Research On Extraction Algorithm Of Palm Fine Lines And Colors Characterizing Physical Health

Traditional Chinese doctors can diagnose health by observing colors, lines and shapes in different position of the palms, however they should possess much medical consultation experience to get these skills. So it is great significant to diagnose health automatically for disease deterioration or prevention by palmprint identification technology.With the rapid development of biometrics over the past ten years, palmprint identification technology as an important member of biometrics has been recognized widely in the world and achieved many research results, but there are still some problems needed to be solved for using palmprint recognition technology to determine people’s health condition. For example, the extracting main palm lines are discontinuous and shallow, and some abnormal lines cannot be extracted as well as the palm color spots cannot be extracted by the existing methods. Therefore, the extraction methods of palm fine lines and colors are deeply studied in this paper, which include five aspects as follows.(1) To solve the problem of discontinuous and shallow main lines, a new algorithm for repairing the discontinuous main lines is presented in this paper. Firstly, the algorithm expands the lines using a Gaussian function based on image energy. Secondly, it computes probability distributions of sub-image and non-highlights spots. Finally, the image pixels are regressed exponentially. Compared with the traditional methods using expansion and thinning, the algorithm can solve the problem that misconnected and discontinuous main lines with long distance cannot be repaired.(2) To solve the problem of extraction of the fine lines, a new method for feature extraction of palmprint line based on NSCT transform and mathematical morphology is proposed. The NSCT transform has the characteristics of anisotropy and multi-scale for palmprint decomposition, and it has the translation invariance which can directly deal with the sub-band coefficients to analyze pixel information for head line features extraction. So we use NSCT to decompose the palmprint images under different spectra, and integrate the coefficients according to the characteristics of fine lines under the single spectral, so as to obtain the fine lines of the palmprint by the morphological thinning after the inverse transformation. The experimental results show that the fine lines extracted by this method are rich and clear.(3) To solve the low recognition rates of the existing algorithms while extracting cross structure and mi structure, a new method using accumulator vote principle of Hough transform is proposed. The multi-spectral palm print collecting platform is used to capture pictures and build the database. The new Gaussian model is used to process the pictures, and then Hough transform is adopted to extract cross structure and mi structure. The experimental results show that this algorithm can improve the recognition rate effectively, and lay the foundation of characterizing physical health.(4) To solve the problem of identifying the colors of the spots on the palms accurately and quickly, a new algorithm called H-S-Gray color spots extraction is provided. This algorithm does not change the image contrast and reduce the color distortion of the color space as little as possible, and it makes the processed images be beneficial to identification color spots by pre-processing such as stretching and quantization of the color space.(5) In order to verify the effectiveness of our recognition algorithms, a multi-spectral palm print collecting platform has been made. To solve the problem of low accuracy with few information of the palmprint images in a single spectrum, the multi-spectral palmprint collecting platform can capture the local palmprint images in six spectrums with much more information than before to reach higher accuracy. Moreover, it supports the palmprint image being not only observed in real time, but also collected and stored selectively.In conclusion, the algorithms of extracting the fine lines and identifying the colors of the spots on the palms are deeply studied, which can provide a theoretical basis for characterizing the physical health automatically.