Abstract: In a video based image recognition system, an adaptive threshold level for each pixel is set based upon a running horizontal and/or running vertical average of the pixel intensities combined with suppression of the running average threshold computation to hold the threshold value constant upon the detection of pixels with an intensity above the threshold, and continuing until the detection of a pixel having an intensity lower than the held constant value. After the edge of an image has been detected, and thereafter upon detecting the first pixel having an intensity less than the held threshold, a formula uses a particular combination of pixel location values to compute an intermediate running average value, after which the normal running average is computed.

Abstract: A tracking system 10 utilizes both a correlation processor 18 and centroid processor 20 to generate track error signals. The operator is given the option to initially select track error signals exclusively from the correlation processor 18 at track initiation. In a mixed mode of operation, track gate error signals from the centroid processor 20 are selected for each axis where the target is contrast bounded while track gate error signals from the correlation processor 18 are selected for each axis where the target is not contrast bounded.