이미지 검색을 위한 관심영역의 오소추출과 데이터 변환 연구 = A Study of Element Extraction and Data Conversion of Region of Interest for Image Retrieval

다국어 초록 (Multilingual Abstract)

In this paper we designed the content-based image system. It discovers the characteristic value of color area for the existing image, when any user establishes the special interesting region for the question images. And we designed the system to be able to search the similar image of the interesting region, compared with the table that saved the characteristic value of moment which is unchangeable for the size and the rotation of from.
In this paper, we suggest th image search model based on neural network. This is able to seek for the aimed image through the feedback to search the nonlinear relations by th method of searching content-based image using SOM neural network.
We make use of the Precision and the recall that are used for the system evaluation using the similar matching method in general. As a result of the evaluation of the content-based image system, we can reduce the time and memory to find out the feature value of the image that is meaningless in the general images.