A deep learning-based algorithm for unitizing ecologically sensitive areas in rural landscapes

The ecological sensitivity of rural landscapes exhibits complexity and diversity. Traditional evaluation methods, which merely take into account a single factor or a limited number of factors, struggle to effectively manage uncertain information. This leads to inaccurate classification of ecological units in rural landscape ecological images, thereby undermining the precise assessment of the distribution of ecological sensitivity in rural landscapes. Therefore, a deep learning based algorithm for dividing rural landscape ecological sensitive areas is proposed. By selecting six major factors that affect the ecological sensitivity of rural landscape, such as geological environment, ecological and hydrological conditions, an ecological sensitivity evaluation index system is constructed, which is used as an input vector, and fuzzy neural network is used to output the ecological sensitivity of rural landscape; In addition, support vector machine is used to divide the ecological units of the collected rural landscape ecological images, and the division results and the sensitivity evaluation results of each unit of rural landscape are used as the input data of ArcGIS software to realize the visual presentation of the unit division results of rural landscape ecological sensitive areas. The results showed that with the increase of slope, the ecological sensitivity of rural landscape showed a trend of first increasing and then decreasing, the vegetation coverage rate decreased, and the ecological sensitivity of rural landscape showed a trend of gradually increasing; This algorithm can effectively evaluate the sensitivity of each unit of rural landscape, and visually present the unit division results of ecological sensitive areas of rural landscape. This algorithm can compare and analyze the changes of ecological sensitivity under different time dimensions.