| Summary: | 碩士 === 國立高雄第一科技大學 === 電腦與通訊工程所 === 96 === In recent years, with progress of science and technology, personal identity becomes more and more important. Iris recognition is a reliable biometric identification technology because the iris is unique and stable. And, its feature do not like other biometric characteristics that is not easy to be influenced by injuring and time.
After studying many iris recognition methods in the literature, three recognition methods are chosen to perform combination by using multiple experts concept in this thesis. The iris image database used in this thesis is the CASIA Iris Image Database obtained from Institute of Automation, Chinese Academy of Science. After simple image preprocessing, iris feature could be extracted and used to perform recognition. In this thesis, there are three multiple experts combination methods including voting method, the fixed weighted average method, and adaptive weighted average method. After multiple experts combining, recognition error rate could be reduced. As a result, a more accurate iris recognition system can be obtained.
|
|---|
