Multiple Camera Sensor System

Offentlig prosjektsammendrag

The project aims to develop an innovative industrial multiple-sensor system that reinvents the traditional camera and how it is being used for multiple applications across different industrial sectors. There are several challenges related to existing camera technology, often related directly to the specific field in which the camera technology is being used. These challenges are mainly related to the camera ´s field of view and the need for moveable cameras to cover an area of interest. Using a multiple camera sensor system where several image sensors with overlapping fields of views one can obtain sensor data from a larger area. This information can be processed and composed into a virtual perception of viewpoint covering an area of interest. Basing research and development on adaption of existing technologies, this project desires to (i) bring forward a multiple camera sensor system with n image sensors with overlapping fields; (ii) develop an interface for concurrent sensor data acquisition; (iii) develop algorithms for image stitching where two images are being stitched into one larger image; and (iv) process and produce a virtual video output frame within the stitched image. Image stitching does in some way simulate how the human brain processes image data collected by two eyes. The human vision is based on images from two retinas (two viewpoints) that are split vertically and assembled in the human brain into a coherent and plausible virtual representation of the environment (binocular stereopsis). Stitching together image data from a multiple camera sensor system with n image sensors (corresponding to n retinas) one can assemble a rather large virtual representation of the environment. The anticipated R&D project is covering challenges within acquisition and processing of large amounts of sensor data in real-time, in addition to producing plausible virtual perceptions.