Font Size: a A A

A Fast Algorithm For Iris Feature Extraction Based On Harr Characters

Posted on:2014-01-17Degree:MasterType:Thesis
Country:ChinaCandidate:L F ShiFull Text:PDF
GTID:2248330395997478Subject:Software engineering
Abstract/Summary:PDF Full Text Request
The Iris recognition system is recognized as the most prospects and most safe technicin all kinds of biological identification system. The main reason is the high stability,unique, non-contact and difficult to counterfeit. Now many products based on the irisidentification system has been applied to many fields and play an important role in thefield of personal identification.The Iris recognition system includes many steps: image acquisition, imagepreprocessing, feature extraction and matching. With the image acquisition, preprocessingand matching step, some scholars have done a lot of research has also made good results,so this paper in these steps directly using the existing methods, this paper focuses on thefeature extraction.The basic principle of the iris localization is to identify the three non-collinear pointson the iris inner and outer boundaries, by using this three point determine the parameters ofinner and outer boundaries. Because the algorithm only involves locating three points, so itcan avoid blind search which can simplify the positioning algorithm to improve thepositioning speed. We can choose three points which not easy to block, can avoid theeffect of eyelash and eyelid, can improve the performance of the algorithm.In the detection process of the eyelash area, not only need to consider the accuracy ofdetection, also need to consider the time spent, and the robustness of the algorithm. Thispaper examined through the method of iris image and normalized by median filtering,through experiments, the method achieves very good results in detection of eyelash andeyelid.Because the existing iris recognition systems require the quality of iris image withhigh definition, and the high complexity of the algorithm. In order to reduce thealgorithmic complexity and sensitivity to image quality, the main work of this paper is thecharacteristics of the iris image extraction algorithm research experiment. By the analysisof the related technical background about real-time moving object recognition algorithm,some characteristic algorithm like Harr characters are applied in the iris recognition system.By projecting different kinds of Harr characters to a projection matrix as the feature vector to classification. This algorithm has the advantage of fast and reduces the complexity offeature extraction.In order to verify the effectiveness of the algorithm, we choose Matlab as theexperimental tool. The tool provides a lot of existing image processing algorithms,especially the preprocessing and matching parts. Bringing great convenience for theresearch. In order to analysis the experiment results, this paper used the Chinese Academyof Sciences iris database to analysis experiment of the algorithm and other relatedalgorithms, and analyzed the final result at last.
Keywords/Search Tags:Iris Recognition, Harr character, Matching
PDF Full Text Request
Related items