Quality Metrics for Image Compression

The aim of the proposed development is to obtain technically sound, perceptually relevant, and broadly useful quality metrics for digital imagery. These quality metrics are targeted for the evaluation and control of image quality of lossily compressed digital images. To achieve our goals, we. will build upon a methodology for the objective evaluation of image.. quality that we have developed. This objective Picture Quality Scale, PQS, provides an excellent correlation with the subjective rating of image quality by observers. For the rating by observers, or mean opinion score (MOS) , we have used the international standard CCIR-500 image impairment scale. The MOS subjective rating has a correlation of .93 with the objective PQS, taken over the entire range of image quality. The PQS methodology will be applied to all the aspects of thd proposed development, with suitable extensions. Extensions are needed for color images and for problems and applications where the image quality is critical, such as in medical imaging. The extensions will provide an image quality evaluation for subareas of images and for restricted quality ranges. The methodology will therefore meet the needs of most applications. Results will be widely applicable to commercial use of compressed digital imagery and lead to new coders for professional use.