为了提高红外热像仪的测温性能，以黑体辐射图像为基准，研究了噪声对红外热像仪测温的影响。首先，分析了室内背景下红外图像中噪声的来源和组成，主要包括高斯噪声和椒盐噪声。接着，建立了红外测温实验系统，获取标定时的黑体辐射图像和实际测温图像。然后，对黑体图像采用人工加噪的方法，研究不同噪声参数和传统降噪方法处理后对测温性能的影响，得到了不同噪声参数及降噪方法对测温性能的影响规律，在传统降噪方法的基础上提出了改进均值法并进行实验验证。最后，将改进均值法应用到室内实际测温的图像上。实验结果表明，改进均值法处理后，温度均方差减小为 0.32，测温误差减少了 1.69℃。效果优于传统降噪方法，改善了红外测温性能。

In order to improve the temperature measurement performance of an infrared thermal imager, the effect of noise on the temperature measurement was studied based on a blackbody radiation image. First, the source and composition of noise in the infrared image with an indoor background are analyzed, including Gaussian noise as well as salt and pepper noise. Then, an infrared temperature measurement experiment system was constructed to obtain the blackbody radiation image at the time of calibration along with the actual temperature measurement image. Subsequently, noise was manually added to the blackbody image to study the effect of different noise parameters and traditional noise reduction methods on the temperature measurement performance and to obtain the relationships between them. Based on the traditional denoising method, an improved mean method was proposed and verified experimentally. Finally, the improved mean method was applied to actual temperature measurement in the room. The experimental results show that after applying the improved mean method, the mean squared error of the temperature reduced to 0.32, and the temperature measurement error reduced by 1.69℃. The mean method is thus better than the traditional noise reduction methods and improves the infrared temperature measurement performance.