中文摘要：

红外热像仪常见的图像格式为IMG，其编码规则有别于通常意义下的IMG图像，经研究发现：图像文件中保存每个像素点信息的数据（本文中称其为像素值）是随着该点的实际温度值单调递增的：利用神经网络得出像素点的实际温度与像素值间的函数关系，仿真误差低于0．0001；据此构造温度值与灰度色彩的非线性关系，本文最后根据BMP图的编码规则，将红外热图转换成Windows下通用的图像格式，从而突破了热像仪器配套软件的禁锢，又可以得到图像最原始的温度信息，为进一步的红外图像处理打下了基础。

英文摘要：

IMG is a common image format used by various infrared thermographs,and its coding is different from the optical IMG imge. It is proved by the research that the decimal data of each pixel in the infrared image file increases by degrees monotonously as the temperature of the pixel rises. Using Artificial Neural Network to simulate the general function relation between the decimal data and temperature of pixel, its error is less than 0. 0001. Constructing the non-linear function between temperature and gray value according to this function, based on the coding ride of BMP this article transformed the infrared image into BMP which is the most common image format under Windows operating system, thereby the image can be recurred without the costly software which is especially made by the thermograh manufacturer, and some more information about the image can be acquired. All the research results in this article will be useful for the further infrared image processing.