Polarization-based shadow analysis

Capturing hidden details with polarized light 

 

Problem: 

Most cameras only capture information on the intensity and color of light. Polarization is another characteristic of light that can provide useful information, but conventional cameras do not provide polarization data. Shadows pose a difficult problem for machine vision and image analysis techniques, because they can hide objects and make edges unclear. Polarization information could provide enhanced object recognition, edge recognition, and detection of shadow-occluded objects.

 

Solution: 

Researchers at the University of Pennsylvania have developed an imaging technique that captures data at 3 different polarization angles using a standard camera. This method of data capture shows more information that would not be detectable by the unaided eye or a normal camera.

Polarization analysis can detect objects in shadows, segment and separate complex overlapping shadows from multiple sources, and help detect edges and hidden details that would not be visible in standard intensity-based images. This process could provide better object recognition for systems that require visual feedback to navigate and interact with the environment.

 

Advantages: 

  • Improved analysis for images with shadows
  • Detection of objects in dark areas of an image
  • Implementation with standard cameras

 

Stage of Development: 

Prototype

 

Intellectual Property: 

US Patent 8,396,319

 

Desired Partnerships: 

License

 

Docket # S4277

 

Patent Information: