Computational Vision

Where computational photography meets computer vision (with a touch of AI)

Computational photography goes beyond traditional photography and here’s the difference. Traditional cameras capture 2D images and output pixels, whereas in computational photography multiple image sensors capture depth and reflection characteristics of what’s in view. Conventional photography captures pixels, whereas computational photography works with light fields, rays of light flying in space. This is Michael Faraday’s dream come true: in 1864 he suggested that light should be understood as a field, just like a magnetic field.

Image processing techniques for computational photography include: brightness and contrast adjustments; image compositions (tone mapping, exposure/focus combining); image manipulation: (denoising, restoration, cropping, alignment, and feature tracking).

Computer Vision is the study of how the machines see things. It enables cameras and computers to replace the human eye and recognize, track, and measure objects. No surprise that Artificial Intelligence (AI) has its thrust from computer vision. Now AI-powered vision systems are the go-to tools for extracting information from images and light fields. Typical uses include: industrial robots; autonomous vehicles; safety monitoring; image organization (database retrieval); industrial inspection; medical image analysis;  and landscape modeling. The signal processing algorithms for computer vision range from object detection and recognition, to face analysis; optical character recognition; feature detection; feature tracking; and image registration.

When combined and powered by AI, the two disciplines above open up vast opportunities for providing groundbreaking multimedia experiences to humans and machines. RayShaper’s mission is to contribute to this exciting field and commercialize its emerging discoveries.