The sampling lattice used to digitize continuous image data is a signi?cant determinant of the quality of the resulting digital image, and therefore, of the e?cacy of its processing. The nature of sampling lattices is intimately tied to the tessellations of the underlying continuous image plane. To allow uniform sampling of arbitrary size images, the lattice needs to correspond to a regular - spatially repeatable - tessellation. Although drawings and paintings from many ancient civilisations made ample use of regular triangular, square and hexagonal tessellations, and Euler later proved that these three are indeed the only three regular planar tessellations possible, sampling along only the square lattice has found use in forming digital images. The reasons for these are varied, including extensibility to higher dimensions, but the literature on the rami?cations of this commitment to the square lattice for the dominant case of planar data is relatively limited. There seems to be neither a book nor a survey paper on the subject of alternatives. This book on hexagonal image processing is therefore quite appropriate. Lee Middleton and Jayanthi Sivaswamy well motivate the need for a c- certedstudyofhexagonallatticeandimageprocessingintermsoftheirknown uses in biological systems, as well as computational and other theoretical and practicaladvantagesthataccruefromthisapproach. Theypresentthestateof the art of hexagonal image processing and a comparative study of processing images sampled using hexagonal and square grids.
Hexagonal Image Processing provides an introduction to the processing of hexagonally sampled images, includes a survey of the work done in the field, and presents a novel framework for hexagonal image processing (HIP) based on hierarchical aggregates.
Digital image processing is currently dominated by the use of square sampling lattices, however, hexagonal sampling lattices can also be used to define digital images. The strengths offered by hexagonal lattices over square lattices are considerable:
higher packing density,
uniform connectivity of points (pixels) in the lattice,
better angular resolution by virtue of having more nearest neighbours, and
superlative representation of curves.
The utility of the HIP framework is demonstrated by implementing several basic image processing techniques (for the spatial and frequency domain) and some applications. The HIP framework serves as a tool for comparing processing of images defined on a square vs hexagonal grid, to determine their relative merits and demerits. The theory and algorithms covered are supplemented by attention to practical details such as accommodating hardware that support only images sampled on a square lattice.
Including a Foreword written by Professor Narendra Ahuja, an eminent researcher in the field of Image Processing and Computer Vision, the book s fresh approach to the subject offers insight and workable know-how to both researchers and postgraduates.
