基于融合分层视觉感知的人脸局部特征识别

Face Local Feature Recognition Based on Fusion Hierarchical Visual Perception

由于已有方法未能在人脸局部特征识别过程中加入人脸图像去噪,导致识别结果不理想,测试时间增加。提出一种融合分层视觉特征感知的人脸局部特征识别方法,通过小波包降噪方法对人脸图像进行去噪。采用Adaboost算法对人脸五官进行检测,利用SIFT算法对人脸五官旋转不变关键点进行特征提取。同时将SITF算法和LBP算法进行融合,对人脸局部特征进行提取,同时建立分层视觉特征感知模型获取人脸局部信息表述,最终实现人脸局部特征识别。仿真结果表明,所提方法能够全面增加人脸局部特征识别率,减少测试时间,获取比较满意的识别结果。

The existing methods fail to introduce the face image denoising in the process of recognizing local face features,so the recognition result is not ideal and the test time increases.Therefore,a method to recognize the local face features combing with hierarchical visual perception was presented.At first,the wavelet packet denoising method was used to remove the noise from face images.Then,the Adaboost algorithm was adopted to detect facial features.Meanwhile,SIFT algorithm was used to extract the features of rotation-invariant points of facial features.Moreover,the SITF algorithm was combined with the LBP algorithm to extract local face features.Furthermore,a hierarchical visual perception model was built to obtain the expression of local information.Finally,the recognition of local face features was achieved.Simulation results show that the proposed method can comprehensively increase the recognition rate of local face features,reduce the test time,and obtain a satisfactory recognition effect.