Feature Based Image Manipulation

Abstract

This paper presents a method for feature-based image representation. In the proposed method, an image is decomposed into a skeleton component and a residual component. The skeleton component is obtained by performing iterative linear interpolation based on high-curvature points, which are extracted from the image surface. These points help in representing the image through a compact and reversible way. The residual component shows the difference between reconstructed component and the original image. The compact representation offers an encrypted and compressed form which can be transmitted over blue-tooth to other devices. Proper authentication offers one more level of security. By efficient manipulation, several applications, such as adaptive contrast enhancement, achromatic information modification and shape editing can be achieved.