Machine learning techniques have been widely used to classify spatial objects and it has become one of the well-established research areas of computer science. The identification of residential, nonresidential areas, cultivated land, non-cultivated land, forest and others in a real time image is a significant problem. Due to the presence of complex patterns of dense forests and residential areas in forests is one of the difficult classification problems because heterozygous image morphologies are very difficult to preprocess. The proposed approach is very useful during natural disaster such as fire in forest, earthquake, flood etc. This paper has twofold object, one is to classify the forest, residential areas and so on whereas the other is to classify the cultivated and non-cultivated areas within the real time picture. In order to mitigate all above stated problems, this research proposes a methodology where in first stage, we prepare the data; in second stage, the Decision Model is constructed using Bayesian Belief Neural Networks. About 98.36% accuracy of the system is measured by using Confusion Matrix.
Volume 12 | Issue 8
Pages: 437-443
DOI: 10.5373/JARDCS/V12I8/20202603