Breaking news events are frequently captured using mobile phone cameras.  Typically, the person holding the camera is excited and moving independently of the action being recording.  This combined motion, along with the camera's automatic focus and progressive imager, can make portions of the video appear hopelessly blurry.  Fortunately, a new video enhancement technology called "motion deblurring" (aka "video deblurring") has emerged to help remedy this problem.

First, each frame of the target video is analyzed to determine the horizontal and vertical motion paths.  This combined motion path (called a "kernel") often has an oddly curved shape that must be negatively applied to the video to reverse the motion blur.  Although the concept is intuitive, the math involved is extremely complex and it can take hours to correct one minute of video.

Newer algorithms are emerging that will dramatically expedite this process by using adjacent frames to estimate, and further refine, each frame's kernel.  This becomes especially beneficial as some video segments may not suffer from any camera motion blur.

The blur corrected video can then be run through a second process to determine the motion path of all the moving objects (e.g. people and vehicles) within the video.  Each object is then isolated and assigned its own motion kernel corrections.  If the object's rate of motion exceeds the imager's capture speed, those objects may have a warped appearance.  Although a separate issue, this can only be corrected once the motion deblurring process is complete.

At this point, traditional enhancement steps like pixilation suppression, focus correction, video stabilization and pixel fusion can be successfully applied.  If this is then followed by cropping, enlargement and lighting corrections, the results can be stunning.

Motion deblurring is in the early stages and still requires significant user interaction.  Attempts to automate this process are unsuccessful when the video is heavily pixilated, has frequent stop-start points or sudden changes in lighting.  As the process matures, these obstacles will become less significant.

Currently, only research labs and video enhancement service providers have full access to this technology.  The better video enhancement labs will offer to improve a brief portion of your video without cost or obligation so you can make an informed purchase decision.  With a little comparison shopping, you can ensure that you are receiving the clearest results available for your video.

Motion deblurring is the latest technology in a rapidly advancing forensic field.  Over the coming years, several companies will begin offering software to automate the entire process.  Competition will quickly drive pricing down while faster computers will shorten the required processing time.  Eventually this technology will be integrated into our mobile phones to allow even the shakiest hands to shoot professional video.

Recommended continued reading  HERE

Doug Carner is president and lead technologist at Forensic Protection
Video deblur example here.  Originally posted here (now inactive).