Sub-pixel classification is a tough issue in remote sensing field. Although many kinds of software or its Module can be used to address this problem, their rationale, algorithms and methodologies are different, resulting in different use of different method for different purpose. This makes many users feel confused when they want to detect mixed feature content within a pixel and to use sub-pixel approach for practical application. It is necessary to make an in-depth comparison study for different sub-pixel methods in order for RS&GIS users to choose proper sub-pixel methods for their specific applications. After reviewing the basic theories and methods in dealing with sub-pixels, this paper made an introductory analysis to their principles, algorithms, parameters and computing process of three sub-pixel calculation methods, or Linear Unmixing in platform ILWIS3.0, Erdas8.5's Sub-pixel Classifier, eCognition3.0's Nearest Neighbor. A case study of three sub-pixel methods was then made of flood monitoring in Poyang Lake region of P.R.China with image data of band-1 and band-2 of NOAA AVHRR image. Finally, a theoretic, technological and practical comparison study was made of these three sub-pixel methods in aspects of the basic principles, the parameters to be set, the suitable application fields and their respective use limitation. Opinions and comments were presented in the end on the use of the sub-pixel calculation results of these three methods in a hope to provide some reference to future sub-pixel application study for the researchers in interest.
Flood vulnerabilities of various flooded entities are dependent on flooding process, e.g. time period, flooding intensity, spatial and temporal distribution, especially for growing crops affected. This information can not be obtained by classical methods (like field survey, middle to high-resolution images and hydraulic models) due to factors of laboring, cost, complicacy and accuracy. It is necessary to develop a new approach of lower cost and reasonable accuracy to gain our ends. This paper made an experimental study of linear unmixing method to try to approximate our goal with easy-to-get,
high frequent revisit, though low spatial resolution NOAA AVHRR image in Poyang lake region of Jiangxi province, P.R.China. This region is located at north Jiangxi province; its low flat plain topography has very little multiplicative spectral reflections, and is suitable for linear unmixing application. After analyzing the histograms of NOAA AVHRR images, we decided to use the difference image of band-2, band-1 and minus band-2 to differentiate water from others. After application of linear unmixing, we assumed area fraction of water in each pixel is approximately equal to calculated spectral fraction of water in each pixel (this is virtually true when there is little multiplicative reflection and refraction), then fine geo-referencing and spatial assignment were made based on the relation between the element (water) to be spatially assigned and other high-resolution thematic factor DEM. This study provide a new Earth Observation
approach for continuously monitoring distribution change of relatively homogeneous large-scale features, like waters, desert, oil spillage extension etc.
In this paper, a new approach for unsupervised change-detection using multitemporal InSAR data is proposed, of which the significant characteristics is joint use of backscattering temporal intensity and long-term coherence based on 2-D (two dimensional) Renyi's entropy. The proposed approach is made up of two steps: feature extraction and unsupervised 2-D thresholding. In the first step, two features are based on the concepts of backscattering intensity variation and long-term coherence variation respectively, and are defined according to the analysis of different signal behavior of interferometric SAR in the presence of land-cover classes within urban area. In the second step, an unsupervised 2-D thresholding technique based on maximum 2-D Renyi's entropy criterion is developed. The thresholding is performed on the two difference images derived from the two features to produce an accurate change-detection map with two classes: changed and no-changed. Primary experimental results, which were obtained from a set of six multitemporal ERS-1/2 SAR images within Shanghai city of China, show the effective of the proposed approach and that ERS-1/2 InSAR data could be exploited for detecting urban land-cover changes.