Abstract: To overcome the limitations of traditional threshold algorithm, an algorithm for adaptive segmentation is proposed. According to the basic concept of the algorithm, a window is defined for every pixel with itself as center, the maximum and minimum value of the pixels in the window are calculated and their average is used as the threshold for image segmentation. Two different image segmentation methods are used, adaptive threshoding and iterative threshoding. The experimental results show that adaptive threshoding achieves higher performance and robustness in practical applications.