モルフォロジー解析を用いた非接触手形画像認証システム

概要

論文の詳細を見る
This paper proposes a non-contacting certification system by using morphological analysis of hand images to access security control. The non-contacting hand image certification system is more effective than contacting system where psychological resistance and conformability are required. The morphology is applied to get useful individual characteristic even if the pose of a hand is changed. First, a hand image is captured using the transmitted lighting. Next, the wrist area is removed from the hand area. The pattern spectrum that represents the form of the hand area is measured by the morphological analysis, and the spectrum is normalized to the invariant pattern to the scale change. Finally, the certification of an individual is performed by the neural network. The experimental results show that the sufficient accuracy to certificate individuals was obtained by the proposed system.
社団法人電気学会の論文
2004-02-01