引入视觉注意机制可变分辨率的遥感图像压缩

Variable resolution remote sensing image compression based on visual attention mechanism

图像压缩是遥感图像处理的重要研究领域,现有的压缩方法要么丢失重要的细节信息,无法满足实际的应用需要,要么压缩率过低,难以达到实时处理的要求。将视觉注意机制引入到遥感图像压缩中,对不同的显著性区域采用不同的压缩率,这样不仅可以对整个遥感图像达到一个高的压缩率,而且还可以保持重要区域的高分辨率,实现了可变分辨率的图像压缩。实验结果表明在前几个显著性区域中,该方法得到的图像压缩性能指标优于传统压缩方法得到的性能指标。

Image compression is one of important research fields in remote sensing image processing. Existing compression methods either miss important details or adopt low compression rate, which can not meet the actual needs of the application and it is difficult to achieve real-time processing requirements. This paper proposes a remote sensing image compression method based on visual attention mechanism. Various compression ratios are adopted according to the different saliency regions. By this method, it not only achieves high compression ratios for entire image, but also keeps high resolution for important regions. Experimental results indicate that in first few saliency regions, the image compression performance indicators of the method are better than that of conventional methods.