With the improvement of urban greening level and residents’ increasing demand for green space, the traditional management system of garden enterprises has been unable to meet the management needs of garden enterprises, and with the promotion of informatization, many industries have felt its strong superiority. This article analyzes, processes and extracts the collected images. The noise of the collected park landscape is processed and preprocessed, and an effective original image is obtained. Then, the characteristic parameters such as gray histogram, oblique square and regional density distribution are used and fused with the grid spacing relationship curve to get the required landscape image. Using smoothing operation, the pixel of the image to be measured is different from the standard value, and the measured value can be obtained. Before adopting the landscape image optimization system based on machine vision, the landscape quality index of green space in C community is 0.3, and after adopting it, the landscape quality index of green space in C community is 0.8. Research shows that this system can not only adapt to the business needs of garden enterprises, but also make appropriate changes to other garden enterprises to make it suitable for other garden enterprises.