Drilling is one of the important means of geological exploration. In recent years, with the development of China’s geology, the storage and sharing of a large number of cores has become an urgent problem to be solved. The problem has been solved through research and development of a core spectral scanner to realize the digitalization of cores. However, the massive production of core spectral data and image data puts forward new requirements for data processing. According to the principles of spectroscopy and spectral analysis methods, spectrum analysis and altered mineral mapping of the spectral data of the core scanner can provide the basis for geological scientific research, deposit analysis, and prospecting. This paper proposes a classification database retrieval method, which not only improves the accuracy of retrieval, but also greatly accelerates the retrieval speed. The core image is also an indispensable part of the core information. Because of the limitation of the core scanner detector, the illumination conditions, and the influence of the cylindrical core, the collected core image will have uneven illumination and radiation distortion. Utilizing the nonlinear bilateral filtering method to sharpen the image, and applying the black and white plate calibration method to correct the core image make the core images closer to the real condition. The automatic image mosaic is completed by detecting feature points using the corner detection method. Core images are spliced into core columns and core trays one by one according to the drilling sequence, making the core images display more direct-viewing. Spectral analysis of minerals is the key to core scanning technology. Different minerals have different peak positions of characteristic absorption. Normally utilize the peak absorption matching method to search minerals, which is suitable for mixed mineral spectral retrieval. The matching of peak positions is based on a standard database. This paper proposes a classification database search method: based on the types of minerals, the standard database is divided into sub-databases of argillization alteration database, porphyritic alteration database, sericite alteration database and so on. According to the images of the samples and the geological environment in which they are located, select the appropriate subdatabase for search analysis. Two experiments were conducted in this paper. The same batch of samples were analyzed using the standard database and the classification database respectively. The results showed that the accuracy of the latter was higher; 141 samples were processed taking 233 seconds and 44 seconds with each method. Experiments prove that the classification database method is an effective method for retrieving large amounts of data accurately and quickly, which can both improve the accuracy of the retrieval, and greatly speed up the retrieval. This method is a novel, unique, and effective means in spectral retrieval. Solving efficiency problems for batch mineral data retrieval is the innovation of this paper. Mineral spectrums contain rich information, whose peak intensity, peak-to-peak- ratio, peak shift, FWHM, and reflectance are related to the relative content of minerals, temperature, cation exchange, crystallinity and color respectively. The metallogenic model can be obtained by comparing and analyzing the information from the same batch of minerals, which reveals the regularity of mineralization as well. This paper takes a drilling in Xuancheng, Anhui Province as an example to process automatically image stitching, spectral analysis and altered mineral mapping of the core spectral data. According to the analysis of information extracted from altered minerals, the area is an acidic, low-temperature geological environment, with darker rocks in the low-temperature areas and kaolinite and montmorillonite in the middle of the low-temperature areas, indicating a good oil storage environment. The experimental results show that this method can not only save a lot of manual workload, but also obtain high-quality core tray and columnar core splices and altered mineral information extraction graphs, which is a practical and reliable method for geological workers to process core data.