Efficient Content Based Image and Video Retrieval System Using Truncation Coding and Multi Resolution Features

Abstract

In this contemporary world, information technology has created a huge impact. The progress in internet began from textual information to a vast and growing collection of images and videos. Due to the quick advancement in technologies, the efficient retrieval of data from large databases have become very difficult. newlineThe considerable increase in the size of data and the size of the storage capability of the database has made it a challenging task for the user to retrieve the appropriate data required. The huge data collections may have millions of videos, images and terabytes of data which makes it very difficult for the user to retrieve the required data. For users to make effective usage of the data stored, efficient searching methods have to be developed. Many Internet clients encountering the proficiency and reliability given by web search engines for example, Google would find it puzzling that current visual retrieval performance is very poor in evaluation to text retrieval. Indeed, text-based search engines have verified to be ineffective in navigating the Web. But, when it comes to visual content search, results seldom match expectations. Visual documents reflect semantic data, but the information is not systematized into a semantic structure. Furthermore, in the case of video, outside the level of frame and frame sequences, the structure is mostly variable and sometimes ambiguous. newlineThe proposed system deals with effective methods for image and video retrieval from a large database which focus on content based image and video retrieval. It is designed using three methods namely Content Based Image Retrieval (CBIR) system using Ordered Dither Bit Truncation Coding (ODBTC) and contourlet features, CBIR System based on Color-Texture Features and Color-Texture Based Feature Modeling for Content Based Video RetrievaL (CBVR). CBIR system utilizes visual contents of the image portrayed as low level features like shading, surface, shape and spatial areas used for image representation in a database. A novel content-