Quantization table design for JPEG compression of angiocardiographic images

The lossy JPEG standard may be used for high performance image compression. As implemented in presently available hard- and software in most cases the so-called luminance quantization table is applied for gray level images, which may be scaled by a quality factor. The questions arise which quality factor is optimal and whether it is possible and worthwhile to specify quantization tables for the particular characteristics of angiocardiograms. Two new quantization tables are derived from the transfer function of the angiocardiographic system, which is a worst-case approach with respect to preserving sharp edges. To assess the performance, evaluations based on Hosaka-plots are developed. These diagrams compare the different errors introduced by lossy JPEG compression objectively.<>