Intersection recognition method based on GoogLeNet neural network

Abstract

The invention discloses an intersection recognition method based on a GoogLeNet neural network. The intersection recognition method comprises the following steps: S1, sampling: selecting road networkdata of main cities throughout the country as a sampling source of a complex intersection sample; S2, constructing a Delaunay triangulation network: carrying out the initial positioning of the complexintersection, and preliminarily determining the central position and spatial range of the complex intersection; S3, enhancing the sample size: enhancing the sample size through sample simplification,sample rotation and mirroring; according to the method, a research hotspot GoogLeNet neural network in the field of machine vision is introduced into complex intersection recognition, and rapid and high-quality positioning of the OSM data complex intersection is realized in a mode of combining vector data and raster images.

Description

Technical field [0001] The invention relates to the technical field related to the recognition of complex intersections, in particular to an intersection recognition method based on GoogLeNet neural network. Background technique [0002] A complex intersection refers to a gathering area connected by multiple intersecting arterial roads, ramps, sidewalks and other auxiliary roads. It is a typical microstructure in the road network, and it is also a basic and important part of the road network. component. In the synthesis of large-scale topographic maps, the identification of complex intersections is one of the key steps in road network selection, simplification, and typicalization. However, due to its intricate structure and diverse morphological changes, how to accurately identify complex intersections in the road network Intersections have always been a difficult point in research. In the past few decades, scholars at home and abroad have conducted a lot of research on the auto...

AI Technical Summary

Problems solved by technology

[0004] The purpose of the present invention is to provide an intersection identification method based on GoogLeNet neural network, to solve the existing complex intersection identification method mentioned in the above-mentioned background technology, the identification accuracy of the intersection depends on the design of the selected index, and cannot effectively distinguish There are complex road intersection structures with interfering road sections, and in some recognition methods based on neural networks, the positioning of the center of complex intersections is not accurate, and complex intersections often deviate from the center of the map, and the identified intersections will be It is biased towards one side of the sampling range, thereby reducing the accuracy of complex intersection recognition and the convenience of subsequent applications. Complex intersections are rich in local detail features. Based on neural networks with low layers such as AlexNet, it is difficult to accurately learn their deep fuzzy features. , the recognition model requires a convolutional neural network with more layers

Images