Abstract: The mineral composition and its embedding characteristics in the ore are closely related to the crushing and grinding effect. Identifying the main mineral distribution characteristics of the ore can realize the reasonable distribution of mineral processing feed, but the conventional chemical detection and phase analysis need a certain time period, resulting in lag in raw ore data collection. This paper attempts to apply the spectral imaging data analysis method to the rapid identification of mineral composition and distribution characteristics of raw ore, so as to provide a basis for in-situ testing of ore properties. It takes molybdenum ore as the research object, prepares rock slice samples, determines the main mineral components and collects the hyperspectral data of the samples, and analyzes the spectral curve characteristics of quartz, potash feldspar, pyrite and biotite. Calculating the Hausdorff Distance and Euclidean Distance, through the spectral curve similarity calculation to identify different kinds of minerals and investigate their spatial distribution characteristics. The results show that quartz, potash feldspar, pyrite, biotite and other minerals in the visible-near infrared band（400-1 000 nm） can be quantified by spectral curve similarity algorithm to effectively distinguish the types of ores, and then the particle size and spatial distribution of the main minerals in the ore can be obtained by statistical analysis. The results of the two algorithms are consistent to a certain extent, but compared with Euclidean Distance, the Euclidean Distance has higher accuracy in identifying fine particles. The research results enrich the analysis method of raw ore properties and provide a basis for improving ore processing technology and realizing energy saving and efficiency.