The embodiments described herein provide devices and methods that facilitate improved input devices. In one embodiment, an input device is configured to determine if a sensed object includes an occluded portion using an image representative of sensor values, determine a portion of the image corresponding to the sensed object, and approximate a boundary representation of the sensed object wherein if the sensed object does include the occluded portion, the boundary representation encompasses at least part of the occluded portion of the sensed object and at least part of a non-occluded portion of the sensed object. The determination of a boundary representation corresponding to a sensed object facilitates improved characterization of the sensed object. For example, the determined boundary representation may be used to more accurately track changes in the position of the sensed object as the sensed object moves out of the sensing region.