Technology Opportunities

Fast method eliminates need for repeated reconstructions of latent image

Highlights

• In many practical scenarios, such as hand-held cameras or ones mounted on a moving vehicle, it is difficult to eliminate camera shake. Sensor movement during exposure leads to unwanted blur in the acquired image.
• Current approaches are based on repeated reconstructions of the latent image, which makes them slow. Furthermore, many methods rely on explicit assumptions that do not always hold and thus undermine the accuracy of the estimated kernel and deblurred image.
• There is therefore a need for a faster, more accurate method for removing blur in digital images.

Our Innovation

Purely statistical approach to recover blur kernel in motion-blurred natural images by extracting a set of statistics from the input image and using them to recover the blur.

Figure: Kernels estimated by different methods and the resulting deblurred images

Key Features

• More robust and accurate recovery of blur kernel.
• Able to cope well with images containing under-resolved texture and foliage clutter in outdoor scenes.
• Input image only accessed once to extract small set of statistics, so the technique depends mostly on blur kernel size and does not scale with the image dimensions.
• Method achieves highly accurate results in scenarios that challenge other approaches, at fast running times.
• Method does not rely on the presence or detection of well-defined step edges at multiple orientations as required by other methods.

Development Milestones

• Seeking licensing opportunities

The Opportunity

• By 2016, the digital photography market should reach $82.5 billion.
• Market researchers have predicted that by 2013, nearly 70% of handsets sold in the United States will come equipped with a camera of 2-megapixels or more and that consumers will respond to this trend by making greater use of mobile imaging software, including photo-editing.