SYSTEM AND METHOD FOR IMAGE IDENTIFIER BASED ON SINGULAR VALUE DECOMPOSITION AND FEATURE POINT

摘要

A system and a method for comparing SVD(Singular Value Decomposition)-based images are provided to search and recognize a still image easily and quickly which has been illegally changed due to copying or intentional attack or a still image whose original characteristics have been changed. A feature extracting module(110) extracts one or more features of an inputted still image by using strength of Gaussian curvature of each pixel from the still image. An identifier detecting module(120) detects a local identifier based on an SVD with respect to a neighbor block of the extracted features. A matching module(20) discriminates whether a still image has been duplicated by comparing the detected local identifier and a still image identifier database(30) storing one or more still images, The feature extracting module extracts the features by using a Y component of the inputted still image, reduces differentiated noise of the Y component, and extracts the features by measuring strength of the Guassian curvature by using an eigen value of a Hessain matrix.