Figure 2From: Modeling and Visualization of Human Activities for Multicamera NetworksConsider views A and B (camera centers C A and C B ) of a scene with a point x imaged as u A and u B on the two views. Without any additional assumptions, given u A , we can only constrain to lie along the image of the preimage of u A (a line). However, if world was planar (and we knew the relevant calibration information) then we can uniquely invert u A to obtain x, and reproject x to obtain u B .Back to article page