Kompressionsalgorithmus

Kompression, in German often written as Kompression, refers to the process of encoding information using fewer bits or units than the original representation. The primary aim is to reduce storage needs or bandwidth requirements while allowing faithful reconstruction of the original data, either exactly or with controlled loss of quality.

Two broad categories dominate the field: lossless and lossy compression. Lossless compression preserves all original information, enabling exact reconstruction. It is essential for text, software, and some data files where any loss is unacceptable. Lossy compression achieves higher reductions by discarding some information, typically imperceptible or statistically unlikely, and is common for multimedia where some quality loss is acceptable to gain substantial size savings.

Common lossless methods include entropy coding (such as Huffman coding), dictionary-based schemes (like LZ77, LZ78, and LZW), and hybrids such as the DEFLATE method used in ZIP and PNG. Lossless techniques balance compression ratio, speed, and memory usage to suit different applications.

Lossy methods rely on perceptual or statistical models to remove information judged to be less important. Transform coding, quantization, and entropy coding are typical steps. Images often use discrete cosine transform (DCT) based methods (as in JPEG) or wavelets. Audio uses perceptual coding with perceptual noise shaping (as in MP3 and AAC). Video combines temporal and spatial transforms (as in MPEG/H.265/HEVC).

Applications span archiving, transmission, streaming, and multimedia storage. Key metrics include compression ratio, bit rate, and distortion measures such as peak signal-to-noise ratio (PSNR) or perceptual quality scores. Standards and formats—ZIP, PNG, JPEG, MP3, AAC, MPEG, and others—govern interoperability and performance.