Optimization of Naïve Bayes Classifier To Classify Green Open Space Object Based on Google Earth Image

Floods, air pollution, and increasing air temperatures become a visible danger for society in urban areas. These problems are caused by uncontrolled human activities and lead to environmental damage. To resolve this problem is by using computerized facilities that can monitor real environmental condition in the form of green open space. The provision of such facilities is enabled by free and accessible Google earth image. However, the information provided by Google earth only shows satellite photos that cannot be used for objects classification of green open space on the surface of the earth. Therefore, it is necessary to develop methods to classify objects of earth's surface, particularly in urban area, using free and available data from Google earth. The case study of this research includes several cities in Indonesia, while the method employed to classify image-based green open space was Naïve Bayes classifier (NBC). In this paper, training and testing are among two stages used in classifying green open space. Training process is to construct a new structure of NBC model which involves several NBC models. While testing process is to classify green open space based on the structure. The experiment results using the test sample show that the accuracy of the new structure of NBC model in the green open space classification is better than the NBC single model.