Novel Techniques for Multi-Resolution Imaging

Efficient image zooming or interpolation algorithms that have low complexity and high quality is an important goal for multi-media distribution. This is because content is consumed with different sizes and sent over networks with different bandwidths. This book discusses four approaches for image zooming and multi-resolution image representation. It also presents a few new tools that can be used in the image/video compression pipeline and demonstrates a novel usage of dictionary coding for improved compression. It analyses a drawback in the hierarchical mode of the JPEG and proposes methods to overcome this. One of the unpleasant results of image interpolation is the introduction of localized artifacts. The methods described herein demonstrates an algorithm to mitigate this called Targeted Image Enhancement (TIE) wherein, a small amount of additional data is used to remove many of these artifacts. Finally, the book proposes a non-linear approach to interpolating digital images called Location and Neighborhood Adaptive (LNA) interpolation. The improvement stemming from this method is demonstrated on binary images.