OTSU is one of the most commonly used image-threshold methods, which based on one-dimension histogram of image’s gray level. The two-dimension OTSU takes the gray level of pixel point, as well as the spatial information between the pixel point and its field into consideration. Because of infrared image, different from the general characteristics of gray image, the two-dimensional OTSU segmentation is targeted adopted. According to the comparison with one-dimension OTSU, it is indicated that the two-dimension OTSU can segment the near-infrared image better and get the furnace image more effectively. By monitoring its temperature in the late period and alarming through sound and light, it greatly enhances the security of high- heat furnace.