Algorithm to correct motion blur

The blur due to a wrong adjust of the objetive of the camera at the time of take the photo, can be fixed slightly applying a focus filter, that really do is overlap of a lowpass filter of the image over itself. The lowpass filter, deletes or attenuate the soft changes between pixels in the image, leaving a grey image that can be superimpose, with different technics, over the original image to create the focus effect. This isn't a perfect tecnic, because leave all focused in the same way, removes the depth sensation of the images.

There are better algorithms than the lowpass filter to focus images a lot of more blurred. These technics are mathematically and computationally complex as the "simulated annealing". The problem is that these algorithms can take several hours in get a result, because try to find the solution testing nearly all the answers, so isn't viable its application in programs of photographic retouching. Due to that, its use is relegated to the correction of atmospheric distortion in space images take it from the Earth.

The new Adobe algorithm is different. It don't try blurs caused by the lense but the produced by the pulse of the person the take the photo. This uses in first place a flow optic algorithm to determine the movement that follow the camera while was taking the photo. After undo the movement over the photo.

In this video of the conference Adobe MAX 2011, that ironically is so blurred, you can watch a presentation of the functionality of the algorithm.

Really the idea isn't new, in the last 5 years or more, have appeared a lot of scientific papers with algorithms of this kind, but this will be the first implementation of one of them in a commercial software.