Abstract:
A direct interpolation method for fractional pixels is proposed to cope with the tremendous memory consumption and high computational complexity of H.264 Fidelity Range Extension(H.264 FRExt) decoding. It overcomes the calculation dependency for 1/4 pixels in H.264 FRExt and only computes the required fractional pixels. An error elimination approach for the method is given to produce the same result as H.264 FRExt. Experimental results show that the method can lower the memory requirement for decoding and save a lot of redundant computation for unused fractional pixels.
Key words:
video decoding,
resource optimization,
fractional pixel interpolation,
Decoded Picture Buffer(DPB),
H.264 Fidelity Range Extension (H.264 FRExt) standard
摘要: 针对H.264 FRExt会大幅增加解码器对内存和计算资源消耗的问题,提出一种动态分数像素插值法,用以优化H.264 FRExt解码所需的存储和计算资源。该方法突破了H.264 FRExt定义中分数点必须按1/2像素到1/4像素分层计算的限制,直接计算运动补偿所需的分数像素,通过误差消除方法保证计算结果与标准完全一致。实验结果证明,该方法可以较大地节省内存,避免对冗余分数点的计算,加快解码速度。
关键词:
视频解码,
资源优化,
分数像素插值,
解码图像缓存,
H.264 FRExt标准
CLC Number:
LIU Ji-Yuan, HOU Fang, HUANG Cheng-Hui. Resource-optimized H.264 FRExt Decoding[J]. Computer Engineering, 2011, 37(7): 276-278.
陆寄远, 侯昉, 黄承慧. 资源优化的H.264 FRExt解码[J]. 计算机工程, 2011, 37(7): 276-278.