Super Resolution Overview Super-Resolution (Super-Resolution) that is through the hardware or software to improve the resolution of the original image, through a series of low-resolution images to get a high-resolution image process is super-resolution reconstruction. The core idea of super-resolution reconstruction is to use time bandwidth (to obtain the same scene of the multi-frame image sequence) in exchange for spatial resolution, to achieve time resolution to spatial resolution conversion.
In the process of digital image acquisition, due to the limitations of machine equipment performance and shooting conditions, will make the collected image resolution is low. This image is more vague, for the latter part of the treatment, the application has a greater impact, so improve the image resolution is that we must solve. The most straightforward way to improve resolution is, of course, to use higher resolution devices, but there are two problems: one is that high-resolution devices are expensive; the other is that each device has its limits, It is difficult to get a truly high-resolution image. Therefore, we can consider the use of software to improve the resolution of the image, which is discussed in this paper to super-resolution (SR: Super-Resolution) image reconstruction.
Super - resolution implementation technology
There are several ways to implement a super-resolution technique. Here we describe several of the most commonly used methods:
1) based on interpolation. This method is the most intuitive method in the current super resolution study. By comparing the multi-frames of the image, the relative relation information between them is obtained, and the pixel value of the high-resolution image at the non-uniform pitch sampling point is obtained. And then through the non-uniform interpolation method, after a certain interpolation, you can get a high-resolution images. Of course, this will get the image noise, fuzzy and other issues, it can be through the image recovery technology for some repair.
2) based on reconstruction. The method mainly has two key steps of registration and reconstruction. In the case of registration, the use of multi-frame low-resolution images as a constraint on data consistency, so you can get other low-resolution images and reference low-resolution images between the sub-pixel precision relative motion. When reconstructed, the target image can be optimized using the prior knowledge of the image. The common algorithms of this method are iterative direction projection, maximum a posteriori probability, convex set projection and so on.
3) based on learning. The premise of this method is that the low-resolution image completely has information for reasoning to predict its corresponding high-resolution portion. This allows a low-resolution image set to be trained to produce a learning model that can compute the image's high-frequency detail information. At present, the commonly used learning algorithm is Freeman et al's Example-based method, Chang et al proposed based on the method of neighborhood embedding and so on.
Super - resolution application scenarios
Super-resolution image reconstruction has a wide range of uses in real life. Here, we list some of the places used in life:
1) Digital HD. In the field of digital television, the use of super-resolution reconstruction technology can be digital television (DTV) signal into high-definition television (HDTV) receiver to match the signal, thereby enhancing the audience experience, video surveillance.
2) Medical images. In medical care, high-resolution medical images are very helpful for doctors to make the right diagnosis. So the use of super-resolution reconstruction to get a clearer image, will make the doctor more accurate, effective, nuclear magnetic resonance imaging.
3) Satellite image analysis. In the military, meteorological field, the use of high-resolution satellite images is very easy to distinguish similar objects from similar objects. So you can use the super-resolution reconstruction technology to obtain high-resolution images, better serve the military security and daily life, satellite imaging: remote sensing, telemetry, military reconnaissance.
4) safety testing. Banks, residential areas, road junctions are the need for safety testing. Although these places will generally install the camera, but the images are very vague. The use of super-resolution reconstruction technology, will help staff get a clearer image, that can help the usual security management, but also in the case is to help the police handling the case.
5) video format conversion, video enhancement and recovery: the old film of the reorganization;
6) micro-imaging, virtual reality and so on
The effect of super resolution
In many cases, people want to get a higher image resolution, the picture is more clear. However, the actual hardware conditions, as well as other factors, we get the resolution of the image can not meet the requirements. We can use the software-based approach, the use of ultra-resolution image reconstruction technology, the image optimization, repair. Through the following pictures, we can see the effect of super-resolution.
1. In Figure 1, the use of large resolution technology to show the vase pattern vague text.
2. Figure 2 shows the application of the super-resolution method in medical images, showing a relatively clear image using super-resolution reconstruction techniques using multiple low-resolution image sequences (carotid MRI).
Figure 2 shows a clear picture of the carotid artery through a set of low-resolution image sequences (carotid MRI)
3. Figure 3 shows the use of super-resolution technology in streaming video enhancement, with limited video bandwidth as much as possible to improve the clarity of the video.
a. A set of low-resolution sequences in a group of sporting events, b. is a high-definition video that is processed by a super-resolution technique
Figure 3 with super-resolution technology to improve video clarity
Super - resolution future development
Because of its wide use of its own, people in the last 20 years of super-resolution image reconstruction technology for a lot of research, the technology has also been rapid development. We have used the technology in satellite weather, medical imaging, image compression and so on. But there are still many problems to be solved in this field, and the degradation model, motion estimation, reconstruction algorithm and real-time application will be the focus of future research. But we have reason to believe that the future will be more and more areas thanks to the further development of super-resolution technology, it can be said that the development of super-resolution image reconstruction technology has been and more in-depth impact and improve our lives All aspects.
Copyright © Jaste Solar Technology Co.,Ltd All Rights Reserved.