Multiple view triangulation with improved robustness to observation errors

A device for determining a location of a scene point in a global coordinate system, said device comprising an input configured to receive location data for image points that represent the scene point in a plurality of overlapping views, and a processing system configured to:

• obtain the location data from the input, the location data comprising location coordinates of a respective image point in a local coordinate system of a respective view, the local coordinate system comprising a distance coordinate axis that is perpendicular to a predefined image plane of the respective view;
• obtain a transformation matrix of the respective view, the transformation matrix defining a transformation operation between the local coordinate system of the respective view and the global coordinate system;
• and
• determine the location of the scene point in the global coordinate system;
• wherein the location is determined by minimizing a plurality of differences between first and second estimated locations of the scene point in the local coordinate system of the respective view, wherein the first estimated location is given by a linear scaling of the location coordinates of the respective image point in the respective view, and the second estimated location is given by operating the transformation matrix of the respective view on the location of the scene point in the global coordinate system;
• and
• wherein the plurality of differences is defined to penalize differences that comprise the linear scaling as applied along the distance coordinate axis.