Skip to Main Content
This paper proposes Geometric features by which shape and margin characteristics of objects present in images can be extracted. These extracted features are being represented as feature vector of an object and used for defining semantic of an image. These features are invariant to rotation and effectively shape and margin of the objects are defined. The feature are extracted for all the images in the image database. For a query image the feature is extracted and compared with the database images using Manhattan similarity measure. The experimental results are encouraging and show that the proposed feature easily captures of images and identify relevant objects based on object characteristics.