WiSe 18/19: Bild- und Videocodierung I: Grundlagen
Today's natural use of digital media has only become possible through efficient coding techniques. The lecture will discuss the basic coding techniques entropy coding, quantization, predictive ... read more
Today's natural use of digital media has only become possible through efficient coding techniques. The lecture will discuss the basic coding techniques entropy coding, quantization, predictive coding, transformation coding, which are the main components of modern image and video codecs. For all the coding techniques discussed, a detailed analysis of the coding efficiency is carried out on the basis of simple signal models. In addition, practical applications in image and video coding will be demonstrated.
The topics covered form the basis for the advanced lecture "Image and Video Coding II: Algorithms and Applications", in which the application of the introduced coding techniques in modern image and video codecs is discussed in detail.
Lossless coding / entropy coding: unambiguously decodable codes; derivation of different entropy measures as lower limit of the average codeword length; Huffman coding; arithmetic coding
Rate distortion theory: maximum coding efficiency of lossy methods; information rate distortion function; differential entropy; Shannon limit
Quantization: scalar quantization with fixed and variable length codes; rate distortion functions for high rates; vector quantization; comparison of different quantizers
Prediction and predictive coding: linear prediction; prediction gain; quantization in the prediction loop; predictive coding efficiency
Transformation coding: Structure of transformation coding systems; Unitary block transformations; Bit assignment for transformation coefficients; Optimal transformation for Gaussian processes; Signal-independent transformations; Efficiency of transformation coding