Iris Recognition Algorithm Using DE_BBO_ELM And Multi-granularity Features

Iris recognition technology,a new kind of biological characteristic recognition technology,is one of the hottest research spot recently.Iris is especially suitable for recognition because of its special properties: the texture of iris is unique for different eye,the characteristic is stable,iris texture is difficult to fake and the acquisition of iris image is not invasive,so iris recognition technology is considered as the most popular technology in the future and is widely used in various fields,such as electronic commerce,finance,security and military etc.This paper summarizes the research status of iris recognition technology both at home and abroad.A multi-granularity feature extraction method is proposed to solve the problem that the features extracted by the existing feature extraction method are not comprehensive.In order to improve the overall performance of iris recognition,DE_BBO_ELM based on DE_BBO and the extreme learning machine(ELM)is proposed,and DE_BBO is used to optimize the ELM.This dissertation of this paper subdivided into three major sections is organized as follows:(1)As the Gabor filter can capture the low and intermediate frequency texture information accurately,and the GLCM has a better ability to obtain the high frequency texture information.In order to avoid the defect of single feature extraction method,a multi-granularity extraction method based on 2D-Gabor filters and GLCM is proposed to generate a multi-feature vector.(2)DE_BBO_ELM,A optimization algorithm based on DE_BBO and ELM is proposed.Firstly,In order to enhance the variability and the traversal ability within the feasible domain of the biogeography-based optimization(BBO),DE_BBO based on the differential evolution(DE)and BBO is proposed by using the DE algorithm to replace the variation operation of the BBO algorithm;then,DE_BBO is used to train ELM to improve the convergence performance of ELM.DE_BBO_ELM is used as the iris classifier.Compared to the support vector machine(SVM)and ELM,the experimental results show that DE_BBO_ELM has better recognition performance.(3)Iris recognition algorithm using DE_BBO_ELM and multi-granularity,A new iris recognition algorithm is established.Firstly,two iris libraries are selected from CASIA-Iris,and 60 iris images with good quality are selected as the experimental sample images in each iris library,after this,all the sample images are preprocessed;secondly,the multigranularity extraction method proposed in this paper is used to extract the features of these sample images to compose the learning and training samples;then,the iris classifier,DE_BBO_ELM based on the DE_BBO and ELM is proposed by using the DE_BBO algorithm to improve ELM;finally,the training samples are classified and identified by DE_BBO_ELM,and the performance parameters such as recognition accuracy,equal error rate and recognition time are obtained.The experimental results show that the iris recognition algorithm proposed in this paper outperforms other mainstream iris recognition algorithms.