Expression based image retrieval using RSCM

Abstract

In recent years, image retrieval has much impact on handhold mobile applications to high-end database processing. Text and content based image retrieval are the common methods in existence. Though it made lot of changes, retrieving human expressions is challenging one for the recent researchers. We introducing a new technique called Ratio of Sum to Centre and Maximum pixel (RSCM) feature which leads to communicate with machines via facial expressions. In the proposed system, first the images in the database get pre-processed and this image can be given to the gabor filter to segment the image as the gabor wavelet blocks. The output of the gabor filter given to the RSCM feature extraction algorithm to extract various expression based features from the image and Artificial Neural Network (ANN) will train the features. Eventually, the comparative analysis performed using the commonly used methods namely precision and recall were clearly shown that the proposed system provide better results than the conventional systems and retrieve the images based upon the expression.