A method, system and equipment for intelligent identification of land functions based on multi-source data

Abstract

The invention discloses a land function intelligent recognition method, system and equipment based on multi-source data. The method includes: acquiring initial data and constructing an area analysis unit; extracting scene points from street view images, and mapping the scene points to the area analysis unit ; Use the first convolutional neural network model to extract features from the street view image of the scene point to obtain the first feature vector; use the second convolutional neural network model to extract features from the remote sensing image of the scene point to obtain the second feature vector; Kernel density estimation is performed on the information points of each category to obtain the spatial distribution characteristics of the kernel density; the feature extraction is performed on the information point data in the scene point to obtain the third feature vector; the vector machine classifier is trained and used to identify the land function type of the regional analysis unit. The invention realizes the training of the classifier by processing and fusing various data, utilizes the classifier to automatically identify the land function, and effectively improves the accuracy of the land function information.

Description

technical field [0001] The invention relates to the field of land function identification, in particular to a multi-source data-based intelligent land function identification method, system and equipment. Background technique [0002] Land use function information can intuitively reflect the basic form of urban layout and regional differences. Based on this, land use function information has very important reference and use value in urban planning, municipal construction, and environmental monitoring. In recent years, the following three types of data are mainly used to judge or identify land use function information, namely remote sensing images, street view images, and POI data. Among them, remote sensing images can obtain the original data of the city through high-resolution images, but because the time, space, and spectrum of remote sensing satellite images are difficult to obtain high resolution at the same time, there is a certain bottleneck in the perception of ground...

AI Technical Summary

Problems solved by technology

Among them, remote sensing images can obtain the original data of the city through high-resolution images, but because the time, space, and spectrum of remote sensing satellite images are difficult to obtain high resolution at the same time, there is a certain bottleneck in the perception of ground objects, that is, It is said that it is difficult to accurately and real-time identify land use function information from remote sensing satellite images; street view images can accurately describe urban scenes, but there are problems such as difficult collection and incomplete coverage of the collection area; POI data, that is, information point data, is used in human activities. It has certain reference value, but its data rules are easily affected by holidays or cluster activities, and it is easy to generate noise, which affects the accuracy of information to a certain extent.

Images