著者

斎藤 隆弘 武尾 英哉 相澤 清晴 原島 博 宮川 洋

出版者

一般社団法人映像情報メディア学会

雑誌

テレビジョン学会技術報告 (ISSN:03864227)

巻号頁・発行日

vol.9, no.12, pp.105-112, 1985-07

---

Formerly we proposed a method for designing parallel gain/shape vector quantizers in the 8*8 DCT (Discrete Cosine Transform) domain [12]. This paper proposes a method for magnifying the block of this quantizer. In this method, the block is magnified by symmetrizing the structure of the shape code book. By magnifying the block to 16*16,at a fixed coding rate signal-to-noise ratio is increased by about 1 (dB), and this coding scheme yields a perfomance advantage of 1〜2 (dB) over the DCT coding scheme with scalor, Max-type quantizers.