Importance of Image Enhancement Methods for Fingerprint Recognition

Anil Osman Tur, Berkay Selbes, Halil Ibrahim Öztürk, Ismail Karakaya, Berkan Demirel

Abstract: In recent years, due to the critical importance of fingerprint detection in security and criminal areas, the number and quality of the work done has been increasing. Fingerprint minutiae must be extracted correctly to be used in these studies. Many methods are suggested for extracting fingerprint minutiae. The proposal methods do not provide the desired results when applied to the raw fingerprint image. In order to have a high performance of the methods, it is necessary to ensure that the fingerprint characteristic on the input image is distinct. In this study, the contribution of applying different image enhancement techniques to fingerprint feature extraction was examined. As enhancement methods, Gabor filter enhancement, Local contrast enhancement, Gaussian Local Contrast enhancement, Short Time Fourier Transform enhancement, Texture Decomposition and Autoencoder were applied separately. The results of these methods were used in 2 different minutiae algorithms. The methods were compared on 4 different data sets according to Top-1, F1, Precision and Recall performance metrics.