查詢結果分析
來源資料
相關文獻
- Iris Recognition Based on Directional Empirical Mode Decomposition
- 以Cantor集觀念解析水文變量在時間尺度上之行為
- 地理資訊系統和碎形維度於森林地景空間變化上之應用
- 碎形維度於淨水處理上之應用
- 應用碎形維度監測森林地景結構與變遷之研究
- 多孔材料之鑑定:展透、碎形及多碎形分析
- 生物認證簡介與應用
- 以自製軟體重現著名碎形並蒐尋未知奇異吸子
- Multifractal Measures of Epicentral Distribution of M ≧ 6 Earthquakes in the Northsouth Seismic Belt of China
- 高雄都會區單顆粒氣膠物理化學特性與污染來源推估
頁籤選單縮合
題名 | Iris Recognition Based on Directional Empirical Mode Decomposition=運用方向性經驗模態分解法於虹膜識別 |
---|---|
作者 | 韓維愈; 李衍博; 李仁軍; 張劍平; | 書刊名 | 中正嶺學報 |
卷期 | 41:1(A) 2012.05[民101.05] |
頁次 | 頁29-42 |
分類號 | 312.13 |
關鍵詞 | 生物認證; 方向經驗模態分解; 本質模態函數; 碎形維度; Biometrics; Directional empirical mode decomposition; Intrinsic mode function; Fractal dimension; |
語文 | 英文(English) |
中文摘要 | 隨著資訊安全與門禁管制的需求逐漸增加,越來越多的個人身分認證傾向於運用生物識別;在現階段生物識別方法中,人類虹膜仍是最有效的方法之ㄧ。本文提出運用方向經驗模態分解(directional empirical mode decomposition, DEMD)及碎形維度(fractal dimension, FD)方法,有效擷取虹膜特徵實施識別。首先將正規化後之虹膜影像,運用方向經驗模態分解法依據不同頻率分解為數張二維本質模態函數(2D intrinsic mode function, IMF)影像,接著運用差方盒計算方法,在數張本質模態函數影像中找出虹膜紋理特徵。為評估方法之有效性,本文運用三種相似度測量法分別實施識別;實驗驗證於兩種公認的資料庫(CASIA及ICE),結果顯示出提出的演算法具有較佳的效果。 |
英文摘要 | As the increasing demand of information security with entrance governing of military building, so does the attention pay the biometrics based, automated person identification. Among current biometrics approaches, one of the most promising techniques is the basis on the human iris. In this paper, we proposed an effective algorithm for iris recognition. The methodology involves an extraction of iris features using directional empirical mode decomposition (DEMD) and fractal dimension. After the preprocessing procedure, the normalized effective iris image is decomposed into different 2D intrinsic mode function (IMF) components according to different frequency by the directional empirical mode decomposition. Then, the texture feature of each intrinsic mode function image is obtained via the differential box-counting method. The experiment results on the CASIA and ICE iris databases show that the presented schema achieves promising performance and is feasible for iris recognition. |
本系統之摘要資訊系依該期刊論文摘要之資訊為主。