Analytic inverse of the supplied pinhole K, mapping pixel homogeneous coordinates to camera rays.
Row-major 3 by 3 supplied pinhole intrinsic matrix K in the resident pixel-centre chart.
Analytic inverse of the supplied pinhole K, mapping pixel homogeneous coordinates to camera rays.
Row-major 3 by 3 supplied pinhole intrinsic matrix K in the resident pixel-centre chart.
Row-major 3 by 3 camera rotation in the operation's declared source-to-camera convention.
Camera translation in the operation's admitted reconstruction scale; two-view direction alone has no metric length.
Signed point depth in the first camera's translation-scale coordinate frame.
Pixel discrepancy between the first observation and its point reprojection.
Angle between the two viewing rays for this triangulated point, in radians.
Physical point in the source frame; camera triangulation uses the admitted translation scale.
True only when the triangulated point lies in front of both cameras.
Signed point depth in the second camera's translation-scale coordinate frame.
Pixel discrepancy between the second observation and its point reprojection.
Triangulate one observation pair in the first-camera, unit-baseline gauge. Original 4-by-4 DLT uses calibrated rays and P1=[I|0], P2=[R|t]. An unresolved third design direction, point at infinity or nonfinite divided coordinate returns null; a finite point behind a camera remains explicitly nonpositive. Invalid or overflowing matrix actions and SVD nonconvergence throw. This algebraic point is fitting evidence, not an independently observed surface.