The enhancement and tracking of moving objects in digital images using adaptive three-dimensional recursive filters

Abstract

A three-dimensional (3-D) adaptive filter control system is described which is capable of seeking, tracking and enhancing an object that moves with time on an arbitrary, but smooth, trajectory in a dynamic digital image. Enhancement of the object is performed by employing a time-varying first-order 3-D planar-resonant recursive filter that is adapted via its coefficients at each time interval to correspond to the trajectory of the object. The trajectory of the object is determined at each time interval by tracking the intensity centroid of the object at the output of the adaptive 3-D recursive filter. The system is initialized for tracking by a seek algorithm which searches the image for the first candidate object having uniform motion and intensity. Experimental evidence is provided showing the effectiveness of the system in locating, tracking and enhancing objects in images containing other objects and high levels of noise.