According to the characteristics of forest fire spread fast, difficult to save, in order to determine the forest fire danger rating quickly and strive for fire time, it is important to calculate the fire area. Considering that there is a close relationship between the gray image of infrared image with temperature, the gray value of infrared image corresponds to high temperature object will be large. At the same time fire occurs, the temperature of the flame is generally higher than the temperature of the surrounding environment , infrared image can be used to exclude a lot of non-fire interference source and provides a simple and convenient criterion for the identification and segmentation of flame area, so this paper presents a calculation method of fire area based on infrared image. The method take full advantage of the significant characteristics of the obtained forest fire infrared image, the coordinate relationship between the image and the camera is established and the camera is calibrated. Then extract the edge of the forest fire spread, utilizing sobel operator rough location, and then grayscale image interpolation, cubic spline interpolation function so that the target to achieve sub-pixel level grayscale images after interpolation, the use of the maximum variance between the threshold is determined to achieve sub-pixel edge detection. The edge of the forest fire spread is extracted, with images used to calculate the forest fire burned area, with the error of measuring result calculated. The results show that the maximum error is controlled within 4.5%. Therefore, the forest fire area calculation methods can be applied to the fire control and post- disaster assessment of forest fire. The method is feasible and convenient calculation and will be of great significance to the forest-fire prevention.