The satellite image classification is one of the important application of remote sensing data.However, the vegetations (such as tree, grass and rice paddy) have very similar spectrum response insatellite images. This spectrum effect will caused to difficult classify each categories by using onlymulitspectral data. To resolve this problem, this study used to the semivariogram theory (textureinformation) to extracted each vegetations feature of the QuickBird image.This research used the Direct-Semivariogram and Cross-Semivariogram to calculate imagetexture. Further, variogram analysis was performed on Quickbird data to determine the nature ofspatial dependence with spectral reflectance for the selected vegetations of land cover systems whichis estimated by the mean of the ranges in a series of variograms. On the other hand, this study alsodiscussion the texture factor by the variogram analysis (such as window size, choose direction andband combination). Finally, the results showed semivariogram texture information which caneffectively improve the classification accuracy in high resolution satellite image.

The major applications of the remote sensing images include the detection of natural resources and the monitoring for geoenvironment. In order to acquire largest possible coverage, the overlapping area of satellite images is small in general. In addition, the converging geometry is poor in common. Hence, the traditional 3-D bundle adjustment is not suitable for orientation modeling directly. The objective of this investigation is to propose a modified block adjustment procedure using DEMs as elevation control for multi-orbit satellite images. The major works of the proposed scheme are: (1) bundle adjustment using DEM as an elevation control, (2) generation of the orthophotos , (3) image mosaicking. Experimental results indicate that block adjustment can reduce the discrepancies for conjugate points between image strips. It is also demonstrated that the mosaicked image is better seamed when tie points are employed in the adjustment.

When detecting changes in panchromatic images taken at different points in time, it is customary to analyze the difference between two images. Areas with little or no change have zero or low absolute values, and areas with large changes have large absolute values in the difference image. If image data gives more than two channels, it is difficult to visualize changes in all channels simultaneously. To solve this problem and to collect information on change, linear transformations of the image data can be considered. Traditionally, we make linear transformation by using principal component analysis via the covariance matrix of difference between two images. In this study, we perform linear transformation by applying multivariate alteration detection (MAD) by cross-matrix between two images. The property of the multivariate alteration detection transformation is the linear scale invariance. So, if we use MAD, preprocessing by linear radiometric normalization is superfluous. To detect the change areas by Chi-Square test, and the major changes is directly related to target changes, not seasonal or atmospheric effects. Results verify the effectiveness of the MAD method for change detection of multi spectral images.

The integration of multi-scale digital maps has recently become an important task because of the popularity of web-GIS and map-related mobile platform. In order to produce different scales of maps from the vector data, numerous map generalization techniques have been developed to automatically generalize the vector data. An ideal generalization technique not only decreases the number of data points but also retains the similarity of the simplified shape to the original ones as close as possible. This paper proposed a new generalization approach; which uses the curvature and the shape distortion as the controlling factors to implement the task. The proposed method classifies the data points as the primary points and the secondary points. Since the primary points are the points that represent the distinctive features of the shape, therefore, if the generalization only accepts the primary points that may result in an over simplification. Hence, it’s necessary to extract the secondary points to compensate the over-simplified situation. In this study, a Shape Distortion Index (SDI) is developed to detect the secondary points that can reduce the degree of the shape distortion efficiently. Two case studies are carried out to compare the proposed method with the commonly used Douglas-Peucker algorithm. The comparisons show that the proposed method has better-simplified results and less shape distortion both perceptually and quantitatively than Douglas-Peucker method.

In this paper, we compared image fusion of optical and radar images by pixel-level image fusion based on Principle Component Analysis (PCA), and feature-level image fusion based on Dempster-Shafer evidence theory. We combined the HV polarization information from SAR and spectral characteristic from SPOT images in an effort to enhance land cover classification. Before the fusion process, wavelet transform was first applied to denoise the SAR image which suffers from speckle contamination due to coherent process. PCA method is then used to fuse the SPOT and SAR images. In so doing, the PC-1 component is replaced by SAR image (approximation image, after wavelet transform) and then the inverse transform is followed. At last, the maximum likelihood classifier was used for both SPOT-XS images and fusion images. In feature-level case, fully polarization information from SAR is used to combine with spectral characteristic from SPOT images, mainly to enhance land cover classification as well. Speckle noise was removed by Lee filter, followed by the maximum likelihood classifier based on different distribution was used for SAR and SPOT images (Based on Wishart distribution and multivariate Gaussian distribution respectively), to extract the conditional probability of each pixel for each class. Finally, Dempster-Shafer evidence theory is then applied, to combine the classified results of SAR and SPOT data. Experimental results show that the classification accuracy is dramatically improved by effective image fusion of SPOT and SAR data. Excellent results were obtained by the proposed method.

