Adaptive invariant content based Image retrieval using semantic Learning

Abstract

With the extensive growth of image databases in the digital world newlineefficient image retrieval has gained much importance Content Based Image newlineRetrieval CBIR is the retrieval of relevant images from the image database newlinebased on the visual content of the image which consists of the low level newlinevisual features such as color texture and shape The performance of most of newlinethe CBIR systems is limited by the low level visual features since they newlinecannot adequately express the high level human perception This is the newline semantic gap problem To reduce the semantic gap semantic knowledge newlineincorporation into the content based image retrieval system has become very newlineimportant This is known as semantic content based image retrieval SCBIR newlineMany recent research works are going on in the field of SCBIR newlineThis research work aims to develop a semantic learning scheme newlinefor predicting the semantic concept of the query image The support vector newlinemachine SVM a supervised machine learning algorithm is used in the newlineconstruction of the semantic learning scheme The SVM binary decision tree newline SVM BDT is constructed using the semantic templates of each of the newlinesemantic concept Using the low level features of the query image the newlineSVM BDT predicts the semantic concept of the query image newline newline