Date of Award


Document Type


Degree Name

Doctor of Philosophy (PhD)


Computer Science

First Advisor

John M. Tyler


Image compression is the process of reducing the number of bits required to represent an image. This can be achieved by reducing (or ideally, eliminating) various types of redundancy that exist in the imaging data. All current image compression methods are based on the same theoretical model, which targets redundancy found in individual images. However, this model ignores an additional type of redundancy that exists in sets of similar images, the "set redundancy". The source of set redundancy is the common information existing in more than one image in a set of similar images. Set redundancy can be recognized by the appearance of similar pixel values at the same image regions, comparable image histograms, similar image features, or analogous distributions of edges. This research explores the concept of set redundancy and establishes its importance for image compression. A new theoretical compression model is proposed, the Enhanced Compression Model, which extends the current model by including set redundancy extraction. The requirements and restrictions for set redundancy extraction are discussed, and practical methods to implement it are developed. These methods are collectively referred to as SRC or Set Redundancy Compression methods. Three SRC methods are presented: the Min-Max Differential (MMD) method, the Min-Max Predictive (MMP) method, and the Centroid method. According to the Enhanced Compression Model, the SRC methods can be combined with any current compression technique to achieve higher compression ratios when compressing sets of similar images. One of the best application areas for the Enhanced Compression Model is medical imaging. Medical image databases usually store large sets of similar images and, therefore, contain significant amounts of set redundancy. Tests were performed by implementing the SRC methods on a test database of CT and MR brain images. The results show an average of as much as two-fold improvement in the performance of standard compression techniques when they are combined with the SRC methods. In addition, the SRC methods developed in this research are fast, lossless, and easy to implement.