Geometrical multiresolution adaptive transforms. Theory and applications (Q2442882)

Nowadays, many image processing techniques are based on multiresolution geometrical methods of image representation. In these methods, a wide family of functions is used and we can divide them into two groups: adaptive (wedgelets, beamlets, surflets) and nonadaptive (ridgelets, curvelets, shearlets). The book is a concise introduction to one of such methods, namely multismoothlets. They belong to the class of adaptive methods. The book is divided into two parts. The first part is devoted to the theory of multismoothlets. In Chapter 2 a family of functions called smoothlets is presented. A smoothlet is a generalization of a wedgelet and a second-order wedgelet. Moreover, it is a continuous functions. This property makes it different from the other adaptive functions used in the literature. Additionally, a sliding smoothlet and the sparsity of the smoothlets are discussed. Next, in Chapter 3, a multismoothlet is defined as a vector of smoothlets. To visualize a simple multismoothlet, two visualization methods are presented. The first method called serial is used to obtain the best quality, whereas the second method called parallel is used to obtain the result relatively fast. Similar to the smoothlets, a sliding multismoothlet is presented and the sparsity is discussed. In Chapter 4, a multismoothlet transform based on moments is presented. The chapter starts with a fast wedgelet transform. Next a smoothlet and a multismoothlet transform is presented. In the case of the multismoothlet transform two versions are introduced: serial and parallel. The chapter ends with a discussion about the computational complexity of the transforms presented in the chapter. The second part of the book is devoted to applications of the presented transforms. In Chapter 5, application in image compression is presented. Compression of two types of images is considered: binary and grayscale. In the former case, a method based on curvilinear beamlets and, in the latter case, a method based on smoothlets is used. The obtained results are compared to the known state-of-the-art compression methods. Next, in Chapter 6, an application of the multismoothlet transform in image denoising is presented. The results obtained with the proposed algorithm are compared to the known state-of-the-art methods of image denoising. The last application of the multismoothlet transform, presented in Chapter 7, is edge detection. Two methods are proposed. The first method is based on the multismoothlet transform and the second one is based on sliding multismoothlets. The two methods are compared to the known state-of-the-art methods. In Chapter 8, some concluding remarks and further directions in the area of multismoothlets are presented. The book ends with three appendices. In the first appendix, benchmark images used in the tests are presented. Appendix B is devoted to the bottom-up tree pruning algorithm, and the last appendix is devoted to the algorithm of smoothlet visualization. The book is presented in a very accessible fashion. The author gives many examples presenting the notations and problems considered, so it makes the learning easier. It is suitable for students, researchers and engineers interested in multiresolution representation of images and its applications.