Composition is an important expression of the artistic power of fine art landscape oil painting, however, its teaching has long faced the dilemma of subjective judgment and vague guidance. This paper takes the evaluation of fine art landscape oil painting composition as an entry point, describes the feature selection method, evaluation algorithm process, and the realization of each feature evaluation in image evaluation. At the same time, combined with the idea of multiple linear regression, it puts forward the evaluation method of landscape oil painting image composition based on multiple linear regression. After the objective composition performance results are obtained through the calculation of the multiple linear regression method, the content-based image scaling algorithm is used. Under the premise of ensuring that the image content is not changed, image scaling is performed to construct a composition optimization model for landscape oil painting images. The model, guided by the results of the regression algorithm, increases the optimization score of the actual oil painting image composition from the original 0.26 to 0.849, showing high feasibility and application value.
