Low Complexity RDO Model for Locally Subjective Quality Enhancement in LAR Coder

Yi Liu (1) , Olivier Déforges (1) , François Pasteau (1) , Khouloud Samrouth (1)

1 IETR - Institut d'Électronique et des Technologies du numéRique

This paper introduces a rate distortion optimization (RDO) scheme with subjective quality enhancement applied to a still image codec called Locally Adaptive Resolution (LAR). This scheme depends on the study of the relation between compression efficiency and relative parameters, and has a low complexity. Linear models are proposed first to find suitable parameters for RDO. Next, these models are combined with an image segmentation method to improve the local image quality. This scheme not only keeps an effective control in balance between bitrate and distortion, but also improves the spatial structure of images. Experiments are done both in objective and subjective ways. Results show that after this optimization, LAR has an efficient improvement of subjective image quality of decoded images. This improvement is significantly visible and compared with other compression methods using objective and subjective quality metrics.

Mots clés: rate distortion optimization; visual improvement; quadtree; image coding

Domaines: Traitement du signal et de l'image [eess.SP]; Traitement du signal et de l'image [eess.SP]

Yi Liu, Olivier Déforges, François Pasteau, Khouloud Samrouth. Low Complexity RDO Model for Locally Subjective Quality Enhancement in LAR Coder. IEEE International Conference on Signal and Image Processing Applications (ICSIPA), Oct 2013, Malaysia. ⟨hal-00920598⟩