Technical Studies, Region-based Multi-source Remote Sensing Image Database Retrieval

The remote sensing image (RSI) retrieval system has been a hot research area recently with an increasing huge amount of image data generated during the two decades. This paper studied the existing technology for image retrieval, then focused on the following aspects:(1) Expression model and description method for color, texture are researched base on theoretic and method of feature-based image retrieval technology. We gain the cnaracteristic of color, texture feature through the experiment of color, texture feature-based image retrieval.(2) We studied the technology for huge image data processing, then present two computation method. We gain some helpful conclusions after the experiment.(3) We studied the fast corner-point detected algorithm be suitable for huge image data. Then this paper presents the new edge-based , side length-constrained algor(?)thm for detecting the corner-point.(4) In this paper, we present a retrieval method base on region for the huge remote sensing image database . We call the significative regions for vision ROI (Region of Interest), which have a affluent of feature information, These regions distributing proportionally. We extraction color, texture, shape features from the ROI and make them eigenvector then access them on the region features database. We call the amount of matching ROI D00 (Degree of Overlapping) in retrieval procedure.