| The iris recognition is a high and new technology, and it has been used in many application areas. As an application-oriented research project, iris recognition integrates mathematic, computer science, optics, electronics and physiology etc. As the development of Information Technology, iris recognition is gradually to be the research focus in home and abroad, because of its excellent characters and wind application future. A unified standard should be established due to the wider and wider used scope. It is useful to accelerate the maturation of the iris technology in order to apply the recognition into practice more widely.The work to establish a standard for the iris recognition has attached more and more attentions abroad, but this work has little attention in home. So the author does some research about this aspect, and the results are the follows:(1) By using the Object-Oriented Analysis (OOA) method, the author analyzes customer demand in details and set up the function model of the system. A logical framework will be presented in the paper. And the framework is designed into five classes: user layer, control layer, process layer, feature database layer and hardware device manage layer. And the interfaces between each layer have been designed according to the layer class structure of the iris recognition system. These are system interface, iris technique processing interface, feature data interface and hardware device manage interface. At the same time, the detail functions of each logic layer and the communication ways have been defined.(2) Each layer of the framework is also a lay-structure and has been divided into several modules in order to keep the consistency of the total system. And the function modules are called individually as follows: hardware device managing, image evaluating, iris locating, iris normalizing, iris featurerepresenting, iris feature coding, feature data safe coding, iris database managing, iris mode matching. The functions of each module, the standard interfaces and the relationships of each layer are defined.(3) Analyzing the requirement of the layers in the system in detail, the author standardizes the interfaces from two aspects: data and interface. The contents of data aspect are standardized feature data model and interchange data model. And the other aspect defines the universal communication method between the application program and iris recognition system. And the method consists of several functions with data structure and other referred information.(4) The object model (OM) of the iris recognition system is set up according to the logic framework structure. The model describes the static data structure of the system. And then the concrete formats of each interface in every layer are described by class diagram, and these are based on the object model referred above and the demand of interface criterion. So far, the design of the framework of iris recognition system has been done.(5) The design of algorithms about the iris recognition system should follow a certain design rule in order to load dynamically and displace the algorithms expediently. The author makes use of dynamic link libraries (DLL) technical. The design of every function model has been realized using DLL. And each function model is allowed to output only a structure object, which records the basic information of the function model such as the edition number, name etc. Naming to a model should embody the function of the model as far as possible. The interface of each model is public functions. The data structure of each model is the attribute defined in the class.(6) The algorithms of every layer of the iris recognition system have been realized. This system is build up according to the logical framework presented by the author in the paper. The author studies, realizes and embeds these algorithms into the platform. And the experiment process on the platform completes logically the iris recognition functions linearly. And explanation isas follows:1) In hardware device manage layer, the author encapsulates and manages the patent product for capturing iris image of the laboratory;2) In process layer, six function models have been realized in the paper. And these models are as follows: image evaluating, iris locating, iris normalizing, iris feature representing, iris feature coding and feature data safe coding. Expatiate in them as follows:In image evaluating model, the author realizes and encapsulates on the quality evaluation of the sequence iris image.In iris locating model, the author completes the experiment for the iris localization. It is called the iris localization method based on Preconditioning Voting and seeking exactitude.In iris normalizing model, three function models have been carried out. They are the iris texture direction revise, the iris texture normalize, the iris texture brightness revise. These algorithms are as follows: A shift invariance preprocessing method based on directional energy distribution sequences is realized to correct the angular rotation of different iris textures, which corresponds to the horizontal shift of normalized iris blocks. The method of the iris texture standardization is as follow: To divide the effectual area of the iris under polar coordinates as the virtual circles, these circles are transformed into the radiate annular district. Bilinear interpolation algorithm is used to map these reference points to image in Descartes reference frame. The area of the image mapped is m×n. In this paper the area is 64×128. The iris texture brightness revise make use of the linearity ray glint model and Histogram Local Equalization approach.In the iris feature represent model, we realize and embed GHM multi-wavelet iris texture multi-scale analysis and representation algorithm.In the iris feature code model, we select some effectively given sub-banks to represent iris feature via testing different sub-banks in different illumination and noise and make multi-scale part directional feature code.3) In iris mode matching layer, we realize the algorithm which is named as the iris recognition arithmetic based on bionics multi-scale. It is the latest technique in the laboratory.In this paper, the object oriented layer class structure unify exoteric iris feature recognition system is proposed and designed. And each layer of the system is realized and inset some corresponding algorithms. These experiments show that the system framework has good logic rationality, and the system platform is stable, reusable, compatible and extensible. . |
PDF Full Text Request