Region-Based Guaranteed Image Quality in JPEG2000

Abstract

This paper describes a spatially variant method that optimises the compression rate within a given image dynamically depending on the application specific requirements and the actual image content. The concept utilises the JPEG2000 standard, however, a generalised region-of-interest (ROI) map is used in this paper together with a different post-compression rate distortion optimisation. The generalised ROI map comprises a multi-level priority scheme as well as continuously adjustable qualities for each individual region to reflect the application specific requirements for a particular image. The key point is that these pre-defined quality constraints are always met for each individual region, thus making the approach suitable for data with inherent reliability requirements. The implementation of the proposed system is transparent and concerns only the encoder side while any standard JPEG2000 decoder can be used for decompression. Results have shown that for the utilised imagery, i.e. satellite and medical images, compression rate improvements of up to 32% in comparison to the JPEG2000 compressor fulfilling the same quality needs are achievable.