The Prediction of the rice crop area using satellite image extraction models with grey level co-occurrence matrix, neural network, and fuzzy logic
Chairuddin, Sony Susanto, Hendra Gunawan, Asto Purwanto, Ketut Wikantika

STMIK IM, ITB

Abstract

The advances in remote sensing technology can provide an alternative for policymakers to obtain geospatial information based on facts. The Information generated from the technology is very dependent on the image resolution and the extraction processes. The image extraction process in this research carried out to predict the harvested area in the distribution of paddy fields through the identification of texture features based on the growth phase. The image is the multi-temporal satellite image taken for four months.

The extraction system model uses the integrated approach of the grey level co-occurrence matrix (GLCM) method and the concept of Haralick features and the concept of artificial intelligence, namely, artificial neural networks and cryptic logic GLCM and Haralick feature-used to form, texture patterns based on the characteristics of each pixel, which corresponds to the growth phase of rice plants. Artificial intelligence models are used to recognize patterns, segmentation, and classify rice plants from their growth phase, so that the distribution of the harvest phase can be known.

The accuracy of the system model-validated by comparing the data on the distribution of plant to harvest area produced by the extraction system with data from the Department of Agriculture. The comparison results show some differences, with relatively low accuracy. The resulting system model implemented to monitor the stage of growth of rice plants to predict the harvested area in an area.

Keywords: GLCM, Haralick, artificial intelligence, texture, classification

Topic: Computer and Mathematics