Electronics Letters vol. 36, no. 15, pp. 1270 - 1271
Publisher
IEE, Stevenage, United Kingdom
Indexed
SCI; SCIE; SCOPUS
Document Type
Article
Abstract
Post-processing, restoration or transcoding of JPEG and MPEG compressed images require a quantization error model of discrete cosine transform coefficients. A method is presented to recover the original probability density function (pdf) parameter from quantized data, assuming that the original pdf of the coefficients is Laplacian. Quantization errors are calculated from the Laplacian model.