Interpretability performance assessment of JPEG2000 and part 1 compliant region of interest coding

Abstract

JPEG2000 is a new international standard for still image coding and provides an extensive set of features vital to many high-end and emerging applications. The default performance metric for JPEG2000 is an optimisation of objective quality versus bit-rate. However, for high compression applications, it would be more appropriate that regions of interest are prioritised for interpretability. JPEG2000 allows for this in its region of interest (ROI) feature. Part 1 of JPEG2000 provides two different ROI coding mechanisms, namely the max-shift and implicit method. With these coders, the objective quality across the image may be selectively sacrificed to achieve higher interpretability in regions of interest. This principle is quantitatively tested using interpretability ratings, assigned using a subjective National Imagery Interpretability Rating Scale (NIIRS), and ROI coding methods are compared in this capacity. Results show that both the max-shift and the implicit method have advantages and disadvantages relative to one another and the choice of methods for ROI coding is very much dependant on the requirements of the application at hand.