Muitispectral Palmprint Image Fine Lines Extraction And Recognition

Palmprint recognition has high recognition accuracy, high speed, low costacquisition devices, user experience high comfort, and it obtains the academiccommunity full attention. Palmprint recognition in identification applications hasmade a series of research results, but the combination of palmprint feature recognitionand palmprint diagnosis of TCM theory is a relatively new field.There are lines on the palm: the main lines, wrinkles, mastoid grains and otherfine lines. The main lines and wrinkles can be extracted from a low-resolution,low-quality images, they are strong in division, very stable and usually used asimportant features for palmprint identification.While, some shallow wrinkles,mastoid grains and other fine lines will rise and fall along with changes in physicalhealth and daily living habits. Fine lines of specific shape appearing on specific areaon the palm means the health problems of corresponding organs. Therefore, extractinga wealth of fine lines on the palm and identifying its morphological features forautomatic palmprint medical diagnosis is of great significance.With the in-depth study of palmprint identification, the combination ofmultispectral technology and palmprint identification obtains a major concern ofscholars. Palm skin absorption and reflectance characteristics of different spectralhave different spectral characteristics. Generally speaking, the longer the wavelength,the stronger the radiation penetrating human skin, the sharper the palm vein patternunder the skin; the shorter the wavelength, the more for a particular imaging surface,the sharper to get fine lines on palm surface. For these features of multispectralpalmprint images, fuse specific spectral palmprint images to obtain clearer andabundant palm fine lines feature. The main work is as follows:Design multispectral palmprint collecting device: We designed a palmprintacquisition device with six spectrum, and it is able to accurately capture images of thesame area in different spectra, it can greatly reduce the error of image acquisition; bythe zoom lens in multispectral barrel, this device can capture the enlarged and clearerpalm fine lines; the movable multispectral camera can capture images at any location.Use this special device collect palmprint images and build multispectral palmprintdatabase, laying the foundation for future extraction recognition of fine lines.Multispectral palmprint image fine lines extraction and recognition: NSCT combinated with mathematical morphology algorithm is proposed to extract a largenumber of fine lines efficiently; in multi-spectral feature fusion stage, an improvedpixel-level feature fusion algorithm, which first feature enhancement for singlespectral image; in identification stage, using neighborhood coding and templatematching algorithm to identify the cross pattern; combined with the theory oftraditional Chinese medicine hand diagnosis,in order to determine the location of thecross pattern on hand, we use the improved SIFT algorithm stitch a full palm imagesand then position nine regions to predict disease of the body.