Analysis Of Influencing Factors About Reconstructing Three-dimensional Fingerprints And Research Methods Of Extracting Its Characteristic

The traditional way of acquiring the Two-dimensional(2-D) fingerprints is contact-based, in which the fingerprints surface could be deformed by the uneven pressure, furthermore, the acquisition accuracy could be affected by human factors such as fingerprint stains, damages, drying and moisture. In addition, because of the 2-D fingerprints can be copied and has a poor ability of anti-counterfeiting, so that it is harmful to personal information protection. To overcome the shortages of 2-D fingerprints identification, the acquisition and recognition research of 3-D fingerprints is gradually appeared. The main purpose in this article is to research and analyze many factors that influence the 3-D fingerprints reconstruction based on grating projection method, then giving reasonable improvement to get a better effect of 3-D fingerprints reconstruction and to lay a good foundation for 3-D fingerprints features extraction and recognition.This article mainly aimed at researching and analyzing the influence factors about image filtering、solve the phase and phase unwrapping which belong to the method based on raster projection to acquire 3-D fingerprints. On this basis, presents a method of extracting the normalized fundamental frequency mode as a means of 3-D fingerprints characteristics and use it can achieve a better recognition rate. The specific work includes:(1)In term of 3-D measurement system, make an analysis from the choice of light source, grating and the CCD camera, at the same time, make a calibration to the parallelism and perpendicularity of the system and get its parameters. Finally, verify the stripes which collected by using the device and the sinusoidal grating stripes which get by computer simulation are basically identical.(2)In terms of image denoising, make a comparison about image filter effect between the method of spatial filtering、frequency domain filtering and the method of wavelet denoising. Finally adopting the method of wavelet to filter out the noise of image. In the aspect of extracting the fundamental frequency. Giving a method of using wavelet multiresolution decomposition to extract it, which can avoid the fundamental frequency spectrum leakage and aliasing problems, however, it can’t be avoided by the method of using traditional frequency domain window functions to extract the fundamental frequency.(3)In terms of phase unwrapping, the traditional methods of phase unwrapping have a poor ability to resist noise, shadow and discontinuous problems. So, considering a method which named quality figure direct path to unwrap phase and compared with the traditional unwrap phase methods, the former get a better 3-D fingerprint model effect.(4)Different from existing 3-D fingerprints feature extraction methods, this article presents a method to extract the normalized fundamental frequency as the 3-D fingerprints characteristics, without fitting the 3-D fingerprints on ellipsoid、cylindrical model and then projecting on 2-D plane to extract the feature and avoid the complex calculation in the processing.This paper through researching and analyzing the influence factors of the method based on raster projection to reconstruct 3-D fingerprints for the purpose of reducing errors and interference to improve the result. Adopted the method named quality figure direct path to unwrap phase and combined with the improvement of each step in the method of reconstructing 3-D fingerprints to get a good result. On the basis, though extracting the normalized fundamental frequency mode characteristic of 3-D fingerprints and use it to do the simulation experiment, prove its feasibility and effectiveness.