Natural and Artificial Object Recognition on Satellite Images Through Level Set Evolution and KNN Classification

Extraction of objects from image seems one of the desired and important steps in image processing. There are many areas where object extraction from image is playing a vital role. To work with satellite images for cropping some useful object body is very much desired motive in mapping and surveying area. Many improvements have been done in this area and some are in progress. The past work has been done for section same kind of items and some demonstrations have been done for enhancing its productivity, effectiveness and its efficiency. We are classifying both kinds of objects natural or artificial objects from satellite image. Most researchers have used Level set evolution (LSE) for clipping out artificial objects from remote sensing images. Level set evolution (LSE) is also gives effective outcomes for clipping physiographical changes and it gives a remarkable outcomes. We have used some geometrical features and texture features for feature extraction and K-nearest neighbor classification for classification of artificial and natural objects which gave us better output performance and we calculated precision, recall, accuracy and finalized our paper with some important conclusions and appropriate results.