Verification of the algorithm for estimating the flow chart of fingerprint images

The construction of the flow chart for fingerprints digital images is an important step in biometric identification and comparison procedure. In particular, the results of building the flow chart significantly affect the filtering of such attributes of images as minutiae, loops, deltas and curls. In addition, improving the quality of the flow chart construction can significantly reduce the overall identification error. This article describes the developed algorithm for constructing the flow chart of fingerprints digital images. The proposed algorithm is based on an improved method of gradients. This method is applied in two stages with different sizes of aperture and two types of smoothing: common and along the direction of the pattern lines. The quality of this algorithm was evaluated using a web-framework created on the platform of the University of Bologna (Italy). This framework is created for automatic, remote evaluation of the results of different fingerprint recognition algorithms including the estimation of the flow chart. The requirements for the program structure, input and output data imposed by this framework were fulfilled. The outputs of the algorithm work evaluation by the selected framework are considered and analyzed. A comparison of the algorithm results evaluation with the original gradient method is made, as well as with the results of other test participants published in open access.