SAMME.RCW algorithm based face recognition optimization method

A face recognition and optimization method technology, applied in character and pattern recognition, computing, computer components, etc., can solve the problems of low recognition rate, improve quality, solve the problem of resampling, and improve the effect of classification accuracy

CN105975902AActive Publication Date: 2016-09-28BEIJING UNIV OF TECH
4 Cites 2 Cited by

Patent Information

Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Publication Date
2016-09-28

Smart Images

  • Figure 1
    Figure 1
  • Figure 2
    Figure 2
  • Figure 3
    Figure 3
Patent Text Reader

Abstract

The invention relates to a SAMME.RCW algorithm based face recognition optimization method, which comprises the steps of firstly carrying out feature extraction on a face image, and carrying out recognition classification by using an image feature vector according to a SAMME.RCW algorithm. Modification is carried out on a weight adjustment process of the SAMME.RCW algorithm, thereby ensuring the weight of every class of samples not to be too small when re-sampling occurs, also enabling weight adjustment after re-sampling to be more partial to minority-class samples, and ensuring classification effects of the samples. A requirement of the SAMME.RCW algorithm for the performance of a weak classifier is that the weight of correctly classified samples in each class is greater than the weight of any other class of samples, and a requirement for the accuracy is performed on each class independently. Through modification carried out on weight allocation in re-sampling, the probability of being selected of each class of samples is ensured to be basically the same, and classification effects of the minority-class samples and majority-class samples in the weak classifier are ensured at the same time. The accuracy of face recognition is effectively improved by a finally acquired strong classifier.
Need to check novelty before this filing date? Find Prior Art

Description

technical field

[0001] The invention belongs to the technical field of machine learning and pattern recognition, and integrates training data to construct a prediction method with strong generalization ability, so as to give accurate estimates to new unknown objects. Background technique

[0002] Face recognition technology is an important technology in image processing, and it is an active research field in biometric recognition. Using computer vision and image processing technology, using the contour features and local detail features of the face to perform face recognition. At present, it has been applied in identity authentication and authority control. However, the low recognition rate is an important reason that has hindered the widespread application of face recognition technology. The research found that the accuracy rate can be improved through the method of integrated learning. Ensemble learning is a new machine learning paradigm that uses multiple base classifi...

Examples

Embodiment Construction

[0037] Provide the explanation of each detailed problem involved in the technical scheme of this invention below in detail:

[0038] The SAMME algorithm requires the correct rate of the weak classifier to be greater than 1 / k. The SAMME.R algorithm, on the basis of the SAMME algorithm, also requires that the weight of the correctly classified samples in each category be greater than the weight of any sample assigned to other classes. In order to ensure that in each weak classifier, the correctly classified samples account for the majority. From a vertical perspective, according to the theorem of large numbers, it ensures that after multiple iterations, the accuracy rate of the final integrated strong classifier is improved.

[0039] The SAMME.R algorithm restricts the weak classifiers obtained each time to ensure that the weights of correctly classified samples in each class are greater than the weights of any samples assigned to other classes. If this condition is met, continu...