快速嵌入式虹膜辨識系統

文化大學機構典藏 CCUR > 理工學院 > 工學院 > 資訊工程學系 > 研究計畫 > Item 987654321/2165

jsp.display-item.identifier=請使用永久網址來引用或連結此文件: https://irlib.pccu.edu.tw/handle/987654321/2165

题名:	快速嵌入式虹膜辨識系統
作者:	李志仁
贡献者:	資工系
关键词:	虹膜辨識嵌入式系統
日期:	2008
上传时间:	2009-09-07 14:57:45 (UTC+8)
摘要:	虹膜特徵是生物特徵辨識中辨識率最高的辨識系統，所以在一些需要高規格的身份認證門禁上都會採用虹膜辨識系統。傳統上虹膜辨識系統的步驟包含：定位、正規化、影像強化、特徵萃取、特徵比對。由於建立嵌入式辨識系統，除了考慮正確率外，還必須考慮辨識速率，所以必須再加速這些步驟，讓嵌入式虹膜辨識系統更容易實現。由於蓋伯函數與視覺系統相近，並且已成功地應用在許多領域，尤其是人臉、指紋和虹膜辨識上。因此所發展的虹膜辨識系統以蓋伯函數為特徵處理層，將可適應不同的應用特性，進而修正函數參數，來萃取有效的生物特徵。此類神經網路不但具有適應學習的能力，而且承襲蓋伯函數的優點──不必經過複雜的前處理程序，直接使用灰階影像，就可以萃取生物特徵，讓所提的辨識系統也可以廣泛地適用到其他生物辨識系統。為展示所提出方法的穩健與可靠，我們不但使用公開的虹膜資料庫，並且自行蒐集各類狀況的資料庫，然後經由「一對一」與「一對多」比對的實際應用與指標評估，來證實所提的方法既快速又正確。最後我們將所提的虹膜辨識系統，實現在TI DSP 的開發平台上，以實現即時又方便的嵌入式虹膜辨識系統。 Owing to high recognition rate, iris is the best features for biometric recognition. Some access control systems, therefore, adopt iris recognition for high security. Conventionally, the procedure of iris recognition includes iris localization, iris normalization, image enhancement, feature extraction, and matching. For embedded systems, both accuracy and efficiency are important. So we need to speed up some steps to implement embedded iris recognition. Simulating vision system, Gabor filters have been applied to many categories of image recognition, such as face recognition, fingerprint recognition, and iris recognition. We will design the modified Gabor filters to adjust adaptively according to iris features. Then based on the extracted features, those of the registered patterns from database are compared. The proposed approach not only has the ability of adaptive learning, but also inherits the advantages from Gabor filters. Without the complex preprocessing, the iris features are extracted from gray images directly. We also apply the proposed method to the other biometric recognition systems. To prove that the proposed approach is reliable and robust, we will not only perform iris recognition on-line, but also evaluate its performance via both one-to-one and one-to-many matching with various kinds of databases including the public and the self-collected. At last, we will implement the proposed embedded iris recognition system by TI DSP platform.
显示于类别:	[資訊工程學系] 研究計畫

文件中的档案:

档案	描述	大小	格式	浏览次数
index.html		0Kb	HTML	530	检视/开启

在CCUR中所有的数据项都受到原著作权保护.

数据加载中.....