An improved content based image retrieval system using wrapper based curvelet transforms and chemical reaction optimization

Abstract

Content Based Image Retrieval (CBIR) which is also known as Query By Image Content (QBIC) has become an active research motivated by the need to search the exponentially increasing space of image and video databases and finds wide range of applications like crime prevention, geographical information and remote sensing systems, medical diagnosis, face findings and so on. CBIR intends to search digital images in huge databases by means the contents of the image such as colors, shapes, textures, or any other information that can be derived from the image rather than the metadata. In the present digital era, there is a critical need to develop robust and efficient techniques to address the challenges of image retrieval from large databases and to retrieve images rapidly. The scope of this research work is bounded to address two major challenges, Feature extraction and optimal feature selection. Feature extraction plays vital role to extract the visual features present in the images. There are many features hidden in a image such as color, shape, texture and so on. The Texture visual features are very important to predict a particular image in the process of image retrieval. There are many texture feature retrieval algorithms are exists. These algorithms can handle linear and rectangular images which is not sufficient to extract complete texture features in an image. To overcome this problem wrapper based curvelet transforms are used to handle both non-linear and curved images effectively. The wrapper based curvelet transform is found to be more robust as well as fasted in the time of computation than that of the ridgelet. The computation performed with correct features may lead to reduce the semantic gap which may occur during the result presentation newline