In detection of distant target, it is difficult to extract target information effectively, which are the temperature, area, surface material type and other characteristics. According to the spectral emissivity, which is the inherent nature of an object, target material characteristics, structural characteristics and the infrared radiation characteristics, an IR composite model is established based on two wavebands. Based on the model and material spectral emissivity, a method of feature extraction of target is proposed. In terms of the special target, the main categories of constituent materials used in constructing them are found. Meantime, the spectral emissivity of materials is measured. The correspondent database is established, which describes the relationship between the categories of constituent materials and spectral emissivity of materials. Based on the model and database, surface material type, radiant temperature and effective radiation area of target are obtained by using chicken swarm algorithm. The method has a good result.