2020/05/16 by Josef Bigün, Bigun, Josef, Fernando Alonso‐Fernandez +1
Computer Science · Earth and Planetary Sciences · Mathematics · #AI in cancer detection #Advanced Neural Network Applications #Algorithm #Artifact (error) #Artificial intelligence #Biometric Identification and Security #Computational Physics and Python Applications #Computer Vision and Pattern Recognition (cs.CV) #Computer science #Computer vision #Cryospheric studies and observations #Estimator #FOS: Computer and information sciences #FOS: Electrical engineering #Fingerprint (computing) #Fingerprint recognition #Geometry #Image (mathematics) #Image and Video Processing (eess.IV) #Linear phase #Mathematical analysis #Mathematical physics #Mathematics #Minutiae #Pattern recognition (psychology) #Phase (matter) #Phase congruency #Physics #Pure mathematics #SIGNAL (programming language) #Scale (ratio) #Seismic Imaging and Inversion Techniques #Statistics #Symmetry (geometry) #Tensor (intrinsic definition) #electronic engineering #information engineering
paper · pdf · doi:10.48550/arxiv.2005.08108
openalex publication_date 2020/05/16 · openalex created_date 2022/07/26 · openalex updated_date 2026/07/28
We reveal that the Analytic Signal phase, and its gradient have a hitherto\nunstudied discontinuity in 2-D and higher dimensions. The shortcoming can\nresult in severe artifacts whereas the problem does not exist in 1-D \nsignals. Direct use of Gabor phase, or its gradient, in computer vision and\nbiometric recognition e.g., as done in influential studies\n citefleet90,wiskott1997face, may produce undesired results that will go\nunnoticed unless special images similar to ours reveal them. Instead of the\nAnalytic Signal phase, we suggest the use of Linear Symmetry phase, relying on\nmore than one set of Gabor filters, but with a negligible computational add-on,\nas a remedy. Gradient magnitudes of this phase are continuous in contrast to\nthat of the analytic signal whereas continuity of the gradient direction of the\nphase is guaranteed if Linear Symmetry Tensor replaces gradient vector. The\nsuggested phase has also a built-in automatic scale estimator, useful for\nrobust detection of patterns by multi-scale processing. We show crucial\nconcepts on synthesized fingerprint images, where ground truth regarding\ninstantaneous frequency, (scale & direction), and phase are known with\nfavorable results. A comparison to a baseline alternative is also reported. To\nthat end, a novel multi-scale minutia model where location, direction, and\nscale of minutia parameters are steerable, without the creation of\nuncontrollable minutia is also presented. This is a useful tool, to reduce\ndevelopment times of minutia detection methods with explainable behavior. A\nrevealed consequence is that minutia directions are not determined by the\nlinear phase alone, but also by each other and the influence must be corrected\nto obtain steerability and accurate ground truths. Essential conclusions are\nreadily transferable to N-D , and unrelated applications, e.g. optical flow\nor disparity estimation in stereo.\n