资料压缩培训资料.ppt

Fundamental Concept;For multimedia, A-to-D ( D-to-D) are a significant stage using various techniques for “compression” with acceptable data-precision loss (precision truncation) ;Why is Data Compression needed ?;Why Data Compression ?;Why Data Compression ?;Why Data Compression ?;Why Data Compression ?;Why Data Compression ?;Why Data Compression ?;Why Data Compression ?;Why Data Compression ?;Why Data Compression ?;Why Data Compression ?;“Requiring” compressed bit rate vs. media complexity assessed by spectral complexity ;objective redundancy :repeated signals, periodic signals nonsense signal (white noise) Subjective redundancy: saving (removing) it will be “unaware/insensible” , “neglectful ”, “acceptable”, ”Tolerable”, and “the unhappy compromise conceded by user” to form the order of increasing compression rates ;The content complexity mainly depends on spectrum (frequency domain). The frequency coefficients can be simply calculated from the DCT discrete cosine transform. The 64 (8 x 8) DCT basis functions are illustrated;頻域轉換 change domains from time to spectrum for signal analysis such as the degree of saving (cutting them out by large quantization step) medium and high-frequency coefficient;f(n) sampled from f(t);Why is block-based coding popularly adopted in compression standards? Different objects are segmented into enough smaller parts, Their similarity will appear and redundancies could be apparent. the perceptual redundancies can be easily found and then largely removed from the frequency domain much over spatial domain. Rather, Computing spectrum transforms such as Fourier, DCT 1 For large block (even whole image) are very time-consuming. Hence, the block of suitable size are segmented from an image as compression unit. ;2. Naturally, different objects and regions in general have distinct “motion” directions. So, for sufficiently finding and then removing the temporal redundancies between adjacent frames, an imag