Video compression

Video compression reduces the data needed to represent moving images by exploiting spatial and temporal redundancy or discarding less perceptible information.

Motion compensation: A prediction method that uses estimated movement between frames to represent image regions efficiently. It exploits temporal redundancy by predicting blocks from earlier or later frames.
Digital video: A representation of moving images as sequences of digitally encoded pictures. Compression operates on sampled pictures arranged over time.
H.264: A widely used video-coding standard also known as Advanced Video Coding. It brought efficient compression to broadcast, discs, web video, and mobile devices.
Lossy compression: Compression that permanently discards information to reduce data size. Most practical video compression uses controlled information loss to achieve small files.
Discrete cosine transform: A transform that represents image data as weighted spatial-frequency components. Transform coefficients concentrate image energy so less important detail can be encoded coarsely.
Chroma subsampling: A method that stores color-difference information at lower spatial resolution than brightness information. Human vision often tolerates reduced color detail, allowing fewer samples per frame.
H.265: A video-coding standard also known as High Efficiency Video Coding, designed to improve compression over H.264. It targets lower bitrates for comparable quality, especially for high-resolution video.
Lossless compression: Compression that reduces data size while permitting exact reconstruction of the original data. It preserves every sample but usually achieves less reduction than lossy video coding.
Quantization: A many-to-one mapping that reduces numerical precision by replacing values with a smaller set of levels. Coarser coefficient levels lower bitrate while introducing lossy distortion.
Color space: A mathematical coordinate system for representing colors as numerical components. Codec design depends on how image colors are represented before transformation and quantization.

Linked from 13 pages