摘要
提出了一种具有分辨率可扩展的适合便携带设备的视频压缩方法。该方法分为两阶段:(1)利用人眼视觉模型对小波系数进行量化;(2)对于低频子带,利用周围的系数对当前系数进行预测,对预测误差进行编码,对于高频子带,先进行Run-length编码,再进行Huffman编码。实验结果表明,在相同的视觉效果下,该方法获得了与MJPEG2000相当的压缩比,并且压缩所需要时间更少。
This paper presents a video compression scheme with resolution scalability and low computation power for portable devices such as PDA and mobile phone. It includes two stages： （1） the wavelet coefficients in a frame are quantized by the threshold formed by the eye＇s sensitivity to their subband. （2） to the low frequence subband, the differences between wavelet coefficients and their prediction values formed by the neighbor coefficients are encoded by the Huffman encoder, and, to the high frequence subbands, the run lengths of wavelet coefficients and their values are also encoded by Huffman encoder. Moreover, statistical frequencies of symbols used by Huffman encoder is resulted from the previous frame. Experimental results show that the scheme not only achieves compression ratio similar to MJPEG2000, but also costs tess time than MJPEG2000.
出处
《计算机工程》
CAS
CSCD
北大核心
2006年第4期225-227,共3页
Computer Engineering
基金
重庆市教委科研基金资助项目(020612)
关键词
视频压缩
小波系数模型
便携带设备
分辨率可扩展
Video compression
Wavelet coefficient model
Portable device
Resolution scalability
