Content based multispectral image retrieval using principal component analysis

Abstract

In most current image retrieval systems, the retrieval process is performed using similarity strategies applied on certain features in the image. This paper presents a novel method for multispectral image retrieval. The proposed method starts with calculation of two features and then it uses Principal Component Analysis (PCA) to extract principal components of the feature values. Later on, feature values of each image are exhibited by a linear combination of these principal components. In the proposed approach, two effective weight vectors are calculated for each image in the system. These two weight vectors are used efficiently in radiance and texture based retrieval process. The proposed method was performed and tested on a set of LANDSAT multispectral images from variant sceneries. Experimental results show the superior performance of this approach.