Matrix methods of constructing one-norm integer cosine transforms of order-8,16 are considered. The one-norm order-8 and 16 integer transforms are proposed and their fast algorithms of low computational complexity are developed whose computational complexity is 3 to 5 time less than that in the well-known algorithms and is 10 time less than one in standard H.265. Figs.: 3. Tabl.: 9. Refs: 18 назв.
Гнатив Лев Алексеевич, кандидат техн. наук, старший научный сотрудник Института кибернетики им. В.М. Глушкова НАН Украины, Киев,
e-mail: levhnativ@gmail.com