Improved techniques for an automatic fingerprint identification

Abstract

This paper compares the technique for increasing the efficiency of an automatic fingerprint identification system. The techniqes considered in this paper are classified into two main processes; the preprocessing and the feature extraction processes. In the preprocesses, tqo techniques (binarized image by lacal thresholding and seperated between background by variance) are considered. In the feature extraction process, the orientation field method is compared with directional image which is a traditional method. The recognition process used in this study is based on the back propagration network. The experimental study was conducted in order to evaluate the efficiency of the methods. The identification was tested on the test image so that the signal to noise ratio of the test image are 16.87, 8.46, 5.71, 4.32 and 3.51. The results showed that the orientation field method improved the efficiency by more than 30% as compare with the directional method. In addition, the seperation between background and foreground increases the efficiency around 4% as compare with binarization image by local thresholding