fbpx

Researchers Develop World’s Best Image Cleanup System

The RCM algorithm – original image on the left – fixed image on the right (click to enlarge)
Researchers from the University of Texas developed a revolutionary method for enlarging and enhancing photos. The team created a special free website which allows users to experiment with the algorithm. We took a test drive and got back with initial impressions.
The current research actually began not as an algorithm for improving images but rather an effort to better understand how animal and human visual systems work. According to Professor Wilson Geisler from the Department of Psychology at the University of Texas at Austin: “Our eyes evolved to process natural signals, so to understand how the visual brain works, we measured the real signals in the natural world — the environment in which the brain is operating. In the process we realized that many tasks the eye performs are the same as what photography requires”.
The University of Texas team developed an approach to image processing based on natural scene statistics rather than an image-by-image method other image processing system use. According to Geisler: “Compared with other photo-enhancement algorithms, we believe this is the best in the world for reducing noise and enlarging images, and also about 50 times faster”.
Besides improving images by reducing noise, the technology developed by the team can also have applications for enlarging and enhancing all types of images. According to Geisler users who are working with very large images will benefit significantly with the high speed of the service.
We have decided to run a quick test of the University of Texas algorithm to see how well it cope with high levels of digital noise. It is important o understand that reducing noise is fairly easy (all camera manufacturers have this feature built into cameras and you can reduce noise using many free and commercial image editing software. The big problem however is maintaining as many details as possible while removing the digital noise. This s actually a very difficult task and this is exactly what we wanted to test with the new algorithm.
Before we talk about the Algorithm itself and how it performs, we wanted to comment on the current web interface. As it happens, the current service has a long way to go before it can reach the standards of a commercial product. The GUI is basic and not very intuitive, performing tasks take time and you do not get any notice that something actually happens.
On the other hand when it comes to what actually matters – reducing noise in an image with minimal reduction of detail, things look much better. We tested an image we shoot at night with a Nikon D5000 with ISO set to 3200 (more or less the maximum you can push the camera without an artificial boost). The original image had a lot of noise – after applying the highest level of “denoise” (very high) on the RCM service, we got an image back which was not free of noise but significantly cleaner. More importantly – much of the details of the original image were preserved.
To sum things up – from our experience the algorithm does seem to do its magic. Whether or not it is “the best of its kind” is an open question but this is definitely something you would like to see in a commercial product.
More information can be found on the University of Texas website. Feel free to test the free algorithm yourself on the RCM tools website.

Related Posts