In the general process, GPS is used for the positioning in the airborne lidar operation. Therefore, most frequently the height obtained is referenced to WGS84 ellipsoid and the measurements are in ellipsoidal height. In order to produce the digital elevation model in the orthometric height, the geoid undulation correction should be performed. This paper investigates the differences of five currently available digital geoid models of Taiwan area as applied to the airborne lidar point cloud obtained for the Kao-Hsiung and Ping-Tung area. Based on the differences between the result from the elevation obtained in the field and the elevation interpolated from the Lidar point clouds, the ellipsoid height are examined. The standard deviation for the ellipsoid heights is 0.172 m. Then, the orthometric height differences among those obtained from different models are compared. The standard deviation for Hwang-2005 is 0.174 m, and 0.199 m for Hwang-2001. Due to the lack of directly observed orthometric heights, the real error in terms of orthometric height is still difficult to confirm.

Three-dimensional (3D) building model is one of the most important components in a cyber city implementation and application. A realistic texture mapping of building models not only supplies authentic appearance of the models in visualization, it can also provide useful and sometimes critical features in complex applications. However, currently, most building models do not have sufficient and accurate texture information. In order to provide accurate texture attributes to 3D building models, this study developed a highly automated facade texture generation and mapping system for 3D building modeling. Seamless and photo-realistic texture mosaics of building facades were generated from video sequences acquired with consumer digital cameras and digital video (DV) camcorders. Extracted frame images were registered using a developed algorithm based on interest points identified semi-automatically to address concerns on the geometric alignment. A polygon-based blending algorithm based on alpha blending was then applied to integrate colors and shadings over the overlapped regions of adjacent images. As a result, the produced facade textures were continuous in both geometric outlines and color domains. Occlusions of generated facade texture mosaics were automatically identified and mended through a series of morphological operations. The corrected facade textures were mapped onto corresponding building facets correctly and efficiently using linear or parametric transformations. Test experiments conducted in this study demonstrated that the resultant building models had more complete and accurate texture attributes as well as near photo-realistic appearances. More importantly, comparing to conventional photo-based building texture mapping techniques, the developed algorithms are highly automated and provide a great advantage in large-scale cyber city implementations and visualizations.

The simulated broadband data of satellites SPOT 5 and FORMOSAT-2 were mimicked from canopy high-resolution reflectance spectra measured near ground, by a field-portable spectroradiometer, and these spectral data were then used to compare the differences in growth estimation and yield prediction of rice crop relative to those of ground measurements. Reflectance of broadband sensors was acquired by using the mean value of reflectance averaging over the respective waveband regions of hyperspectral data. Results indicated that the calculated values of normalized difference vegetation index (NDVIs) from hyperspectral (i.e. NDVINB) and simulated satellites data (i.e., NDVISPOT 5 and NDVIFORMOSAT-2) were all nonlinearly distributed during the first and second cropping seasons of rice. The curves were nearly parallel to each other, but values of NDVISPOT 5 and NDVIFORMOSAT-2 were lower than those of NDVINB. The relationship between the measured leaf area index (LAImeasured) and NDVINB was best fitted to an exponential growth function, by which the estimated values of LAINB, LAISPOT 5 and LAIFORMOSAT-2 were obtained using the respective NDVIs as inputs. Values of LAINB were found higher than those of LAISPOT 5 and LAIFORMOSAT-2 on the same sampling dates yet all the estimates were lower than the corresponding values of LAImeasured after ca. 55 days after transplanting. As the simulated satellites data obtained lower values of NDVI and LAI than those of NDVINB and LAINB counterparts, yields predicted from satellites data were less than yield predicted from ground-based hyperspectral data. These predicted yields were all lower than the measured yields.