Study On Detection Of Fingerprint Recognition Based On Hyperspectral Imaging Technology

With the high-speed development of social information and networking, peopleput forward higher requirements for personal identity security, accuracy andpracticality. The fingerprint has characteristics such as universality, uniqueness andpermanent, it has been widely used in the fields of finance, banking, security andgovernment agencies, but for the field of public security organs and the criminalinvestigation, the traditional fingerprint identification method has far cannot meet theneeds of the actual situation, especially potential fingerprints and fold fingerprints,brought a lot of trouble to the police solve crimes, the fingerprint identification needto inject new methods and new ideas.Hyperspectral images has the characteristics of mapping one, not only containsthe image information, but also contains the spectral information. This paper mainlystudied how to use spectral information for fingerprint recognition, combined withthe traditional method of fingerprint identification, use of the combination of the twomethods to determine the personal identity, the main contents are as follows:(1) Structures hyperspectral images acquisition system, shoot a fingerprintimage under different wavelength by hyperspectral cameras, obtain the originalhyperspectral fingerprint image. Filtered the original image through principalcomponent analysis method, gained the feature image of415nm,550nm and630nm.(2) Extracted the light absorption value to different wavelength image in a pixelsby optical fiber spectrometer and Matlab7.0, obtained the value for each wavelengthof the image in the pixel point, then processing the data by the data processingsoftware for data, get the fingerprint in the spectrum of the pixel curve.(3) Established the fingerprint spectra database and proceed fingerprint matchingand recognition for different fingerprint, and analysis of the spectrum curve andvalidation, including error analysis and error analysis of curve data.(4) Characteristics image were fused by wavelet fusion technique, obtainedbetter experimental results. Preprocessing the fusion image, mainly includes thedirection field calculation、filtering、binarization and thinning, then feature extractionand feature matching.(5) Combining the image information and spectral information, compared high spectral fingerprint identification method with the traditional fingerprint recognitionmethod, analysis the advantages and weaknesses of the high spectral fingerprintidentification method.By the contents of the study proved the feasibility of the high spectralfingerprint identification method.