Research On Face Recognition Complex Illumination

With the rapid development of computer vision,face recognition technology has been widely applied in various fields.However,in the actual scene,the interference of various external factors greatly reduces face recognition rate,and the complex illumination condition is one of the key factors affecting the performance of the face recognition system.In this dissertation,face recognition under complex illumination conditions is the main research object,focusing on the in-depth study of normalization of illumination algorithm and illumination invariant feature extraction.Low-illumination image enhancement algorithm based on RetinexNet in HSV space and the face recognition algorithm based on the feature fusion of Center Symmetric Local Binary Pattern and Histogram of Oriented Gradient operator are proposed.The main work of this paper is as follows:Aiming at the problem of how to make an illumination compensation for the uneven illumination,an improved algorithm based on RetinexNet algorithm is proposed.First of all,based on the visual characteristics,the image is converted into HSV color space,and the value component is enhanced by the relatively independent characteristics of each channel.At the same time,the saturation component is adjusted adaptively along with the value component.Finally,the edge details of the image are enhanced by the sharpening algorithm to improve the feature expression of the image.Finally,the experimental results show that the improved algorithm could effectively improve the image quality.In the complex illumination conditions,the face recognition algorithm based on the feature fusion of CSLBP operator and HOG one is proposed.More effective facial features are obtained by fusing gradient histogram extracted by HOG and the texture feature by CSLBP.The fused features combining the local feature with the global feature of the face,enhance the description ability of the face image,and effectively improve the face recognition rate under complex lighting conditions.Finally,theexperimental results verify the effectiveness and robustness of the proposed feature fusion algorithm.Finally,a multi-feature face recognition algorithm based on the improved Retinexnet algorithm is designed.Firstly,the algorithm of illumination enhancement is used for preprocessing,then the CSLBP-HOG operator is used for feature extraction,and finally use the nearest neighbor classification for classification and recognition.Finally,the experimental results verify the effectiveness of the proposed algorithm.