In the field of target recognition, target detection and tracking can be achieved by measuring the temperature of it. At present, most of the temperature measurement technologies are used for surface targets. Dim small targets are often faced with several problems during temperature measurement, such as the low filling rate of field, unknown emissivity and serious noise interference. Referring to the current issues about dim small targets temperature measurement, this paper built a new model for dual-waveband thermometry of them based on the wavelet analysis theory and neural network theory, obtaining the dual-waveband thermometry results of the dim small targets which are very close to the theoretical temperature. What’s more, the model validation is carried out by using the measured data of dim small targets. Analysis results show that the new model is more suitable to measure the dim small targets temperature of the radiation intensity signal-to-noise ratio within the scope of 3-8, laying the theoretical foundation and technical foundation for the recognition of dim small targets.