Image expansion using segmentation-based method

Abstract

Precise image expansion techniques are required for several applications in image processing field. These include reconnaissance photography, cartography, medical images, and satellite imagery. Several methods have been employed for this purpose: (1) pixel replication; (2) area sizing; and (3) interpolation and spline methods. All these methods generate distortion and noticeable degradation in the quality of the image especially on and near edges. We introduce a segmentation-based method that produces significantly improved expanded images and maintains high quality edges. This method segments the image into nonstationary regions and homogenous regions and then expands them separately and via different procedures. Nonstationary regions are expanded using an elaborate look-ahead-and-back procedure. Homogenous regions are expanded using an expanded linear prediction approach. The experimental simulation results show that the expanded images are aesthetically and objectively better than those of other methods.