论文部分内容阅读
提出了一种分层图像压缩框架:图像=边缘轮廓+纹理。对原图像进行了一种自适应的多尺度Wedgelet分析,抽取并编码了图像的边缘轮廓。基于Wedgelet分析了在残差图像中引入的伪迹所具有的局部振荡特性,采用自适应局部余弦变换分析了以纹理为主要内容的残差图像,在将变换系数重组成与小波系数类似的树形结构后,采用零树编码获得了嵌入式码流。实验结果表明,该算法的重建图像质量优于SPIHT算法,在较好地保留原图像边缘轮廓和有效地减少边缘附近振铃伪迹的同时,较清晰的保留了原图像的纹理特征。
A layered image compression framework is proposed: image = edge contour + texture. An adaptive multi-scale Wedgelet analysis of the original image was performed to extract and encode the edge of the image. Based on Wedgelet, the local oscillatory characteristics of the artifacts introduced in the residual image are analyzed. Residual images with texture as the main content are analyzed by adaptive local cosine transform. After the transform coefficients are reconstructed into trees similar to wavelet coefficients After the structure, the embedded code stream is obtained by zerotree coding. The experimental results show that the reconstructed image quality is better than that of SPIHT algorithm, and preserves the texture features of the original image while keeping the edge contour of the original image and effectively reducing the ringing artifacts around the edge.