The Estimated Spectrum Adaptive Postfilter (ESAP) is an adaptive digital image processing tool that enhances standardized digital JPEG imagery, MPEG video, and HDTV at very low bit rates (less than 0.25 BPP). Since these digital technologies can be scaled down to accommodate the low bit rates needed in constrained bandwidth space exploration applications, ESAP can be used to compensate for these low bit rates where applicable.
ESAP looks into the decoded JPEG image and determines the location of the edges of the image. An edge is the spatial transition from a human face, for example, to a landscape background or the silhouette or cartoon of a person or object. In the areas away from the edges, ESAP performs adaptive filtering to remove the pixelation (or blocking artifacts) created by highly compressed JPEG images.
In general, ESAP improves both the subjective visual quality of highly compressed JPEG images as well as their objective quality measure known as the Peak-Signal-to-Noise Ratio (PSNR), as compared to baseline JPEG images. It does this without requiring any additional overhead in the data stream.