Chinese Text Detection And Spatial Temporal Distribution Analysis Of Capital Cities In Southeast Asia Based On Street View Images

The main content of the Belt and Road is the realization the five major goals "policy coordination,facilities connectivity,unimpeded trade,financial integration and people-to-people bonds" of countries along the Belt and Road.Systematically and quantitative evaluation of the basic situation of the Belt and Road construction can provide the important data support for scientific decision-making and regional cooperation.Language is the foundation of the connectivity projects of the Initiative construction,while character is the main constituent part of language.The usage of Chinese characters in countries along the Belt and Road can reflect the communication with China,which can provide method of grasping the situation of the connectivity projects of the Initiative construction,especially the knowing the situation of tourism,cultural,educational,scientific and technological exchanges.A quantitative study on the spatial distribution of Chinese characters of capital cities in Southeast Asia can provide application demonstration of the Chinese characters spatial analysis of coutries along the Belt and Road.Traditional data acquisition methods are difficult to obtain large scale,spatial distribution of Chinese character spatial information.Street view map shows the details of street facade,including the text used.What's more,street view map has the advantage of geographical location,wide coverage and free access to users,which make it possiblely provide data support for acquiring the spatial distribution information of Chinese characters.Technique of text detection of natural images is very mature,but the research of Chinese characters detection from multi language remains improved.Subject to the constraints of data acquisition,the research of spatial analysis of Chinese characters is still blank.This paper focus on sloving the technology difficulty of obtaining the Chinese characters spatial distribution information and the research gap of related spatial analysis,proposed a feasible technique of Chinese characters spatial data acquisition based on street view images,and established an effective system of analying and evaluting the distribution of Chinese characters.The main research contents are as follows:(1)Chinese characters detection of street view images.The paper proposed a technique flow "text detection-Chinese character distinguish" based on online street view map.The street view images of capital cities in Southeast Asia were downloaded based on the network data acquisition technology.Three methods:stroke width transform,improved maximally stable extremal regions,connectionist text proposal network,are performed to realize text detection.The algorithm result was chosen as the data source of Chinese character discrimination by calculating the accuracy and recall index.According to the analysis of the characteristics of different language characters,a new method for distinguishing Chinese characters from other language characters was developed based on character segmentation and character feature calculation.Then Chinese spatial information data of capital cities in Southeast Asia was obtained.(2)Analysis of Chinese character spatial distribution.Based on Chinese spatial distribution point data,the research used mathematical statistics and spatial analysis technology and method to explore the number and density of Chinese characters,the pattern of Chinese character distribution of capital cities in Southeast Asia by calculating the degree of agglomeration,the degree of equilibrium,the discrete trend.The spatial correlation analysis of the Chinese characters and the road network was conducted to find out the region location of the Chinese characters in the city.The central area theory was used to calculate and evaluate the radiation area and capacity of Chinese characters in different cities.Thus,a comprehensive understanding and comparision of Chinese characters spatial distribution of capital cities in Southeast was obtained.(3)Spatial temporal distribution analysis of Chinese characters in Singapore.The research was performed based on the Chinese characters distribution data between 2008 and 2015.Using mathematical statistics and spatial technique to know the change of the number and density,the principal distribution direction,the degree of agglomeration,the degree of equilibrium,the region location,the radiation capacity of Chinese characters in Singapore during the periods.The changes of spatial distribution of Chinese characters and regional differences are revealed.The main research results are as follows:(1)Among the seven capital cities in Southeast Asia,Chinese characters' number and density was highest in Kuala Lumpur,Jakarta is the least.Not matter which city,Chinese characters represented cluster distribution,and the degree of agglomeration was highest in Kuala Lumpur,Bangkok is the least.For the distribution balance within study cities,the balance was best in Phnom Penh,the worst in Manila.For the relation of road network,Chinese characters mainly distribute in residential roads and the density was positive correlated with road network centrality.The Chinese characters radiation space was biggest in Phnom Penh,and the minimum in Jakarta,as long as the influence.(2)Between 2008 and 2015,the number Chinese characters in Singapore were increasing year after year,the area density of all districts were increased.The highest kernerl density point moved toward Central Area,and the overall distribution center moved to the southeast.The degree of agglomeration of Singapore's Chinese characters didn't change,while the balance of distribution decreased.Besides,the increased Chinese characters mainly distribution in residential roads and the radiation space area has increased.It means the chance for residents'access to Chinese language and the influence of Chinese language to Singapore enhanced.The paper analysed and discussed the Chinese characters distribution of capital citiesof Southeast Asia from different aspects,and achieved good results,but there are also some shortcomings.The Japanese characters or some alphabetic writing text didn't completely eliminate based on the Chinese characters distinguishm method,and didn't performd text content recognition.How to improve the effectiveness of discriminant Chinese characters and the text recognition needs to be further studied.In addition,this paper only consider the spatial location and road network to analyze the Chinese characters distribution.In the further study,some factors such as the Chinatown,the commercial center,foreign investment,the impact of the policy,the local Chinese people change,could be added to find out the Chinese characters distribution mechanism.