CLASSIFICATION ACCURACY OF SAR IMAGES FOR VARIOUS LAND COVERS

V. Baby Vennila, R. K. Gnanamurthy

Abstract: In recent times, the mixture of K-means clustering and Artificial Neural Network Classifier (ANN) has been often and productively useful to image categorization owing to their balancing result. Though, the amount of lessons is typically wanted to be allocated physically. This correspondence presents a disciplined unconfirmed semantic categorization technique for sky-scraping declaration protectorate descriptions. We insert tag price, which can punish a answer footed on a set of labels that come into view in it by optimization of power, to the chance meadows of dormant matters, and an iterative algorithm is thereby planned to create the quantity of lessons lastly become together to an suitable height. Evaluated with additional declared categorization procedures, our technique not only can get hold of precise semantic segmentation consequences by superior level arrangements but also can mechanically allocate the quantity of fragments. The investigational consequences on a number of prospects have established its efficiency and forcefulness.

Keywords: Label cost, K-means clustering, satellite images, Artificial Neural Network classifier (ANN).

DOI: https://doi.org/10.15623/ijret.2014.0319011