Bringing High Level Meanings from an Image

Abstract

Content-based image retrieval is a difficult area of research in mult imedia systems. The research has proved extremely difficult because of the inherent problems in proper auto mated analysis and feature extract ion of the image to facilitate proper classificat ion of various objects. To segment the image to ext ract mean ingful objects and then classify it in high-level like table, chair, car and so on has become a challenge to the researchers in the field. The gap between low-level features like color, shape, texture, spatial relat ionships and high-level definit ions of the images is called the semantic gap. It is very important we find a viable solution of how to ext ract meaningful definition of an image fro m low level features so that we can identify the image automatically without any human intervention. As we know billions of images are being generated fro m various sources all over and it's extremely time consuming and expensive to identify these images manually. When we can identify these vast number of images automatically without human intervention then we can classify them and create databases based on these classifications. These databases could then be used to enhance mach ine learning of artificial intelligence. Until we solve these problems in an effective way, the efficient processing and retrieval of info rmation fro m images will be difficu lt to achieve. In this paper we explore the possibilities of how we can ext ract high-level meanings fro m an image before or after the seg mentation of the image in an automat ized way. We attempt to use a database of data dictionary for the purpose which we believe has the potential for solving the problem of semantic gap in content-based image ret rieval.