A method of dynamically adjusting and predicting a segmentation threshold is provided. The method comprises calculating a threshold factor, calculating an actual average image pixel intensity for an entire image frame, and predicting a future average image pixel intensity for future time t, wherein the predicted future average image pixel intensity is based on past actual average image pixel intensities. The method also comprises combining the predicted future average image pixel intensity with the threshold factor to provide an adaptive threshold that adapts to changes in pixel intensity due to movement of an object in the image frame.