基于Multibandelets的自适应图像压缩

为了在图像压缩时更好地保护具有方向性的几何结构信息，该文构建了一种新的基函数，称为Multibandelets，并结合Shannon编码用于自然图像的压缩。实验结果表明：与多小波、具有同样消失矩的小波和Bandelets相比较，基于Multibandelets的图像压缩在视觉效果和客观衡量指标两方面都有改善，尤其对具有方向性的细节和纹理信息具有更好的表示。

Adaptive Image Compression Based on Multibandelets

For protecting directional geometric structure information, a novel basis function called multibandelets is presented in this paper. And then an image compression algorithm is proposed based on the multibandelet transform and Shannon encode. Experiments show that the multibandelets-based compression provides improvements both in visual effects and quantitative analysis, especially for the detail information containing directional geometric structures. The compared methods are multiwavelets, and those of the wavelets and bandelets with the same vanishing moments, respectively.