透過您的圖書館登入
IP:18.188.111.130
  • 會議論文
  • OpenAccess

利用雷射光斑指紋影像進行身份識別之分析

摘要


根據雷射光斑影像的高解析度與不變形之特性,可以從中獲得非常精密之定位資訊,並實際使用於機械精密加工等應用。本論文擬將雷射光斑取像方式從平面材料延伸至手指紋路,並探討其紋理特徵擷取方式與識別演算法。在實驗階段,我們以連續及快速方式,取得雷射指紋光斑影像,並分別以手指固定與滑動的兩種方式進行取像及建立資料庫。同時,我們利用SURF演算法及2DPCA演算法進行影像前置處理,從光斑影像當中擷取特徵點與使用原始影像進行統計分析。接著,再以特徵為主及以區域為主的影像比對方法進行識別分類。因受限於取像裝置之設計,實驗結果顯示,以手指固定方式取像及使用2DPCA演算法之組合才能獲得較佳之結果。

並列摘要


According to the properties of high resolution and invariance from Laser Speckle Patterns(LSPs), positioning information can be obtained and used in the applications of mechanical precision manufacturing. This paper extends the image capturing from surface materials to fingerprints. Also, texture features and classification algorithms are studied. In the experiments, LSPs are captured by fixing and sliding the finger to create a database. Meanwhile, SURF and 2DPCA algorithms are used to preprocess the images. Feature points and original images are statistically analyzed. Next, feature-based and area-based methods are adopted for image classification. Due to the design limitation of image capturing device, experimental results show that best results can only be obtained from the combination of finger-fixed images and the 2DPCA algorithm.

延伸閱讀