摘要
针对手机和其他嵌入式平台特点,对AVS-P7 Jiben解码器进行了C语言级优化.通过对流缓存、指数Golomb解码、VLD码表、反量化、反变换、宏块解码、运动补偿插值等模块的优化处理,降低了运算复杂度,减少了内存开销.在S3C2440(ARM920T)平台上的测试结果表明,解码器在同等PSNR、无失真情况下比原有解码器提升了8~10倍的性能,基本满足了实时解码的需要.本文最后采用标准的AVS-P7一致性测试码流对优化后的解码器进行了一致性测试,发现同AVS标准组织提供的wm2.7相比,测试结果更加理想.
According to the characteristics of the mobile and other embedded system, the AVS-P7 Jiben profile decoder was optimized with C language. After the optimization of the streambuffer, Exp-Golomb decoding, VLD lookup table, inverse quantization, inverse transformation, macroblock decoding, motion compensation and interpolation modules, the computational complexity and the memory were greatly reduced. The simulation results in the S3C2440 (ARM920T) showed that the optimized decoder was improved 8-10 times as compared with the original decoder with the same PSNR without any distortion, which meet the basic real-time decoding requirements. Finally, the AVS standard conformance test using the AVS-P7 conformance test bitstream showed that the test results are more satisfied and ideal than the wm 2.7 provided by the AVS standard organization.
出处
《浙江大学学报(理学版)》
CAS
CSCD
北大核心
2008年第4期416-421,427,共7页
Journal of Zhejiang University(Science Edition)
基金
韩国三星电子公司资助研究
关键词
解码器
运动补偿插值
边界填充
宏块解码
decoder
motion compensation interpolation
boundary padding
macroblock decoding