Investigating the Efficiency of a Three-Dimensional Facial Tracker Using Stereo Camera Arrangements

Throughout the history of visual tracking technology, it was generally thought that smooth object movement occurred without abrupt changes. However, significant technological developments have been recently made in this field; algorithms have been developed to tackle issues like drifting, abrupt appearance change, and disappearance from scenes. In visual tracking, a 3D world is projected onto a 2D image. This results in a loss of information and various other problems including cluttered backgrounds, partial or full occlusions, realtime processing requirements, image noise, complex object motion, and illumination changes. Issues that may arise in visual tracking include partial and full occlusions, changing illumination, cluttered backgrounds, and complex object motions. Face tracking systems require basic gestural understanding as defined by the interaction environment and basic facial understanding. A system's “robustness” refers to the property that enables it to keep track of a concerned object even under complicated situations. Obtaining robust facial features that are accurate and easy to find requires evaluating multiple options and techniques. Numerous methods have been proposed and developed to combat these challenges. This paper presents an extensive analysis on the performance of stereo camera arrangement methods in detecting and tracking faces. The methods investigated offer a state-of-the-art performance in terms of improved speed, more effective handling of occlusions, and a higher quality of resulting images.