Efficient Face Recognition Research Based on Random Projection Dimension Reduction

With the rapid development of economy, nowadays the accuracy requirements for the image of face recognition are becoming increasingly higher. In this paper, the random projection in dimension reduction is realized by analyzing the method of dimension reduction and improving the limitations of high-dimensional data. This paper takes 100 images of 50 individuals in the face recognition database of University of Essex as the sample of the experiment and reaches the following conclusions: there are advantages in accuracy and stability for random projection with low computing complexity, independent data and constant distance.