Postprocessing of MPEG(-2) video is widely used to attenuate the coding artifacts, especially deblocking but also deringing have been addressed. The focus has been on filters where the decoder has access to the code stream and e.g. utilizes information about the quantization parameter. We consider the case where the coded stream is not accessible, or from an architectural point of view not desirable to use, and instead estimate some of the MPEG stream parameters based on the decoded sequence. The I-frames are detected and the quantization parameters are estimated from the coded stream and used in the postprocessing. We focus on deringing and present a scheme which aims at suppressing ringing artifacts, while maintaining the sharpness of the texture. The goal is to improve the visual quality, so perceptual blur and ringing metrics are used in addition to PSNR evaluation. The performance of the new `pure' postprocessing compares favorable to a reference postprocessing filter which has access to the quantization parameters not only for I-frames but also on P and B-frames.
|Title of host publication||Conference proceedings : 2009 IEEE 16th International Conference on Image Processing|
|Number of pages||4|
|Volume||IEEE Catalog No: CFP09CIP-USB|
|Publication status||Published - 2009|
|Event||IEEE 16th International Conference on Image Processing ICIP 2009 - Cairo, Egypt|
Duration: 7 Nov 2009 → 10 Nov 2009
|Conference||IEEE 16th International Conference on Image Processing ICIP 2009|
|Period||07/11/2009 → 10/11/2009|
Bibliographical noteCopyright 2009 IEEE. Personal use of this material is permitted. However, permission to reprint/republish this material for advertising or promotional purposes or for creating new collective works for resale or redistribution to servers or lists, or to reuse any copyrighted component of this work in other works must be obtained from the IEEE.
Forchhammer, S. (2009). Postprocessing MPEG based on estimated quantization parameters. In Conference proceedings: 2009 IEEE 16th International Conference on Image Processing (Vol. IEEE Catalog No: CFP09CIP-USB, pp. 3929-3932). IEEE. https://doi.org/10.1109/ICIP.2009.5414026