Abstract: The existing methods of the target searching in remote sensing image database are discussed, with their defect in mind, and a target’s matching method, from “roughness” to “fine” is presented. The primary target location is found by the use of direct two values correlation method based on edge features, and with the invariant moments and statistical features of the image histogram, combined with similarity measurement, the target can be found quickly. Experimental results show that with high matching precision and rapid matching speed, this method can meet the need of rapid target searching in remote sensing image database.

Key words: remote sensing image；histogram, invariant moment, statistical feature；similarity, matching