計算機内における画像のデータ圧縮

概要

論文の詳細を見る
When large images such as ERTS multispectral images are processed by a system using a minicomputer, data compression is desirable for more efficient uses of anxiliary memories. Although much work has been done on bandwidth compression in image communication, it is not directly applicable to data comptession within computer. Major differences are: (1) Exact restoration is required in data compression, whereas distortions unnoticed by human eye may be allowed in bandwidth commpression. (2) Random access to data is required within a computer, whereas serial transmission is sufficient for communication. (3) In a computer, complicated processing and abundant use of memories and registers are possible, which are not practical in communication. (4) Error control is important in communication, but can be neglected in a computer. With these points in mind, this paper presents an algorithm for data compression and restoration, which is based on Huffman code. An implementation in a minicimputer (HP-2100) shows that the compression ratio of 1/2.54 for ERTS data is obtained and the processing time is comparable in comparison with the time for data transfer.
一般社団法人情報処理学会の論文
1977-08-15