Given two views of the same point and a known 3D transform checks to see if the point is in front of both cameras. This is the positive depth constraint. A class is provided instead of a function to reduce computational overhead each time the function is called. Memory only needs to be declared once. Also less chance of messing up and only checking one view instead of two views if you use this class.
Triangulation is done in homogenous coordinates so that points at infinity can be handled
COORDINATE SYSTEM: Right handed coordinate system with +z is pointing along the camera's optical axis,
checkConstraintChecks to see if a single point meets the constraint.
viewA- View of the 3D point from the first camera. Calibrated coordinates.
viewB- View of the 3D point from the second camera. Calibrated coordinates.
fromAtoB- Transform from the B to A camera frame.
- If the triangulated point appears in front of both cameras.