[Application of Two-Dimensional Near-Infrared Correlation Spectroscopy in the Specificity Analysis of Noninvasive Blood Glucose Sensing].

In this paper, two-dimensional (2D) correlation spectroscopy analysis was applied to investigate the influence of the main component in blood and the systematic drift during the measurement on the specificity of glucose in the near-infrared (NIR) spectroscopy. First, the NIR transmittance of glucose aqueous solutions was measured and the 2D correlation NIR spectra were calculated under the perturbation of glucose concentration. Based on the comparative analysis for synchronous and asynchronous 2D correlation spectra, the characteristic absorption peaks of glucose in the combination band and the overtone band were determined. Then a small amount of albumin was added into glucose aqueous solutions, and the transmittance was recorded to perform 2D correlation spectroscopy analysis under the perturbation of glucose concentration. However, the absorption of glucose in the first overtone band (1590nm) and second overtone band (1195nm) was no longer homologous in the 2D correlation spectra, which means that the albumin may reduce the specificity of glucose. Further, the oral glucose tolerance test of healthy volunteer was conducted and the NIR diffuse reflectance of left palm was collected in vivo. The 2D correlation analysis results showed that, the homology of glucose in the diffuse reflectance was also destroyed. Moreover, as the spectral variation from the glucose concentration change is too low to be covered by that induced by systematic drift easily, some background correction methods were usually required. For the transmittance experiment of glucose aqueous solutions and the diffuse reflectance experiment of human body, the pure water sample and 5% diffuse reflectance standard were used as the reference, respectively. Then 2D correlation spectroscopy was developed under the perturbation of measurement time. Results showed that, smaller band shift was observed in the slice spectra of 2D correlation synchronous spectra after the corresponding background correction, and the specificity of glucose was improved both in the in vitro and in vivo experiments. So for the non-invasive glucose sensing by NIR spectroscopy, the wavelengths should be chosen carefully to avoid the absorption band of some interfering components which may destroy the homology of glucose and make spectral interpretation more complicated. And the selection of reference samples for relative measurement is also important to improve the specificity of glucose.

A novel classification method of halftone image via statistics matrices.

Existing classification methods tend not to work well on various error diffusion patterns. Thus a novel classification method for halftone image via statistics matrices is proposed. The statistics matrix descriptor of halftone image is constructed according to the characteristic of error diffusion filters. On this basis, an extraction algorithm is developed based on halftone image patches. The feature modeling is formulated as an optimization problem and then a gradient descent method is used to seek optimum class feature matrices by minimizing the total square error. A maximum likelihood method is proposed according to priori knowledge of training samples. In experiments, the performance evaluation method is provided and comparisons of performance between our method and seven similar methods are made. Then, the influence of parameters, performance under various attacks, computational time complexity and the limitations are discussed. From our experimental study, it is observed that our method has lower classification error rate when compared with other similar methods. In addition, it is robust against usual attacks.