To improve the accuracy of land-cover classification using the hyper-spectral data (i.e., hyperion data), a GA (Genetic Algorithms)-based Band Selection algorithm (termed "GBS algorithm") for hyper-spectral classification has been proposed. The required condition for improving the classification accuracy is not only to maximize the classification accuracy for the training-and the reference data sets used to evaluate the overall accuracy, but also tc minimize the error rates of omission- and commission-error. For satisfying those conditions simultaneously, the effective bands for classification are searched through GA operations. Toward the end of the run, "51 bands" out of all hyperion bands were selected as the effective bands for classification. As the results of maximum likelihood classification based on the selected bands, the average of classification accuracy with respect to training classes has increased from 80.5% to 98.1 %, while the average of error rate has decreased from 40.7% to 3.4%. Furthermore, PCC (Probability of Correct Classification) with respect to the reference data has increased from 72.2% to 81.4%. Those results indicate that GBS algorithm might be useful for selecting the effective-band for hyper spectral classification.
Wetland ecosystem is highly sensitive to environment and climate changes, so its changes are not only changing locally but also globally. And, wetlands play an important role in the control of global warming, which sink the methane as a source of greenhouse gasses. Therefore, it is urgent necessity to develop the methodology for monitoring the distribution of wetland vegetation. The purpose of this paper investigates remote sensing methodology for monitoring the distribution of wetland vegetation by using multi-frequency and polarimetry SAR (Synthetic aperture radar) data. Vegetation classification was carried out with single temporal X-band and L-band Pi-SAR (Polarimetric and interferometric-SAR) images used unsupervised classification by minimum distance classifier. And, in order to investigate the change of classifications accuracy by reducing speckle noise, the classification was performed difference kind of filter and its window size. As the result of unsupervised classification for the wetland vegetation, a classification map was produced which evaluates the calculation of classification accuracy each test area based ground truth.
Many optical sensors equipped on earth observation satellites have onboard calibration units. A solar diffuser is one of such units and a pattern of reflected radiance yielded by the solar diffuser directly affects the onboard calibration accuracy. In this paper, we present the results of our measurements of bidirectional reflectance distribution functions (BRDFs) for three witness samples of the solar diffuser in the Global Imager (GLI), a wide-field, multi-color radiometer equipped on the earth observation satellite ADEOS-II, to comprehend the scatter properties of the diffuser before degradation by UV radiation in space. Three witness samples are diffuser samples for GLI engineering model (EM) and pre-flight model (PFM1 and PFM2). The measurements of BRDFs were performed at wavelength of 490, 565, 666, 763, and 865nm by using the multidetector hemispherical optical scattering instrument. In the case of small angle of incidence (θi ≤ 30°), our results show that little azimuth differences in BRDFs appear, in contrast with the clear appearance of zenith angle dependence of BRDFs. Furthermore, it is found that the strength of BRDF decreases as the zenith angle increases. On the other hand, at large angle of incidence (θi=60°), a pattern of BRDF approaches that expected in a Lambertian surface, except in the certain direction related to the specular angle. At the direction close to the specular angle, the strength of BRDF increase remarkably as the zenith angle increases. Concerning a wavelength dependence of BRDF, we measured a relative reflectance from the wavelength of 400nm to 2.2μm by using a spectrophotometer. The results show no appearance of significant wavelength dependence of BRDFs in our sample. It is found, moreover, that the discrepancy of the intensity of BRDFs occurs between the EM witness sample and PFM samples. That is, the EM sample has systematically larger BRDF compared with BRDF of the PFM samples. The maximum difference between the BRDFs for the EM and PFM samples, i.e. about 20% in strength, occurs in the scattering angle larger than 30°. On the other hand, the BRDFs of two PFM samples were agreed within the 1% difference. Since our measured was performed for the non-degraded witness samples, measured BRDFs would be able to use as a reference to evaluate the degradation trend of the on-board diffuser.
This study sought to develop a method that effectively uses multi-spectral data to estimate photosynthesis. Information from multi-spectral data was concentrated in a vegetation index based on pattern decomposition (VIPD). In particular, we investigated the relationship between VIPD from multi-spectral data and the lightphotosynthesis curve of a leaf or tree crown. To examine this relationship, we measured the light-photosynthesis curve and the reflectance of leaves and whole crowns of three plant species. We then developed a method to estimate photosynthesis using the VIPD and confirmed that estimated photosynthesis approximated measured photosynthesis. The VIPD-based photosynthesis model was found to be both simple and useful for estimating gross photosynthesis. Broad-scale distribution of net primary production (NPP) can be obtained by applying this method to remotelysensed data by satellite or airborne.
This paper describes two effective techniques for reducing most of the static types of blur effects in remote sensed images. The first method is based on the sub-pixel shifted multi-observation of a scene. It attempts to sharp the image by down-sampling and matching a set of sub-pixel shifted frames, and calculating the statistical weighted average within the correspondent aligned pixels of the multi-frame set. The second technique reduces the blur of a single frame image by re-assigning the maximum and minimum intensity values of the blur-width area to the correspondent pixels in the direction of the gradient of the blur. Then, the area is shifted 1 pixel and the same process is done again, up to cover the full image. Real patterns are used for the analysis. The effectiveness, adaptability and simplicity of the methods presented here, is demonstrated.
The authors proposed previously a method for evaluating rock mass slope stability by using the remote sensing on the ground. The method is based on the observation of the spectral reflectance (Lab value) and temperature of the slope surface. From the observed data, we estimate the seismic wave velocity and the ultrasonic wave velocity of slope surface, and evaluate the slope stability by using the conventional and empirical relationships. This paper applies the method to a lot of rock mass slopes, and verifies that the method can be applied effectively to various types of rock slopes.