摘要
针对无损图像压缩编码,提出了一种新颖的图像分解去相关方法。当前的无损图像编码方法主要有CALIC和JPEG-LS,两者都在空域直接作预测,导致编码码流不具有分辨率可伸缩性。结合小波提升模式与边缘自适应预测研究实现了一种比二维小波变换性能更好的分解方法。首先,对图像的每一列样值进行一维小波分解;然后,对高频子带进行边缘自适应预测,减少残留的信息。针对低频子带图像进行同样的两步操作,就完成了对图像的一次二维分解。对低频图像进行多次迭代操作后即形成了对图像的一个多分辨率分解。实验结果表明,与JPEG2000的无损模式相比,由于边缘自适应预测的引入,提出的分解模式获得了明显的编码增益。
This paper proposed a new decomposition scheme for lossless image compression by incorporating edge- directed adaptive prediction with wavelet lifting scheme. A vertical one-Dimension Discrete Wavelet Transform （1D-DWT） was first applied to images by means of lifting scheme. Second, edge-directed adaptive prediction procedure was applied to those high-frequency sub-band coefficients generated by the previous DWT. And then, a similar horizontal decomposition was performed in the low-frequency sub-band generated by vertical decomposition. A multi-resolution representation was thus acquired by an iterative repetition at the produced low-resolution approximation. Unlike the well-known coder CALIC and JPEG-LS, this scheme can provide a resolution scalable code-stream due to DWT. In addition, the experimental results indicate, due to the edge-directed prediction, this decomposition scheme has achieved noticeably better performance of lossless compression than JPEG2000 which supports resolution scalability.
出处
《计算机应用》
CSCD
北大核心
2013年第6期1697-1700,共4页
journal of Computer Applications
基金
国家自然科学基金资助项目(61201452
61272278)
关键词
无损图像压缩
多分辨率表示
分辨率渐进性
边缘自适应预测
小波变换
lossless image compression
multi-resolution representation
resolution scalability
edge-directedprediction
Discrete Wavelet Transform （DWT）
