Abstract: This study proposes a target clustering recognition method based on the spectral fusion of short-wave and mid-wave dual-band spectral information to address classification and recognition challenges in infrared detection targets. The article first analyzes atmospheric transmission characteristics under short-wave and mid-wave conditions, as well as differences in target response signals between these spectral bands. Then, based on the KANN-DBSCAN clustering recognition algorithm, the study completes the identification of three different types of targets. This method adds spectral dimension information to conventional recognition elements, is robust, and fully exploits the fault tolerance of the clustering recognition model. The simulation experiment provided response signal data for three types of targets under short- and mid-wave conditions and performed target recognition. The simulation results show that the target clustering model based on spectral fusion of short-wave and mid-wave data has strong recognition capability, with accuracy exceeding 90%.