METHOD AND APPARATUS FOR CONTENT-BASED IMAGE RETRIEVAL WHICH IS ROBUST AGAINST RESIZING, ROTATION, AND CROPPING

摘要

PURPOSE: A method and apparatus for content-based image retrieval are provided to increase the accuracy of searching for images by searching for a specific region of an image based on the content of an image and extracting a characteristic vector from the specific region. CONSTITUTION: If a search image is inputted, the search image is decoded and decompressed(S2). The size of the decoded search image is normalized to a certain size(S3). The specific region brighter or darker than surroundings is extracted by searching the normalized search image(S4). The fingerprint vector of the specific region is extracted(S5). The fingerprint vector is compared with the fingerprint vector of the image saved in the video information database, and the similar fingerprint vectors are matched each other. The similarity is identified based on the rate of the matched fingerprint vectors(S6).