Fast video image sampling __ image processing

Summary

We propose a simple and efficient method of sampling on the top. This method can automatically enhance the resolution of the video image, while maintaining the important structure information of the image. The main advantage of our approach is a feedback control framework that can accurately restore high-resolution images from low-resolution images without imposing local structure constraints on images learned from other samples. This makes our approach independent of image quality and high quality images learned through mass sampling. Usually a large number of sample learning algorithms, can produce high-quality image quality without detectable unsightly artificial traces. Another advantage of our approach is that it can be naturally extended to the top sampling of the video, while the transient continuity of the video can be maintained automatically. Finally, our algorithm runs very quickly. We demonstrate the effectiveness of our algorithm through different video image data.

Note: This article is my 10 years of translation of the Chinese University of Hong Kong Jayaya published in SIGGRAPH ASIA 2008 article, a lot of local translation is not good, please forgive me.

Please download the translated version from here.

Click to open the link

Original Download Address: http://appsrv.cse.cuhk.edu.hk/~leojia/projects/upsampling/index.html

Their processing results show that the sample is pretty good and they claim to be able to process the video in real time. But in the absence of the GPU is very slow and slow, on my PC test, the 720p image magnification twice times a few 10 seconds.

Referring to their ideas, I used a sparse prior distribution of deconvolution algorithm to achieve a bit, the actual effect is not their good, but better than bicubic. And their treatment effect is much better than bicubic, and their test diagram is as follows: