红外双色比能够表征目标的温度信息,但点目标双色比受噪声、探测器盲闪元以及跨像元因素干扰,难以准确测量,影响测温精度。应用递归图法定性分析和判断了点目标双色比的非平稳性。针对这一特性,先对点目标信号应用移动平均滤波做预处理,再通过小波软阈值去噪对双色比进行降噪处理。试验结果表明,在信噪比大于12的条件下,该方法可以实现静态点目标1 K温度分辨率,慢速动态点目标2 K温度分辨率。 Infrared two-color ratio can characterize the target’s temperature information, but the point target two-color ratio is affected by noise, detector blind flash element and inter-pixel interference, it is difficult to accurately measure and affect the temperature measurement accuracy. The recursive graph method is used to qualitatively analyze and determine the nonstationarity of the two-color ratio of the point target. In view of this characteristic, the moving average filter of the point target signal is used as the preprocessing, and then the noise reduction of the colorimetric ratio is performed by the wavelet soft threshold denoising. The experimental results show that the method can achieve 1 K temperature resolution of static point target and 2 K temperature resolution of slow dynamic target under the condition of SNR greater than 12.