The Research On Key Technique Of Iris Recognition On Moschus Berezovskii

Moschus berezovskii wildlife is the first-class protected wild animal in China, of which musk is an irreplaceable material of medicines and perfumes. It is endangered on account of being hunted with reckless abandon. Our country has been working on artificial rearing in the Moschus berezovskii to protect them, while the Moschus berezovskii individual identification becomes a significant problem. The traditional methods of identifying individual animals not only do harm to the animals to some extent, but also the flags are susceptible to wear and modifying, which subsists a lot of hidden trouble.Therefore in recent years, they are replaced by identification technology based on biological characteristics. Among which the iris recognition becomes the safest and most accurate method of identification with its high universality, uniqueness, stability and other characteristics.Almost all of currently existing iris recognition research is based on human beings, but rare research on animals has been recorded at home or abroad. Human’s iris structure differs from animals’, so in the Moschus berezovskii iris recognition it is impossible to follow the example of the human’s,especially in the most critical and difficult step of iris localization.This thesis unprecedentedly does the research of iris recognition on Moschus berezovskii’s individual identification. In the thesis, a new optimized localization method based on method of least squares will be proposed according to the structure of the The elliptical ring structure of Moschus berezovskii iris and the quality of the collected images. Using this method, results are better in decreasing the influence of dead pixels and fast localizing of iris. In normalization, an effective method based on tiered oval loop sampling is proposed.The thesis is committed to the key technologies of all aspects of the Moschus berezovskii iris recognition. The main work is as follows:(1) Moschus berezovskii iris image collection. The thesis analyzes the requirements, difficulties and equipment of iris image.(2) Moschus berezovskii iris detection and localization. In this thesis, the classic human eye iris recognition localization methods will be analyzed, and then a new localization method based on method of least squares will be proposed according to the structure of the Moschus berezovskii iris.(3) Moschus berezovskii iris normalization. A new normalization method based on tiered oval loop sampling is proposed according to the elliptical ring structure of the Moschus berezovskii iris.After that image enhancement processing will be done.(4) Moschus berezovskii iris feature extraction and matching. In this thesis, two-dimensional Gabor filter method and Hamming distance method will be used in extracting iris feature and iris matching.The test proves that the positioning method proposed in this thesis gets higher precision and speed. And the Moschus berezovskii iris identification process established on the methods proposed in this thesis acquires preferably effect.