论文部分内容阅读
传统提升小波变换无法有效重构遥感图像中的非水平与非垂直高频信息,导致这些地方的高频小波系数仍然较为显著,降低了遥感图像的编码效率。提出了一种新的基于方向优化的提升小波框架(DOLW)。设计基于梯度的方向预测模型获得提升小波的最优变换方向;沿最优变换方向对图像进行先垂直后水平的方向提升变换,削弱遥感图像高频子带中非水平与非垂直方向上的边缘与纹理能量;利用抽样函数完成分数像素上的插值预测。针对遥感图像的实验表明,与传统的提升小波变换相比,新算法获得的重构图像无论峰值信噪比还是主观质量都有显著提高,对今后遥感图像的压缩编码研究具有重要价值。
Traditional lifting wavelet transform can not effectively reconstruct non-horizontal and non-vertical high-frequency information in remote sensing images, resulting in the high-frequency wavelet coefficients still prominent in these places, reducing the coding efficiency of remote sensing images. A new lifting wavelet framework (DOLW) based on direction optimization is proposed. The gradient-based direction prediction model is designed to obtain the optimal transformation direction of the lifting wavelet. The image is upgraded and transformed in the direction of the first vertical and horizontal direction after the optimal transformation, and the edges in non-horizontal and non-vertical directions in the high frequency sub-band of the remote sensing image are weakened And texture energy; using the sampling function to complete interpolation prediction on fractional pixels. Experiments on remote sensing images show that compared with the traditional lifting wavelet transform, the reconstructed image obtained by the new algorithm has a significant improvement both in peak signal-to-noise ratio and subjective quality, which is of great value in the future research on compression coding of remote sensing images.