Journal of Forensic Sciences · 2002 · 27 citations · 3 references
EngineeringBiometricsCross CorrelationImage ForensicsFast Fourier TransformImage AnalysisPattern RecognitionDigital Image CorrelationCharacter RecognitionStatisticsOptical Character RecognitionFourier AnalysisStatistical Pattern RecognitionFunctional Data AnalysisSignal ProcessingDiscrimination MethodFourier TransformFourier TransformationDocument ProcessingPattern Recognition Application
Abstract A non-destructive method for discriminating between different types of paper has been proposed, using image analysis, Fourier transformation, and cross-correlation matching. A fast Fourier transform (FFT) is used to extract the periodicity in the structure of paper that results from the manufacturing processes. The light-transmission images of the paper to be Fourier transformed are obtained from a flatbed image scanner. The similarity between the power spectrum of the FFT of the sample and that of a reference is quantified using a cross-correlation matching method. An advantage of using frequency analysis is that periodicity can be detected even if the sample is damaged or is printed on. The technique works on samples as small as 2 cm 2 .
3