Datakompression

Datakompression, also known as data compression, is the process of reducing the number of bits required to represent information. It can be lossless or lossy, depending on whether exact reconstruction is possible. The goal is to save storage space and bandwidth while preserving the essential content. The efficiency of a compression method is described by the compression ratio, and for lossy methods, by the level of distortion relative to the original data.

Lossless compression exploits redundancy without discarding information. It relies on entropy coding and dictionary-based methods to encode frequently occurring patterns with shorter representations. Common algorithms include Huffman coding, LZ77, LZ78, LZW, Deflate (used in ZIP and gzip), and Brotli. These are used for text, executables, and other data formats where exact recovery is required.

Lossy compression sacrifices some fidelity to achieve higher compression. It uses transform coding, quantization, and perceptual models to remove information deemed less important. Widely used examples are JPEG for images, MP3 and AAC for audio, and H.264/AVC, HEVC, and AV1 for video. The resulting bitrate and perceptual quality depend on quantization levels, encoding settings, and the chosen model.

Data compression is central to archiving, media delivery, and communications. Standards and tools continue to evolve to balance quality, speed, and complexity. Recent work includes learned or neural compression methods that optimize representations with machine learning, often for specific data types. Effective compression depends on data characteristics and the acceptable trade-off between size and fidelity.