A system and method for determining positions of multiple targets on a planar surface is described. The targets subject to detection may include a touch from a body part (such as a finger), a pen, or other objects. The system and method may use light sensors, such as cameras, to generate information for the multiple simultaneous targets (such as finger, pens, etc.) that are proximate to or on the planar surface. The information from the cameras may be used to generate possible targets. The possible targets include both "real" targets (a target associated with an actual touch) and "ghost" targets (a target not associated with an actual touch). Using analysis, such as a history of previous targets, the list of potential targets may then be narrowed to the multiple targets by analyzing state information for targets from a previous cycle (such as the targets determined during a previous frame).