Content Based Image Retrieval and Classification Using Principal Component Analysis

Roshani Mandavi, Kapil Kumar Nagwanshi

Abstract: Content based image retrieval (CBIR) systems has become more popular nowadays in many applications. The main issue in content-based image retrieval system (CBIR) is to extract the image features that represent the image contents in a database. Such an extraction requires a detailed analysis of retrieval performance of image features. In this research we will propose a retrieval system which uses color, texture and shape features of an image that represent the contents of image. Proposed system extract the color, texture and shape feature using color moment, grey-level co-occurrence matrix (GLCM) and Fourier descriptors respectively. After the features are selected, a PCA based classifier classifies an image based on trained feature database and results of this system retrieves the relevant images. It will propose an efficient retrieval system which is able to select the most relevant features to analyze new encountered images thereby improving the retrieval efficiency and accuracy.

Keywords: CBIR content based image retrieval, Feature extraction, Color moment, Grey level co-occurrence matrix GLCM, Fourier descriptor, PCA principal component analysis, Feature similarity