基于图像类推的遥感图像超分辨率技术

Image-analogies based super resolution for remote sensing images

由于卫星上相机距离拍摄景物较远,所以遥感图像分辨率一般较低。为了获得较高分辨率的图像。提出将图像类推技术(IA)与立方卷积插值法相结合的方法;并在学习样本集合建立过程中引入一种新的思路,直接对高分辨率图像的高频细节信息进行学习。实验结果表明,该方法不仅提高了放大图像的清晰程度,而且较一般的图像类推算法,能产生更为合理的细节以增强图像,使放大后的遥感图像更为逼真。

Due to the long distance between satellite camera and the site, the resolution of remote sensing image is low. In order to get high resolution image, Image Analogies （IA） technology and cubic convolution interpolation algorithm were combined, and a new idea that studied the high-frequency details of the high resolution image directly in the learning set was proposed. The experimental results show that this method can not only amplify images much more sharply, but also generate more reasonable details to enhance the image than general image analogies method. Thus, the amplified remote sensing image becomes more vivid.