Cassava Planted Area Classification Using Object Based Image Analysis and Decision Tree : Case Study of Mahasarakham Province, Thailand

Abstract

Classification cassava planted area using LANDSAT 8 from satellite data with technical analysis, object-oriented and decision tree: case studies Mahasarakham province. The study aims to identify the cassava planted areas with technical analysis, object-oriented data and decision tree classification. Researchers have taken techniques used in object-oriented image classification. By dividing the image data (Segmentation) before distribution to group the data. Another factor in the segmentation data including Scale parameter Shape parameter and Compactness parameter to image objects corresponding to the object type. The average value of the object that was used to classify. The results of the research can be segmented image data 11,800 objects by the classified the vegetation has 8,783 objects or 3,042,100 rais and a series of plants to identify plant cassava under the classification in groups. Various conditions of each classification. The end result is a process that in ordered by 4th of classified. Cassava has classify by the object has 2,998 objects or 2,168,112 rais, representing 65 percent amount of area in the province. To check the accuracy of classification can determine where the field survey of 60 points on the object image classification to determine that the object image falls on Yucca 48 basis points to 80 percent, which is the result. The research was necessary on this occasion can be applied in the management of farmland. Managing
agricultural zoning and about the area used for agricultural cultivation. And can predict the amount of acreage and yield by using technology to help explore. Reduce the cost of space exploration actually. It can also be integrated on track in terms of providing compensation to farmers or to register more easily with one
another.