Interpretation of the geologic structure of the earth's surface using satellite data reflecting the subsurface fractures in the target area is the significant method for the exploration. Fracture analysis would be an unusually simple and valuable tool for understanding the structure in subsurface. This study is aimed at the analysis of relation of the earth's surface character and data processing methods using extracted lineaments from Landsat TM VNIR data. It is well known that the lineament patterns are the significant features to indicate the historic results of structural motion. In this paper, two VNIR Landsat TM data were chosen from the two categories of surface character, vegetated mountainous, and vegetated flat areas, to apply three types of digital filter, such as median (MD), bidirectional high frequency component enhancement (TY), and omnidirectional high frequency component enhancement (OM) . Types of the filters were selected from the points of enhancement and preservation of the edges contained in the original images. Four images, Original, MD, TY, OM, were generated in each area. Using these enhanced images, the lineaments were extracted and analyzed statistically. As the results, images of median filter in both category area preserve the lineament pattern of non-enhanced original images, but bidirectional and omnidirectional high frequency component enhancement images do not preserve the original image's character. The relation between lineaments and active faults existing in each study area, we can clearly find those active faults as lineaments on the images. Apparently, northward active fault running from Yoneshiro river can easily be recognized in each processed images of studied Noshiro area. Active faults running around Mt. Iwaki can also clearly be seen as lineaments in Hirosaki area. The matching ratio among active faults indicate that omnidirectional high frequency component enhancement images are in maximum in both studied areas.
The development of automatic system to detect cracking information on pavement surface has been expected in order for the maintenance of a proper condition and repair planning of ever-expanding road-network. It is difficult to develop such a system that substitutes the conventional human-dependent interpretation and evaluation applying computer-based image analysing methodology due to a wided range of variation in images caused by the sundry pavement surface conditions and tough shooting environment, and such endeaver has not been successful in reaching the practical stage yet. This paper presents a cracking extraction system employing a combination of processing methodologies, which comprises of a process to enhance the linear continuity of cracking and a pavement surface density compensation process, as a processing algorithm for stable automatic extraction of pavement surface cracking. The following results were obtained through the aforementioned set of processes. 1) The linear continuity of cracking could be enhanced and clearer cracking skeltal structure was obtained. 2) The fluctuation of the intensity of reflected light caused by uneven lighting and surface conditions could be compensated. 3) Moreover, a process to extract cracking lines stabely could be established.
A contour to grid conversion method is proposed. In the proposed method, Euclidean distance transformation is used to determine the nearest points on the contour with which given point is to be interpolated. By using this approach, complicated searching process of standard points for interpolation could be avoided, which results in simplified and stabilized algorithm. Also, the method can be applied to the contours which are not completely closed. Hence, the burden for restoration of contour can be reduced. The conversion accuracy is evaluated by comparing the result with human-interpreted grid data. An average error for closed contours is 37cm when the elevation interval for contour is 10m, which is thought to be acceptable for many applications. Even in the case of not completely closed contours, the average error is 38cm, which is negligible degradation of accuracy compared to completely closed case.
The relative land surface temperature distribution can be recognized by using thermal imageries, which obtained by Landsat, NOAA and GMS. These thermal informations in the wide area, have been used for some thermal phenomena investigations, such as heart-island, thermal belt. However it is necessary more detailed' time dimensional and spatial information in order to exclaim some factors of these phenomena. This paper describes on the preliminary result of an application of multi-temporal thermal imageries for the thermal environment analysis. In the site investigation of this study, thermal imagery observation was performed through full day. The time dimensional change of thermal environment was investigated by the relationships between land surface temperature and land cover condition, height above the sea level, atmospheric phenomena. Some of typical daily change patterns in vegetation and sub-urban area, were recognized as the result of the data analysis.