关键词: 人脸识别, Gabor分析, 信号归类, 加权匹配

Abstract: To improve the Gabor ability to retain face structure and content information，and solve the problem of low anti-noise property in human face recognition，this paper proposes a new face recognition algorithm based on weighted analysis and modified Gabor filter.This method analyzes normalized face in multiscale by Gabor filter，and classifies Gabor signal under the principle of uniform filter parameter.Owing to this，classified Gabor signal can be a decision threshold by weighted match.Experiments show that this method not only robust in retaining face structure and content information，but also has strong anti-noise ability in face recognition.

Key words: face recognition, Gabor analysis, signal classification, weighted match