The development of an efficient ground sampling strategy which can sample the natural dynamics of variations in variables of interest, is critical to ensuring the validation of remotely sensed products. This study attempts to take a fresh look at geostatistical methods for ground sampling and pixel-mean estimating in remote sensing validation campaigns. Spatial random sampling (SRS), Block Kriging (BK), and Means of Surface with Non-homogeneity (MSN) were implemented to estimate the fractional vegetation cover mean values at GEVO1 1 km2 pixel level using Landsat 8 OLI and SPOT4 HRVIR1 fine-resolution FVC maps respectively derived from a homogeneous area covered by forest and a heterogeneous area covered by crop. The GEOV1 FVC product was validated using the means estimated by SRS, BK, and MSN. Root square error (RMSE), mean absolute percentage error (MAPE) and product accuracy (PA) were used to evaluate the validation. Results showed that the MSN method performs well for estimating the means of the surface with non-homogeneity, with a high accuracy of the GEOV1 FVC product (RMSE=0.12, MAPE=29.37 and PA= 77.39%). The statistical values outputted by BK were respectively 0.13, 31.46% and 76.21%. These values of SRS were respectively 0.13, 31.16% and 76.10%. For homogeneous surface, the statistical parameters outputted by these three methods were similar. These results revealed that MSN is an effective method for estimating the spatial means for heterogeneous surface and validating remote sensing product. We can conclude that choosing an appropriate sampling method has a significant impact on the validation of remote sensing product.
Damage to vegetation caused by secondary disasters of the Wenchuan earthquake in severely damaged counties was estimated through the visual comparison of SPOT images acquired before the earthquake and ADS40 aerial images acquired after the earthquake, and a series of spatial analyses. In this paper, we (1) interpret 2-meter resolution aerial images that cover areas severely affected by the earthquake, and obtain statistical information on vegetation damage for the counties of Beichuan, Wenchuan, Maoxian, Lixian, Pingwu, Qingchuan, Anxian and Jiangyou; (2) spatially analyze the relationships between vegetation damage and slope gradient and distance from active faults using ArcGIS software to obtain information on vegetation damage under different geologic and geomorphologic settings; and (3) estimate the area of vegetation damage for the whole region using the above results for the areas covered by imagery. The results indicate that (1) farmland and grassland were less damaged than forestland was since they are mostly located on less steep slopes; (2) Wenchuan was the worst damaged county; and (3) the proportion of damage to vegetation first decreased and then increased with increasing distance from the three main faults of the Longmenshan fault zone owing to the combined effects of the three faults and the effects of regional geology and landforms.
KEYWORDS: Probability theory, Roads, Image classification, Computer simulations, Data fusion, Image fusion, Image processing, Error analysis, Data modeling, Monte Carlo methods
There has been substantial effort dedicated to the issue of how to incorporate spatial information to improve the classification accuracy in past decades and some excellent methods have been developed. Each method has its own advantages and disadvantages for different images and user requirements. This paper proposes a new classification method, which introduces multiple-point simulation to improve the classification of remotely sensed imagery data by incorporating structural information through a training image. This new method named CCSSM is the derivation of two classifications and based on spectral and spatial information, which then are fused. For validation purpose, a real-life example of road extraction from Landsat TM is used to substantiate the conceptual arguments. An assessment of the accuracy of the proposed method compared with results using a maximum likelihood classifier shows the overall accuracy improves from 48.9% to 82.6%, and the kappa coefficient improves from 0.12 to 0.55 and therefore, the new method has superior overall performance on the classification of remotely sensed data.
Speckle noise is a common phenomenon in SAR images. The reduction of Speckle is necessary for any further processing of SAR image such as segmentation, classification and other procedures for information extraction. In this paper, after a brief review of conventional filters for SAR speckle reduction, a wavelet-based soft-thresholding filter for SAR speckle reduction is presented. To evaluate the performance of this filter, the adaptive local statistics filters, which include Lee, Frost, Enhanced Lee, Enhanced Frost, Kuan, and the Gamma-Map filter are applied to the speckle reduction for a same typical SAR image. The performances are compared in several aspects including Radiometric preservation, feature preservation, speckle reduction in extended uniform regions and the absence of artifacts. The results show that the wavelet-based speckle reduction filter performs better in every aspect in evaluation than the conventional filters do.
Conference Committee Involvement (1)
International Conference on Earth Observation Data Processing and Analysis
Access to the requested content is limited to institutions that have purchased or subscribe to SPIE eBooks.
You are receiving this notice because your organization may not have SPIE eBooks access.*
*Shibboleth/Open Athens users─please
sign in
to access your institution's subscriptions.
To obtain this item, you may purchase the complete book in print or electronic format on
SPIE.org.
INSTITUTIONAL Select your institution to access the SPIE Digital Library.
PERSONAL Sign in with your SPIE account to access your personal subscriptions or to use specific features such as save to my library, sign up for alerts, save searches, etc.