[Tongue Coat Information Extraction of the Traditional Chinese Medicine with Hyperspectral Image].

Tongue coat information extraction plays an important role in disease diagnosis for the traditional Chinese medicine. For the purpose of quantifying the tongue coat properties in traditional Chinese medicine, most of the existing methods are based on computerized image analysis, carrying out RGB color space in a tongue imaging captured with digital camera. However, those methods cannot meet the requirements of clinical medicine. To explore more information about the tongue objectively, a new approach to analyze tongue information based on hyperspectral images is presented. Hyperspectral images are acquired using the hyperspectral imaging system in the spectral range of 370.200 0 to 992.956 0 nm (343bands), and the traditional Chinese medicine clinical diagnosis information is recorded. The main region of interest (ROI) in the samples is extracted while the background is removed from the tongue image, then tongue information of ROI is analyzed. The largest different spectral characteristics between tongue proper and tongue coat are found in the wavelength range of about 525 to 600 nm. Nine wavelengths tongue spectral images from 382.108 0 to 963.668 0 nm are extracted, then comparing with the actual tongue situation, we find that the spectral image at 527.548 0 nm band can better reflect the actual tongue situation than others. The experiment results show that hyperspectral imaging technique is very helpful for tongue coat information extraction of the traditional Chinese medicine, and this new analysis approach can provide a fast and simple non-invasive detection method for tongue coat segmentation and tongue coat information extraction.

Analyzing Spectral Characteristics of Water Involving In-Situ Multiangle Polarized Reflectance and Extraction of Water-Leaving Radiance.

Polarization is defined as an asymmetry in the direction of vibration with respect to the direction of light propagation. Polarization information is an important component to remote sensed data, which comprises spatial, spectral, and radiation information. In optical remote sensing, polarization information supplements spectral information. Polarization-based remote sensing has a significant application potential for analyzing the spectral characteristics of water bodies, wherein a very important technique is eliminating the mirror reflection caused by skylight on the water surface and extracting water-leaving radiance that carry the constituent information. The incident sunlight on the surface of water either reflects or scatters owing to the existence of particles in water, which results in water-leaving signals with strong polarization characteristics. The ongoing experiments on remote sensing involve water polarization cover either clean ocean waters under natural light or indoor simulations of water under artificial light; however, turbid inland waters under natural light have rarely been investigated. Through the combination of a field spectroradiometer and a Thompson polarizing prism, this study obtained in-situ measurements of the spectral polarization reflectance over inland waters under natural light conditions. Using the obtained multiangle polarized reflectance spectra, the polarization spectral characteristics of water under multiangle viewing conditions were quantitatively analyzed, and the water-leaving radiance was achieved by eliminating skylight reflection. When observing water bodies at an azimuth of 135° and a zenith of 53°, the measurement of polarization to eliminate skylight reflection had better elimination efficiency than at other viewing angles, and this observation angle was recommended for conducting spectral polarized above-water observations. Compared with the traditional methods including non-polarized above-water measurements, the proposed method is less prone to being affected by changes in weather conditions, and it can extract water-leaving radiance more accurately.

Airborne validation of ground-object detection from polarized neutral-point atmosphere.

Based on the object's polarization effects, polarization is a newly emerging method in the field of remote sensing. Both objects and atmosphere have polarization effects, however, the atmosphere's polarization effects are much stronger than that of objects'. Consequently, atmosphere polarization effects will interfere or even cover objects' when observing with sensors. How to maximally eliminate the polarized effects generated by the atmosphere is a crucial problem in polarization remote sensing. Atmospheric neutral point is an area where the degree of atmosphere polarization is near to zero; therefore, if sensors are set up in this area, atmosphere polarization would be greatly eliminated, which is the main content of separating the effects between objects and atmosphere by its neutral point method. In this paper, after processing and analyzing the experimental data got from the first polarization remote sensing flight experiment with atmosphere neutral point, the degree of polarization images captured in neutral and non-neutral point area were obtained, and it can be seen that the main value of polarized degree of images got in neutral point area was obviously smaller than that in non-neutral point area. The results showed that the theory mentioned above was logical and practical. An innovation in our study is that the requirements needed in polarization remote sensing flight with neutral point were clarified. In the meantime, a qualitative conclusion was drawn that observing with longer wavelength is more applicable to polarization remote sensing.

[Research on the model of spectral unmixing for minerals based on derivative of ratio spectroscopy].

The precise analysis of mineral abundance is a key difficulty in hyperspectral remote sensing research. In the present paper, based on linear spectral mixture model, the derivative of ratio spectroscopy (DRS) was introduced for spectral unmixing of visible to short-wave infrared (Vis-SWIR; 0.4 - 2.5 microm) reflectance data. The mixtures of different proportions of plaster and allochite were analyzed to estimate the accuracy of the spectral unmixing model based on DRS. For the best 5 strong linear bands, the Pearson correlation coefficient (PCC) of the abundances and the actual abundances were higher than 99.9%, while the root mean square error (RMSE) is less than 2.2%. The result shows that the new spectral unmixing model based on DRS is simple, of rigorous mathematical proof, and highly precise. It has a great potential in high-precision quantitative analysis of spectral mixture with fixed endmembers.

[The multi-angle polarization spectral character of water and its applications in water color remote sensing].

The reflectance of pure water is very low at visible and near infrared bands. Its spectral characteristics are not obvious. Water always shows dark hue in optical remote sensing images. This dark hue causes the difficulties in water remote sensing identification. There is an interesting phenomenon when the authors research the water polarization spectroscopy. The authors measured water's polarization spectra and reflectance spectra at different view zenith angles using the ASD spectrometer. When the view zenith angle was zero (measured vertically), as the spectrum people commonly measure, there was no polarization phenomenon at the water surface, and the reflectance was low at each band. Along with the increase in view zenith angle, the DOP spectra curves increased evidently, while the reflectance curves only changed a little. When the view zenith angle was over 30 degree, the values of DOP spectrum were much larger than the reflectance spectrum values at the entire visible and near infrared bands. At some bands, the DOP value was several dozen times than its reflectance value. This phenomenon shows that the water's brightness in DOP image is much higher than its brightness in intensity image under the same condition. This rule was verified by the PARASOL multiangle polarization satellite data. Comparing the average brightness of DOP images with the average brightness of intensity images at 490, 670 and 865 nm band, the former is higher than the latter apparently. The brighter DOP images are better for water remote sensing identification It is the first time that the authors found this special multiangle polarization spectral character of water. It revealed the advantage of water detection using the multiangle polarization remote sensing data. This method solved the low reflectivity problem of water color remote sensing. It will greatly improve the capability of water remote sensing identification and the retrieval accuracy of water quality parameters.