Core is a part of subsurface rock formations, and rock classification can be achieved by analyzing lithological characteristics such as color, texture, or shape. This is an essential step in oil and gas exploration. In the field of geology, core image analysis is a method for studying the micro-features of rocks, utilizing color and texture characteristics from core images for lithology identification. This paper establishes three models: one utilizing color moments for neural network training, another using gray-level co-occurrence matrix for neural network training, and a third employing a fusion of color moments and gray-level co-occurrence matrix for neural network training. The results from the three models are compared to assess the experimental accuracy. The findings indicate that the multi-feature fusion approach demonstrates higher precision in core image lithology identification.