Research on Optimization Method of Landscape Architecture Planning and Design Based on Two-Dimensional Fractal Graph Generation Algorithm

The development of modern mathematical theory, especially two-dimensional fractal graph algorithm, provides a possibility for large-scale landscape data processing. Landscape digital identification technology is an innovative technology based on digital landscape technology and computer identification of experimental data. It is an important artificial intelligence technology, which includes three steps: landscape acquisition, landscape processing and landscape identification. The characteristics of the scene in the landscape picture can be collected by special instruments, such as cameras, etc., and then the collected data can be processed by two-dimensional fractal graph algorithm, and finally realice the automatic identification of the landscape. For images with significant boundary characteristics, we can extract the boundary of the región quickly and accurately, so as to realice the segmentation of the region. However, when the edge features of the image are not good enough, there is little color difference between the background and the region, or there is some interference, the result will be very bad. In this paper, based on the two-dimensional fractal graph generation algorithm, a series of optimization of landscape architecture planning and design. The accuracy of landscape prime number can reflect whether specific types of landscape pictures can be correctly identified and divided. 200 Pictures are divided into six categories, namely Water scene, landscape scene, living scene, sky scene, architecture and transportation then exact ratios of two-dimensional fractal graph network -8s, two-dimensional fractal graph network - 16s, two-dimensional fractal graph network -32s and two-dimensional fractal graph network -32s. It reached the best level in pixel accuracy, average accuracy, average IU, etc., and the pixel accuracy has reached as high as 100%, average accuracy has reached 100%, average accuracy has reached 100%. When compared to the recommended algorithm, the 2D fractal graph generation algorithm has the highest accuracy (94.52%), precision (93.34%), and recall (94.18%) in the classification process.