Image thresholding is one of the main techniques for image segmentation. It has many applications in pattern recognition, computer vision, and image and video understanding. This paper formulates the thresholding as an optimization problem: finding the best thresholds that minimize a weighted sum-of-squared-error function. A fast iterative optimization algorithm is presented to reach this goal. Our algorithm is compared with a classic, most commonly-used thresholding approach. Both theoretic analysis and experiments show that the two approaches are equivalent. However, our formulation of the problem allows us to develop a much more efficient algorithm, which has more applications, especially in real-time video surveillance and tracking systems.