To recognize the point target maneuvering mode from the spectral characteristic dimension, the mapping relationship between the point target maneuvering mode and the spectral signals is built, and the multi-spectral radiation characteristics of point target with a maneuvering status in the direction of observation are investigated. The features of multi-spectral radiation signals are extracted to establish a dual-color-ratio-feature spatial model. The clustering method based on the Gaussian mixture model is used to analyze deeply the features of migration and separability of the dual-color-ratio-feature space. The migration vectors of feature sub-space of different maneuvering modes and the change of the cosine of adjacent vector angle are obtained. In addition, the smallest attitude angle change of feature sub-space and the separable distance threshold are obtained as Δα=6.25° and Dth=2.6, respectively. It provides the basis and feasibility for the recognition of point target maneuvering modes. According to the characteristics of dual-color-ratio-feature sub-space, a method for the recognition of point target maneuvering modes based on sequential-feature sub-space is proposed, which is verified by simulation as simple and feasible as well as possesses high sensitivity and separability in the recognition of point target maneuvering modes. These results have great significance for the acquisition of point target maneuvering information in the beyond-visual-range air combat.