The Study Of Identity Information Retrieval And Application Based On Face Feature

The study of identity information retrieval and application based on face feature isdetermined the identity of the person through facial feature. It is an important part ofbiometric identification technology, and it also has many advantages that other biologicalcharacteristics of human body do not have. Now it is widely used in the field of criminalinvestigation, safety verification, and credit card recognition and so on. But face recognitionis generally implemented in the specific environment now, so it still faced with the influenceof various factors, especially the influence of illumination, noise, time and some other factors,which will seriously affect the recognition.The object of this paper is the identity of the information retrieval base on face feature,which the key of it is face recognition. The changes of natural conditions and other factorswill affect the face acquisition, so image fusion algorithm is proposed in this paper forillumination preprocessing and the method for median filtering of noise in imagepreprocessing. In order to reduce the dimension and better feature extraction for face image,the paper proposes the improved DCT transform combined with2DPCA on the base of theanalysis of face recognition methods.Firstly, the paper analyzes the influences to the face image coming from the factors ofillumination and noise, expounds several common methods for light processing now, andpresents the integration of histogram equalization and logarithm transform in the illuminationpreprocessing method. In the aspect of preprocessing for noise, compared with severalcommonly noise processing methods include Wiener filtering and median filtering, itproposes the improved method of adaptive median filter to process the face picture whichinfluenced by the noise.Then the simulation is carried out through the Yale face database,ORL face database and FERET face database. It is verify the effectiveness of the improvedillumination preprocessing algorithm through the contrast.Secondly, the paper analyzes the principal component analysis, linear discriminateanalysis and independent component analysis and it describes several simulation experimentson the face database to compare the performance of these methods that have been talkedbefore. And under such condition, it puts forward a face recognition method which iscombined the improved DCT transform method and2DPCA method. After DCT transform, itcan eliminate redundancy and high frequency noise, while preserving the illuminationinsensitive features, then though the improved image enhanced algorithm in DCT domain. Itcan enhance useful information and restrain the block effect. And complete the facerecognition through2DPCA in the end. It carries out the simulation experiment in the facedatabase to verify the effect of the algorithm to improve the face recognition rate. Finally, the paper expounds the application of the identity information retrieval based onfacial features in the criminal investigation detection. Then design the system which aims atthis application. The paper examines the feasibility of the simulation system through thecollected face images.