<p>Oil paintings have attracted numerous art enthusiasts with their unique brushstrokes and rich color layers, and achieving effective recognition of oil painting brushstrokes is of great significance for the analysis and protection of art works. This article aims to propose a method for oil painting stroke recognition based on multi-scale feature extraction and support vector machine (SVM) algorithm. A study proposes an image recognition technique that combines multi-scale feature channel attention mechanism. Under this framework, firstly, important local features in oil painting images are obtained through multi-scale feature extraction techniques, and then support vector machine models are applied to classify the extracted features. The use of selective feature fusion strategy further improves the robustness and recognition accuracy of the model. In the performance analysis of model parameters, we evaluated the impact of different parameter configurations on recognition results to optimize algorithm performance. A novel oil painting multispectral data acquisition method was proposed by studying and designing a process for identifying stroke flow lines in oil painting, combined with a local lighting model for oil painting. The experimental results show that the algorithm performs well in oil painting image recognition tasks, effectively identifying stroke details in oil paintings and providing a new approach for in-depth analysis of oil painting works.</p>