Content based image reference apparatus and method for relevance feedback using fussy integral

摘要

PURPOSE: A contents based image searching system and method using fuzzy integral are provided to automatically calculate significance of image characteristics by reflecting a user's opinion in the calculation to improve searched results. CONSTITUTION: A contents based image searching system includes an image input unit(10) for receiving an image from a user, a characteristic extractor(20) for extracting image characteristics required for calculating image similarity from the input image, and a small image generator(60) for dividing the input image to generate small images. The system further includes a similarity calculator(40) for calculating similarity between a selected inquiry image and a DB image by characteristics using the image characteristics, and a similarity measurement unit(50) for integrating the similarities using fuzzy integral to measure the finial image similarity. The system also has an inquiry processor(80) for interpreting an inquiry from a user to measure similarity and sequentially retrieve the small images according to the measured similarity to transmit the images, and a user interface(90) for reflecting a user's opinion in the image data transmitted from the inquiry processor and delivering information selected according to the result of reflection of the user's opinion to the inquiry processor.