Blurry photos: New program fixes the unfixable
Vladimir Yuzhikov, a computer engineer specializing in image and signal processing, has created software called SmartDeblur, a free software used for de-blurring images.
In the days of my misspent youth, I was a semi-professional photographer. This essentially means that people occasionally paid me money for photos (and not the kind taken after bursting out of the closet of a cheap motel room).
In learning the technical aspects of the art — then done with light-sensitive film and chemicals — I was told that most exposure errors could be fixed (at least to a certain extent) in the darkroom. Underexposed film could be “pushed” in processing, and overexposures remedied in printing.
What couldn’t be fixed was bad focus. You miss your focus with the lens, and you need a mulligan. That has held true for the digital photography era, but there may now be a fix for that, as well.
Vladimir Yuzhikov is a computer engineer specializing in image and signal processing. He clearly knows so much more about the science of reconstructing poorly focused images than I do that I’m not even prepared to try and explain how his method works.
He offers to explain it on his website and the explanation includes several complex formulas I might have been able to work out when I was 18 years old — then again, maybe not. Even if we had the formulas then, the fastest computers then in existence wouldn’t have had a prayer of running them. Now, the processor in your laptop can handle it.
SmartDeblur is his free, no-questions-asked software used for (as the name suggests) de-blurring images. It doesn’t even have to be installed on the computer that runs it. You open the application from an executable file and are greeted with a window and several slider controls.
Load your blurry image into the program, and start monkeying with the sliders. A progress bar at the bottom of the window moves to show how the image is being processed according to each change in parameters, and on my reasonably powerful desktop machine, each change took about 15 seconds to resolve.
The software has its limits. If an image is too blurry, the best result you can get is still pretty useless. But if you’re trying to see detail that is just out of reach in the blurred image, you may be able to clear it up enough to see what you need to see.
The example furnished by Yuzhikov simulates a car license plate that is illegible in the original, but easily read in the processed version. His website includes several other images processed by his software, and the results are impressive.
Adobe Photoshop has included an “unsharp mask” filter for some time that can simulate sharpness in a softly focused image, but it doesn’t do much to clear up details. It’s better adapted for removing artistic blur effects from photos and making them look more utilitarian and businesslike. SmartDeblur uses an entirely different approach to the problem.
The current version of the software is free, but I expect it will soon be licensed to a commercial developer and improved so that detail will be recoverable from images even more poorly focused than the example. What it won’t do is produce information that isn’t there.
Popular crime TV shows depict the gearhead techie grabbing a tiny portion of a video frame and “enhancing” it to produce a portrait-like image of the bad guy, reflected in the chrome bumper of a passing car. Until we have video hardware capable of recording 50 megapixels of data in every video frame (a standard video frame has about 307,000 pixels), that ain’t happening.
Get this software while it’s still available for free. You may not need it today, but a future case might include a blurry still image you want to clear up for detail.