Continuous fingerprint image quality estimation based on neural network

Fingerprint image quality estimation is crucial in eliminating poor fingerprint images, which will affect the performance of the automatic fingerprint identification system. In this paper, we present an image quality estimation method based on neural network. Unlike other methods, which are also based on neural network, we directly take the gray value of fingerprints as inputs into the network. This algorithm is tested for a large number of fingerprint images with the varied quality, and is found to be effective and fast. The results of segmentation of fingerprint images using our method proposed here are presented.