Z-Images

Abstract

Local patterns and other patch based features have been an integral part of various computer vision applications as they encode local structural and statistical information. In this paper, we propose an image coding technique that utilizes Zeckendorf representation of pixel intensities and basic mathematical operators such as intersection, set difference, maximum, summation etc. for summarization of image regions. The algorithm produces a Z-coded image that tells about the homogeneity or the contrast in image regions with all codes in a range of 0 to 255.Vigneron-This research was supported by the program Cátedras Franco-Brasileiras no Estado de São Paulo, an initiative of the French consulate and the state of São Paulo (Brazil). We thank our colleagues Prof. João M.T. Romano, Dr. Kenji Nose and Dr. Michele Costa, who provided insights that greatly assisted this work.