Colla corii asini is a kind of traditional and precious Chinese medicine. The main raw materials are donkey skin, cow skin, etc. Different raw materials determine the medicinal properties and efficacy of ass hide glue. In order to realize the nondestructive testing of colla corii asini raw materials, this paper uses the space offset visible near-infrared spectroscopy technology combined with diffusion mapping and K-nearest neighbor algorithm to carry out the identification and classification of colla corii asini raw materials. This research realized a space offset visible near-infrared spectroscopy technology based on the spatial offset theory, carried out the spatial offset spectral experiments of donkey hide, cow hide and their mixtures of different concentrations, realized the dimensionality reduction of high-dimensional spectral data through the diffusion mapping algorithm, and then completed the identification of donkey hide and cow hide colla corii asini based on the K nearest neighbor algorithm. The recognition effect of the model was evaluated through the comprehensive evaluation index of the classification algorithm confusion matrix and accuracy. The results show that the diffusion mapping combined with K nearest neighbor algorithm proposed in this study has high feasibility and reliability in assisting space offset visible near-infrared spectroscopy to identify colla corii asini raw materials. This paper proposes and verifies a new method for rapid identification of colla corii asini raw materials, which provides a new methodological reference for relevant technicians.