Аппаратурно-ориентированные алгоритмы обработки матриц кватернионов
Today many DSP-problems use quaternionic matrices for signal and image representation. Singular value decompositions (SVD) and eigenvalue decomposition (EVD) are important problems in this row. The most widely employed method of solving SVD problem is zeroing the entries in a matrix by a sequence of rotations or reflections. In many such computations, it is necessary to zero some quaternion elements selectively with matrix similarity transformation. High practical importance of these problems leads to the need of designing hardware-oriented algorithms to speed up their implementation. The Jacobi method is the frequently used tool and very high throughputs are required from the special processors which will be used as array’s elements for the parallel decomposition of quaternionic matrices. This book contains basic results of some resent publications of authors.