论文部分内容阅读
为了降低低码率视频编码算法的块效应与环效应，提出了一种基于块区域子波变换和重叠运动补偿的视频编码算法。通过分层游动窗块区域形成算法检测出帧间预测误差场的高能量区域，仅对这些区域进行子波变换和量化，有效节省码率，或在保持码率不变的条件下，有效的降低振荡环效应；根据人类视觉系统（ＨＶＳ）的空－频和时－频特性，给出基于ＨＶＳ的量化模型，以期在信噪比约束下获得最好的主观测试质量。模拟结果表明，对于公共交互格式（ＣＩＦ）图象，在低码率和甚低码率下仍能得到满意的图象质量。该算法可以应用于低码率视频会议系统中。
In order to reduce the block effect and ring effect of the low bit rate video coding algorithm, a video coding algorithm based on block region wavelet transform and overlapped motion compensation is proposed. High-energy regions of the inter prediction error field are detected by the layered moving window region formation algorithm, and the wavelet transform and quantization are only performed on these regions to effectively save the code rate or keep the code rate unchanged Based on the space-time and frequency-frequency characteristics of human visual system (HVS), a HVS-based quantization model is proposed to obtain the best subjective test quality under the signal-to-noise ratio constraint. Simulation results show that for CIF images, satisfactory image quality can still be achieved at low code rates and very low code rates. The algorithm can be applied to low bit rate video conference system.