The authors propose a method for realizing a high accuracy image target centroid--center of mass (COM) detection system. To do this, they make use of an adaptive K-winner-take-all circuit [1]. They also use a two-dimensional (2D) dynamic element matching algorithm [2] intended for image sensor arrays. The system suggested by the authors yields a high accuracy COM position for the most prominent target in a programmable active region of the field of view, for use in tracking stars. This makes it appropriate for real-time applications. The system permits target selection and locking, along with the potential for tracking manifold targets.
Fish et al. make use of a separability property of the COM in order to diminish the complexity involved, and employ one-dimensional circuits. High output accuracy is ensured by using a dynamic element matching algorithm. This algorithm is often used in circuits that transform analog signals to digital signals, and vice versa. The authors perform some simulations using the MATLAB software and real images, in order to show that high accuracy can be obtained. They also propose a potential low-level hardware realization.