適用於行動運算環境之語句式指紋分類模型的研究

A Study on Syntactic Model of Interpreting Fingerprint Topology for Mobile Computing Environment

在這篇論文中，我們擬針對無線通訊技術頻寬限制情況下發展合適指紋分析及其分類與辨識相關的技術。因為頻寬受限制，因此接受比對之指紋影像所需傳送之資料量必須做相對應之縮減。同時希望在資料量縮減下，還能保持指紋特徵的不失真。除此之外，行動環境之計算能力相對薄弱，也有必要研發另一套有別於現存之幾何形式指紋特徵比對法，以加快行動身份認證的速度。觀察各類型的指紋，我們發現組成所有指紋的紋路只有十種基本的樣式(fundamental ridges)，而不同類型的指紋，在不考慮紋路的分支(bifurcation)與斷線(fragment)的情形下，都只是由這十種不同的指紋紋路樣式依照特定的順序所組合而成的。基於這個原理，我們提出一種語句式(syntactic)表達法的新模型，以不同的句子來代表個別的指紋，句子中的字母(alphabet)就是那十種指紋紋路的基本形式。並且以此方法為基礎，發展出一個指紋分類系統的新模型。另外，我們也嘗試利用紋路分佈所具備的特質來定義出一個清楚的拒絕分類標準。這個分類法除了能完成Henry分類的那七種有效分類以外，也可以使資料庫中的子集合數目大量增加。對於類似全民指紋建檔等的超大型指紋資料庫，或許才可能真正落實。

In this paper, a syntactic model for interpreting fingerprint structure which can be use in a mobile computation environment is proposed. Because bandwidth is limited, therefore the amount of fingerprint image data transfer must be reduced. From our observation, there exist only ten basic ridge patterns which constructing fingerprints. Fingerprint classes can be interpreted as a combination of these ten ridge patterns with different ridge distribution sequence. The classification task is performed depending on the global distribution of the ten basic ridge patterns by analyzing the ridge shapes and the sequence of ridges distribution. This classification scheme can accomplish the seven-classes discrimination of the Henry's classification. Moreover, the number of the subsets in database can be increased largely by combining the concept of structural classification and the number of ridges. For this reason, we will design the ridge number calculation and the singular point detection method in gray-scale domain. We also will define a rejection standard by the appearance of the type line.