Japanese mountain areas have abundant snowfall in the Sea of Japan side, and avalanches tend to cut off the traffic and crush houses. Thus, many avalanche disasters brought heavy damages. Therefore, it is important to improve avalanche risk estimation method and to map the risk. Avalanches are likely to occur at slope where snow depth is deep, slope inclination is steep, vegetation is sparse, and there are shallow furrows on bedrock incised by past frequent avalanches. Previous study scored these four explanatory variables, and mapped the risk based on the accumulated scores. However, it is difficult to estimate adequate weights on the four variables in accumulating them. This is because the relation between actual avalanche and the four variables is seldom to be revealed simultaneously. This study aims to investigate the relation between them and propose an improved avalanche risk map. Study area was selected in Yamakoshi district in Nagaoka city where some 2, 000 avalanches, whose distribution was mapped, occurred under the influence of the recorded snowfall in 2004-2005 winter. In the investigation and the mapping, five-meter-grid data derived from aerial photo interpretation and image processing, and airborne LIDAR data were used. At each grid, ideally, the higher the accumulated score is, the higher the avalanche risk. Comparing with the previous study's risk map, this study could present a better risk map where the accumulated score showed very high positive correlation with avalanche ratio.
Recently, there were various disasters such as Katrina flood, seismic wave in Sumatra, great earthquake in several countries. Inthese disasters, it has become one of the most important rescue activities to detect quickly the victims. If automatic detection of these victims in vast disaster fields is feasible, searching efficiency will be improved spectacularly in comparison with traditional visual search from aircrafts. In this paper, we propose Normalized Difference Human Index (NDHI) in order to extract human images out of various background objects efficiently. For the NDHI, we examine spectral characteristics of the human skin in short wave infrared band, and chose the best wavelengths for its calculation. For marine peril or water disasters, we have conducted experiments with human extraction in and out of the mud water using NDHI. In order to extract objects in the mud water, it is necessary to calculate the threshold operation on the reflectance of 1070nm wavelength as preprocessing of NDHI. In addition, we examined allowable water depth to extract objects for each transparency (intermediate-water, 40, 20, 10cm) . Finally, comparable studies with the conventional thermography measurements were also carried out.
This paper shows the object-based forest type classification using texture features from a panchromatic (PAN) image and spectral features from a multispectral (MS) image obtained by the QuickBird satellite.To investigate the performance of each feature, first, only texture feature is applied to image analysis, then spectral feature, and lastly combination of texture and spectral features.In this analysis, we use common segments obtained from a pansharpen image in order to compare the difference only between texture and spectral features.Distance between supervised classes is used to find well distinguishing feature combinations for classes.For PAN image analysis, 4 texture features from 8 candidates generated from co-occurrence matrix were selected. For MS image analysis, 9 spectral features from 10 candidates, such as 4 bands value and 6 differences between 2 bands from 4 bands, were selected.For PAN and MS analysis, 3 texture features from 8 candidates and all 10 spectral features were selected.Overall accuracy and Cohen's kappa of 6 forest types classification were 32.6% and 20.4% for PAN image, 74.6% and 70.6% for MS image, and 79.3% and 76.0% for PAN and MS images.This study demonstrated that combination of texture and spectral features exceeds a single feature in accuracy.
This paper introduces the robust matching algorithm of time series laser scanner data for indoor map making. We first describe the background of our research. Then we explain the matching algorithm in detail that is basically improved on “robust fuzzy clustering” method. Through experiments, effectiveness of the algorithm is confirmed in relatively small rooms with a few walkers and wide area with many walkers, respectively. Finally, this paper is concluded with some issues for future study.