Automatic Feature Extraction for CBIR and Image Annotation Applications

Abstract

In the area of information technology, organizing and indexing of the digital information is a primary concern. In CBIR system, one of the most significant issues is a semantic gap. Semantic gap refers to the difference between the features extracted from image and interpretation of features content in the image or within the regions by human. Hence, automatic image annotation has achieved momentum in the last few years. The objective of the automatic image annotation (AIA) is to allocate textual labels to the image that clearly describes content or objects in the image. Accuracy of the automated image annotation algorithm depends upon the feature extraction process. Therefore, effective feature extraction algorithm is essential. In this paper, feature extraction algorithm using Gabor filter is presented. Gabor filter through its multi-resolution capability successfully extracts effective features from images or regions obtained after segmentation. It is demonstrated that the Gabor filter generates low level, less number of features and accurate description of the image if filter with frequency response in the band of 50–75% of the total frequency is selected. These extracted features further reduce the complexity in the classification algorithms developed using statistical models or soft computing techniques.