Integration Of The Human Ear And The Face Of Multi-modal Identification

Information fusion is through the use of computer technology from the sensor observations of information in a certain standards to achieve the comprehensive analysis, decision making and tasks of the task of a kind of information processing. Information fusion center integrated and comprehensive problem is how the information from different sources. According to the level of information processing and information fusion can be divided into data fusion, characteristics and decision-level fusion. In the tertian, data fusion of fusion is mainly to the original observation data fusion, thus keeping the integrity of data, but the system requirements, the data is very high capacity. Decision-level fusion processing speed faster, but some relative information, information loss compression is getting higher. The fusion feature information leakage and processing speed between data fusion and decision-making level fusion, so has greater flexibility. This article mainly from the face and ears more modal identification of fusion technique characteristics of exploration and research.The identification process of feature core is extracted. Biometric technology for extracting features, the key is to seek effective feature extraction method. In the traditional feature extraction method, the fusion of general linear characteristics is operated, while ignoring the influence of non-linear characteristics, and because of the existence of information between inevitable in the capacity of data, virtually increased a lot. In the related field research foundation, this paper proposes a method based on improved nuclear fusion of independent component analysis method for extracting features of order statistics, from high characteristic, using the improved selection algorithm of network centric RBF neural network to classify characteristics, because this algorithm in overcoming nonlinear effects and sunshine has good robustness, experimental analysis, get more satisfactory effect.On this basis, in order to further enhance the recognition, this paper further and proposed based on independent subspace of independent component analysis method. So-called independent component analysis method is put all the ingredients into a number of independent space within the group, groups of interdependence, mutual independence, then respectively tc projection vector and fusion of subspace on feature extraction. This independent subspace o: independent component analysis method, reduced because algorithm between independent component of defects, further to overcome the recognition and illumination change. improve the effect of the generality of the products, expand the scope of application of the algorithm.This paper considers synthetically the ear and face image fusion after the superpositior and image of the information from the main illumination change posture, the fusion feature extraction technique research and exploration, and achieved good results, the rate of singler biometric identification had the very big enhancement.