Digital image processing techniques

         Digital image processing techniques is destined mainly to the students of Economy Informatics master program and Economy Informatics doctoral school of the Faculty of Cybernetics, Statistics and Economy Informatics from the Bucharest University of Economics. It is also useful for all specialists and others interested in this field.

             The monography cover some of the usual problems in digital image processing like: representation and statistical characterization of digital images, image enhancement, statistical transformations and frequency analysis for detection of image signal characteristics, restauration methods for noise removal and also removal of other types of convolution-based perturbations (atmospheric turbulence, horizontal/vertical/mixed movement in continuous or finite domain) and some of image registration techniques. In the following a brief description of techniques provided in each chapter is provided.

          Chapter 1 presents introductory notions, including general concepts related to image types, classified by the way chromatic/achromatic light intensity is measured, statistical image characterization elements based on first and second order statistics and also arithmetical operations with images, with considerations regarding their practical use. The second chapter, Image transforms, includes some of the most widely used image transforms in order to obtain representations that facilitate later processing, namely the SVD (Singular Value Decomposition) of images, the Fourier transform, Cosin, Sin and Hartley transforms, unitary transforms that use, like TFD, sinusoidal base functions, the Karhunen-Loeve representation and principal component analysis (PCA).   The most widely used techniques of image enhancement are described in chapter 3 and include: Log and Gamma transforms, linear/polynomial operators (with interpolation), histogram equalization and histogram specification methods. Next, spatial/frequency filtering methods for image smoothing are described followed by methods for enhancing object edges.       Chapter 4, Image restauration methods, includes a series of classical and novel techniques used to remove or reduce the unwanted effects (noise, movement etc.) from digital images. Restauration techniques are mainly oriented towards creating a mathematical model of the degradation, followed by applying the reverse process in order to obtain the initial image. This type of approach usually involves the definition of performance criteria leading to optimal estimations of desired result. One of the most common tasks in image processing is to match two or more pictures of the same scene taken at different times or/and different sensors and viewpoints.  Special tailored techniques, called image registration, are employed in order to identify the differences between the images; some of the most commonly used methods being presented in Chapter 5.

Each presented technique is accompanied by experiments and considerations regarding its application. Also, the experiments are described in the context of adequate information measurements, in order to highlight the main characteristics of associated methods.