The development of digital modeling technology opens up endless possibilities for three-dimensional modeling of urban landscapes. Based on computer vision methods such as stereovision and optical flow, this paper investigates how to carry out three-dimensional digital modeling of urban landscapes. We utilize the method of combining UAV and ground shooting to obtain image data of urban landscapes, then process the collected data with image smoothing, enhancement, edge detection, and segmentation. The Moravec operator and Hough variation are used to extract point and line features from the cityscape image, respectively. The beam leveling algorithm is used to reconstruct the cityscape in three dimensions, and it is optimized by combining the Kalman filter correction. This paper uses the landscape of Li City, Quanzhou City, as the experimental object for digital modeling. A comparative analysis of the three groups of experiments, A, B, and C, shows that the digital modeling technology of urban landscapes, based on image processing technology, achieves the highest degree of restoration of the original image. The errors for landscape elevation and non-elevation are 0.69% and 0.37%, respectively, which are the smallest errors among the three groups of experiments.