tailieunhanh - Coconut inventory and mapping using object oriented classification

Object Based classification involves grouping of pixels based on spatial relationships like similar colour, tone, Shape with surrounding pixels. Object-based information extraction depends on spectrum character and geometry and structure information. Object based classification interprets an image that is represented not only by a single pixels, but also in meaningful image objects and their mutual relationships. In this study has been attempted mapping of coconut growing areas for Kozhikode taluk, and LISS-IV data was used in this study for classification of coconut using Object-oriented classification techniques. Main objective of this study is Comparison of different classifiers for better accuracy. Cartosat data gave the spatial information of object and LISS-IV data gave the spectral information of the object. Multi resolution Segmentation process was performed for classification. Multi resolution Segmentation is nothing but images subdivide into separate regions based on the spatial and spectral heterogeneity. Using eCognition software as the platform, this study carries two kinds of supervised classification and Rule based classification. Methodology used in this study was SVM Classifier (Support Vector Machine), KNN classifier (K-Nearest Neighbour) and Rule based classification. Parameters used in the rule set was NDVI, Maximum Difference, Brightness, Mean, standard deviation, Asymmetry, shape index. Kozhikode had a scattered settlement so there was a chance to settlements can be classified under the coconut classification. Young plantations were difficult to classify and Inter crops or mixed crops like arecanut also been a big challenge. | Coconut inventory and mapping using object oriented classification