Contribuţii privind clasificarea imaginilor în funcţie de conţinut (content based image retrieval)

Abstract

Research field Definition: In the fields of electrical engineering and computer science, the image processing concept is composed of any signal processing technique, where the input is represented by an image (a standalone picture or a snapshot taken from a video sequence) and the output is represented by a different image or by a set of characteristics or parameters which describe it. Most of the image processing algorithms treat the image as a bi-dimensional signal on which they apply standard signal processing techniques. Definition: The Content-Based Image Retrieval – CBIR concept is also known as Query by Image Content (QBIC) or Content-Based Visual Information Retrieval (CBVIR) and represents a solution to the problem of searching digital images in a large database. The "Content-Based" term implies that the CBIR techniques use the real image content, as opposed to the information extracted from the metadata fields, like key words, labels, tags or any other form of description associated to the image. The image processing techniques are mainly used in two distinct areas:  Improving the image quality in order for it to be observed and analyzed by a human person. We are mentioning in this category the image printing and image transmission fields;  Image analysis – in order to obtain a set of characteristics and structures. The image decomposition algorithms are usually based on techniques which are targeting edge detection, unique light intensity, texture or colour recognition, or a combination of all. Shape is a more difficult concept but there are some methods to describe it mathematically. All these techniques are used to classify or recognize objects; this paper is located in this area, in the general concept of CBIR.