Abstract

Motion compensated frame interpolation (MCFI) is a technique used extensively to enhance the visual quality of low frame rate videos. It requires true motion vectors (MVs) to ensure the high quality of the interpolated image. However, many current motion estimation techniques are block-based, which cannot always provide reliable MVs. In this paper, a novel motion vector processing method is presented to obtain the reliable MVs for MCFI. By introducing the interpolated artifact information of each MV as a new reliability metric, the proposed method first classify the input motion vector field (MVF) into reliable and possibly unreliable groups. A correlation-based MV reliability determination method is then used to identify the unreliable MVs from the possibly unreliable group with the reference of the reliable MVs. In the motion correction stage, these unreliable MVs are corrected properly with an region classification method. By using the updated MVF as new input, the above processes are performed iteratively until no new MV reliability changes. Finally, an occlusion handling process is performed to solve the unreliable MVs within the occluded areas and remove the small motion inconsistencies. Experimental results demonstrate the superior performance of the proposed method. Compared with existing benchmark algorithms, the frames interpolated by the proposed method provide better image quality both subjectively and objectively.

© 2014 IEEE

PDF Article

References

You do not have subscription access to this journal. Citation lists with outbound citation links are available to subscribers only. You may subscribe either as an OSA member, or as an authorized user of your institution.

Contact your librarian or system administrator
or
Login to access OSA Member Subscription

Cited By

You do not have subscription access to this journal. Cited by links are available to subscribers only. You may subscribe either as an OSA member, or as an authorized user of your institution.

Contact your librarian or system administrator
or
Login to access OSA Member Subscription