Satellite Image retrival based on sensitive content method

Abstract

-The satellite cloud image is a valuable source of information in weather forecasting and early prediction of different atmospheric disturbances such as typhoons, hurricanes etc. Due to the increased number and resolutions of the Earth imaging sensors and image acquisition techniques, the satellite image data is growing enormously which makes it difficult to store and manage. The traditional image retrieval technique is inefficient in retrieving these images. Content-based image retrieval is an approach from data mining community which provides the solution of managing this huge quantity of data. In this research, a Content-Based Image Retrieval (CBIR) system has been applied on Geospatial Images of fire and forest, Clutter and water, cyclone and water etc. Geospatial images are processed using K-means clustering algorithms to obtain a highdimensional feature vector. The Feature vectors include HSV Histogram, LAB features, color autocorrelation, color moments, Gabor features. Then Train a KNN classifier using those features using different distance metrics. The images and the extracted feature vectors are stored in the database. Distance metric is used to compute the similarity between the images. The system is robust as it provides search based on the multiple features. The performance of the system was evaluated by analyzing the retrieval results using precision. Many past result was evaluated and based on that results and method the aim was to find the best outcome among all.