An N valued input symbol is encoded by using a tree structured binary arithmetic coder wherein N is an integer larger than 2.sup.M-1 but not greater than 2.sup.M with M being a non-negative integer. First, the input symbol is converted into M bits of binary symbols. Thereafter, a context for a...http://www.google.co.uk/patents/US5694128?utm_source=gb-gplus-sharePatent US5694128 - Tree structured binary arithmetic coder
