based on this, scholars such as Lei Zhang proposed a collaborative representation based classification (Collaborative representation based classification, CRC) method, which overcomes the shortcomings of the slow calculation speed of the SRC method. However, from the perspective of non-negative matrix decomposition, in CRC Negative values ​​do not have any physical meaning, these are the shortcoming and deficiency of algorithms such as CRC, and the method of the present invention hopes to solve this shortcoming

Images