Recompression of Still Images
Syin Chan · Kent Academic Repository (University of Kent) · 1992
The effects of subjecting a digital image to more than one step of the lossy JPEG baseline data compression are studied. The performance in terms of the signal-to-noise ratio as well as visual artifacts are observed and presented. Two types of visual artifacts have been identified and they are found to correlate closely to the two types of errors introduced when the Discrete Cosine Transform coefficients undergo repeated quantisation as part of the JPEG compression algorithm. A method is proposed to estimate the amount of errors generated in the process, which may be used to determine if any special processing is required to reduce these errors. Several methods to overcome the more severe artifacts in the images are investigated and the experimental results are presented.