Fast multiple reference frame selection algorithm for H.264/AVC
Yao Cheng
Abstract
Yao Cheng
Abstract
H.264/AVC uses multiple reference frame motion estimation to enhance coding efficiency.But the computational cost linearly increases with the number of reference frames.In this paper,the candidate reference frames are determined based on edge detection and correlation information.The motion characteristics of the macroblock are judged by motion vector field with P8×8 mode.A fast multiple reference frame selection algorithm is proposed.Experiments on JM10.2 are performed and results show that the proposed algorithm significantly saves encoding time with neglectable loss of image quality.
A significance statement is not available in the OpenAlex record.
A contribution statement is not available in the OpenAlex record.
Method details are not available in the OpenAlex metadata.
Findings are not separately available in the OpenAlex metadata.
Limitations are not available in the OpenAlex metadata.
Application details are not available in the OpenAlex metadata.
H.264/AVC uses multiple reference frame motion estimation to enhance coding efficiency.But the computational cost linearly increases with the number of reference frames.In this paper,the candidate reference frames are determined based on edge detection and correlation information.The motion characteristics of the macroblock are judged by motion vector field with P8×8 mode.A fast multiple reference frame selection algorithm is proposed.Experiments on JM10.2 are performed and results show that the proposed algorithm significantly saves encoding time with neglectable loss of image quality.
Key concepts: Macroblock, Reference frame, Motion vector, Inter frame, Computer science, Coding (social sciences), Motion estimation, Algorithmic efficiency