Thresholding method is one of the most common methods for image segmentation, however; thresholding methods based on 1-D histogram are easily ruined by the noise. Thresholding based on 2-D histogram and Fisher criterion function can overcome the shortcoming of 1-D threshold method, which has better segmentation performance. But due to huge computation is required for 2-D Fisher criterion function its speed is very slow. Commonly used optimization methods to speed up thresholding based on 2-D Fisher criterion function like particle swarm optimization and genetic algorithm are easily to fall into the local optimum. Cuckoo search is a newly proposed meta-heuristic optimization algorithm; testing results on some benchmarks indicate that cuckoo search has better global convergence ability than particle swarm optimization and genetic algorithm. In this paper, by employing cuckoo search algorithm, a segmentation approach is proposed based on 2-D Fisher criterion function. The experimental results show that the proposed method significantly decreases the seeking time of optimal threshold, which is a well performing method and is more suitable for real-time image segmentation.