Bayesian classification and class area estimation of satellite images using stratification

Abstract

The paper describes an iterative extension to maximum a posteriori (MAP) supervised classification methods. A posteriori probabilities per class are used for classification as well as to obtain class area estimates. From these, an updated set of prior probabilities is calculated and used in the next iteration. The process converges to statistically correct area estimates.The iterative process can be combined effectively with a stratification of the image, which is made on the basis of additional map data. Moreover, it relies on the sample sets being representative. Therefore, the method is shown to be well applicable in combination with an existing GIS.The paper gives a description of the procedure and provides a mathematical foundation. An example is presented to distinguish residential, industrial, and greenhouse classes. A significant improvement of the classification was obtained.

Multivariate texture‐based segmentation of remotely sensed imagery for extraction of objects and their uncertainty

“…Split-and-merge segmentation consists of a region-splitting phase and an agglomerative clustering (merging) phase (Haralick and Shapiro 1985, Horowitz and Pavlidis 1976, Gorte and Stein 1998, Lucieer and Stein 2002, Lucieer et al 2004. Supervised segmentation uses explicit knowledge about the study area to train the segmentation algorithm on reference textures.…”

“…Segmentation techniques extract spatial objects from an image (Gorte andStein 1998, Lucieer andStein 2002). It extends classification, as spatial contiguity is an explicit goal of segmentation whereas it is only implicit in classification.…”

Land Cover Classification of Remotely Sensed Satellite Data using Bayesian and Hybrid classifier

“…If the data are complex in structure, then to model the data in an appropriate way can become a real problem. [6] have used Bayesian classification for class area estimation of satellite images using stratification. LAU C.C.…”

Spatial statistical modelling of urban particulate matter

“…Gorte and Stein (1998) showed that higher overall accuracy can be achieved by localized k -nearest neighbour estimation of the probability density P (x p |C i ).…”

“…In each training region 14 areas were selected representatively in terms of land cover classes and variations in soil types. From each training area, 20-30 pixels were selected resulting in a training sample size that was within the recommended range for maximum likelihood classification (Gorte and Stein, 1998). The training samples were the basis for signature development for the ML classifier and for determining NDVI thresholds for the bare ground, water and vegetation classes.…”