Title :
Uniform scalar quantizers-distortion and the additive noise model
Author :
Marco, Daniel ; Neuhoff, David L.
Author_Institution :
Dept. of Electr. Eng. & Comput. Sci.,, Michigan Univ., Ann Arbor, MI, USA
fDate :
29 June-4 July 2003
Abstract :
High-resolution infinite level uniform scalar quantizers are considered for the two cases when cell midpoints and cell centroids are used as reconstruction levels. First, it is shown that asymptotically, as the cell size goes to zero, midpoints induce the same MSE as centroids. Hence, they are asymptotically MSE optimal. Secondly, sufficient conditions are found under which the commonly used additive noise model, when midpoints are used, is asymptotically correct. It is also shown that for input probability density functions (pdf´s) that have a single jump discontinuity not at the origin, such a model is not well defined and consequently incorrect.
Keywords :
distortion; mean square error methods; probability; quantisation (signal); signal resolution; MSE; additive noise model; cell centroid; high-resolution infinite level uniform scalar quantizers; mean square error methods; probability density functions; Additive noise; Density functional theory; Error correction; Quantization; Random variables; Sufficient conditions;
Conference_Titel :
Information Theory, 2003. Proceedings. IEEE International Symposium on
Print_ISBN :
0-7803-7728-1
DOI :
10.1109/ISIT.2003.1228479