摘要
将感兴趣区编码和多描述编码相结合,提出一种噪声信道下保护图像感兴趣区的多描述编码方法。该方法将图像分为感兴趣区和背景区,并采用多描述编码方法,将感兴趣区和背景区码流编码成多个描述。同时对感兴趣区和背景区描述码流采用不同的冗余树分配方案,以进一步增强图像感兴趣区的抗分组丢失能力。实验证明,该方法即使在信道拥塞较为严重的情况下,也能为图像的感兴趣区提供令人满意的重建质量。
This paper proposed a new multiple description coding for protecting regions of interest in images over noisy channels by combining the advantages of multiple description coding and regions of interest coding.The SPIHT algorithm was used to generate multiple description streams,and divided these description streams into regions of interest streams and background streams.For the protection of the regions of interest,the method assigned different redundant trees to the two kinds streams,the full coding trees were assigned as the redundant tress of the regions of interest streams,the partial coding trees were assigned as the redundant tress of the background streams.Experiment results show that the method can further increase the reconstructed quality of the ROI over noisy channels compared with other ROI multiple description coding methods.
出处
《计算机应用研究》
CSCD
北大核心
2010年第11期4345-4347,共3页
Application Research of Computers
基金
国家自然科学基金资助项目(60572011)
关键词
噪声信道
感兴趣区
多描述编码
noisy channels
regions of interest
multiple description coding
