Optimization of Pixels with Data Mining and Image Segmentation for Landuse Land Cover Maximum Likelihood Classification Algorithms

Abstract

The image segmentation is a classification issue where every pixel is classified into different groups. A variety of image segmentation methods have been developed for image processing and computer applications of pixels of satellite data. The data mining of data of remote sensing data in which it assumed that optimization of pixels is typically labelled as single land cover and land use class. The pixel level and texture features are selected from the transformed colour image. The pixels are classified the spectral variables and informations. These pixels are classified by two methods of unsupervised and supervised classifiers algorithms. Pixels satellite images are natural grouping of digital value using maximum likelihood and selforganizing data analysis (ISODATA) algorithms. An analyst selects training sample sites with known class types and representative samples. The pixels are labelled by decision rules by their spectral properties with maximum likelihood classifier (MLC) algorithms.