This paper takes garden landscape as the research object, discusses the application of digital media art in assisting garden landscape design, and obtains the optimization probability of various types of garden landscape types in landscape spatial layout through patch generation using the change simulation (PLUS) model. And after using the multi-scale Retinex algorithm to make image enhancement, a fusion of single-scale enhancement results in achieving the initial optimization of the image, after a dynamic interception and stretching operation to restore the enhancement effect to achieve the optimization of the landscape image. Finally, a group of landscape design images are selected as experimental objects to test the effectiveness of digital media art-assisted landscape design. The results show that with the assistance of digital media art, the modulus of the change distance of neighboring units is between 2.21 and 10.89, and the relative change rate takes the value between 0.45 and 5.21. The method is capable of balancing the ups and downs and repetitive rhythms in the design, ensuring that the landscape has good brightness and visual effects.