The Flathead Valley’s Leading Independent Journal of Observation, Analysis, & Opinion. © James R. Conner.


6 August 2012

You’re on Trailhead Camera, part 2 — deconvolving the smile.

Last week I reported that the trail cameras employed by the current Glacier visitor use study are defocused so individuals cannot be identified. That’s been the protocol for a long time.

But after corresponding with the study’s director, University of Montana professor Wayne Freimund, there was one loose end: Freimund said he had not run tests to determine whether image processing technology — deconvolution — could bring blurred images back into focus.

Twenty years ago that was not an issue. Deconvolution has been around since the 1950s, but it’s so CPU intensive that until recently only mainframes had the processing power to deconvolve an image in less than a human lifetime. That’s changed. Now even my little Mac Mini has the muscle to deconvolve megapixel magnitude images in just a few minutes — and deconvolution software is available for just a few hundred dollars.

So, I decided to see whether I could bring severely blurred images back into focus. I shot six images of my Heidi and Jemima test subject; one sharply focused, the rest defocused to varying extents. I processed the images with Adobe Photoshop CS5 Extended and Pixinsight, an astrophotography application with more powerful deconvolution algorithms than Photoshop.

You’ll find the six originals at the end of this page. The originals and my attempts at extracting a sharp focus are on a very slow loading page.

Optimized for fast processing, Photoshop’s Smart Sharpen function employs one-step deconvolution (probably Wiener deconvolution) and integer math. Optimized for power, Pixinsight provides iterative deconvolution (Richardson-Lucy and Van Cittert), and a one-step Restoration Filter (Wiener deconvolution; also, a least squares solution). Pixinsight also offers a 64-bit floating point format, a significant advantage over Photoshop. I used 16-bit images in Photoshop, 64-bit images in Pixinsight.

My tentative conclusion? Even consumer grade software has the power to refocus severely blurred images. It might be a good idea for researchers employing defocused cameras to buddy-up with their colleagues in astronomy and electrical engineering to determine how much protection current degrees of defocusing trail cameras actually provides.