Fusing range and 2-D images in multi-sensor for robot vision

Abstract

A multi-vision-sensor for an intelligent robot has been introduced, which integrates a 3D surface profile based on a synchronized laser scanner, one imaging laser radar and two revolving CCD cameras. The work mode of multi-vision-sensor, applied within the different ranges, has been put forward. The feature-based registration among medium range and 2D CCD images has been obtained, and the mapping relationship between the 3D surface profile and 2D CCD images has also been discussed. Moreover, the matching relation between two CCD images can be achieved corresponding to two methods, including the same features by registration within medium range and the common mapped pixels on the 3D surface within near range respectively, so that the intersection measurement based on the two CCD images can also be realized. Fusing range and 2D images in this multi-vision-sensor can be ensured for the different functional modules of the intelligent robot vision.