Study On GIS Similar Legend Recognition System Based On Deep Learning

Deep learning technology has been developing rapidly in recent years.It has been widely applied in computer vision,image recognition and speech recognition,and is a hot research area at present.In recent years,the GIS technology develop rapidly due to the national support.A graphic symbol,including a graphic legend on a map or a mark to be used in conjunction with a variety of spatial objects,is a language of the GIS map.In the process of GIS research,we will need to combine two or more graphic legend layers,which means each layer' s graphic marks should be merged.This merge process is similar to mathematical union operation.The merged layer contains both unique legends and similar legends.It' s a time-consuming effort if we work done by hand.If the problem of similar legend recognition is solved through deep learning,the merging process of the legend layers can be greatly simplified.This paper introduces a GIS symbols recognition model based on deep learning.Training the model with GIS graphic legends,and repeating the training by using the comparing result outputted by a graphical user interface built by MFC.With these repeating training the model can recognize similar legends from two different layers efficiently and accurately and complete the combination of two graphic legend layers.