Paper Title
Mining Of Overhead Images From Large Database Using Feature Extraction Methods

Abstract
Abstract: This paper is based on the mining of overhead images from the large database using input image. Local feature extractions are particularly well suited for the newer generations of aerial and satellite imagery whose increased spatial resolution. Feature extraction have been successfully applied particularly for challenging tasks such as detection and classification. We can also perform an extensive evaluation of local invariant features for high resolution image mining of land-use/land-cover (LULC) aerial imagery.Scale Invariant Feature Transform (SIFT) is used to matching the image in the large database. Bag-of-visual-words (BOVW) representation like saliency- versus grid-based local feature extraction on the basis of a number of design parameters. BOVW representation is used to measurethe size of the visual codebook, the clustering algorithm used to create the codebook andcompare the dissimilarity. We perform comparisons with standard features such as color and texture. We can also describe interesting findings such as the performance-efficiency tradeoffs that are possible through the appropriate pairings of different-sized codebooks and dissimilarity measures in addition to reporting on the basis of the core design parameters. While the focus is on image mining, we expect our insights to be informative for other applications such as detection and classification.