This paper presented a very efficient method for face recognition. The previous work, linear regression classification (LRC), is a fairly simple method (which is actually a nearest subspace method finding the minimum distance from query to every class-specific subspace), but obtains very high recognition accuracy on various facial databases. The proposed methods in this paper are still classifiers based on linear regression. To the best of our knowledge, this paper first gives the detailed analysis of the effectiveness of NS classifier.
Competing interests declared:
Furthermore, the experimental results showed that the proposed classifier is superior to LRC classifier. Figure 1 and 2 are very good to illustrate the idea.
