Extraction,Comparison And Verification Of Image Elements Of World Cultural Landscape Heritage Based On Photos Data Mining

As an important achievement of the continuation and progress of human civilization,World Heritage is a non-renewable and irreplaceable precious resource.Protecting,inheriting and making good use of these resources is an inevitable requirement for the continuity of human society and the sustainable development of the world.As a product of the interaction between nature and humans,the emergence of cultural landscape types breaks the barriers of cultural and natural values in the original World Heritage classification,and is a new breakthrough in people’s cognition of traditional heritage.Clarifying the status quo of social cognition of the World Cultural Landscape Heritage has guiding value for its utilization and protection,especially for promoting its sustainable development.Based on the above background,this study takes urban imagery,tourism imagery and tourism gaze as the theoretical basis,and mines the image data of 70 world cultural landscape heritages on the Flickr photo sharing platform as research samples.With the help of Google Cloud Vision deep learning algorithm to analyze the content of the image representation,after the image data is textualized,the LDA topic model algorithm,topic feature word construction and correspondence analysis methods are used to extract the overall image elements of the world cultural landscape heritage,and the heritage based on image elements.This paper analyzes and studies the image elements of the world cultural landscape heritage from three aspects: land classification,comparison between this classification and two traditional classification methods,and verifies the scientificity of the relevant research in this paper through the empirical analysis of the Hani terraces case.The main research contents and conclusions are as follows:(1)From the analysis results of the picture data,there are five image elements in the world cultural landscape heritage from the perspective of tourists.By modeling the label content of 70,000 images with the LDA algorithm,it is calculated by the elbow method that the image elements of the world cultural landscape heritage are clustered into five dimensions(ie image elements)with the best effect.By extracting the high-frequency thematic feature words of the five image elements,and superimposing the content of the pictures to retrospectively analyze the connotation of each element,the overall image elements of the world cultural landscape heritage can be divided into five types,namely natural landscape,landmark buildings,recreational life,public space and Local symbols,the proportions are 43.75%,23.34%,10.24%,9.59% and 13.05%.(2)This paper proposes a classification method of world cultural landscape heritage based on the composition of image elements.First,a chi-square test was performed to determine the significant correlation between the world cultural landscape heritage sites and the composition ratio of image elements,that is,the classification method of image elements of the world cultural landscape heritage is scientific.After processing the proportion data of the image elements of each heritage site through SPSS correspondence analysis and dimensionality reduction,the classification results of the world cultural landscape heritage were obtained by combining 6 kinds of correspondence analysis graphic methods.In order to test the accuracy of the classification results,a research model of heritage site imagery was further constructed,which confirmed the accuracy of the classification results from three aspects: the dominant direction of heritage site imagery,the degree of heritage site imagery characteristics,and the similarity of heritage site imagery.Comparing the classification method of world cultural landscape heritage based on image elements with the two traditional classification methods,it can be found that the classification results of image elements are generally consistent with the traditional classification.Iconic buildings and local symbolic heritage sites;organic evolution heritage sites with rich natural landscape image elements are mostly natural landscape heritage sites;agricultural,settlement,and leisure heritage sites are mostly due to the rich image elements of human activities.Complex imagery heritage site.However,compared with traditional classification,the image element classification method has two major advantages:(1)the classification of heritage sites is highly differentiated,which solves the problem of overlapping classification results in traditional classification;The continuous development and adjustment of its classification results are more conducive to analyzing the status quo of heritage sites.(3)The verification research of this paper is carried out by taking the Honghe Hani Terraces as a case.The first part is to verify the rationality of using Flickr image data to extract image elements.By importing the latitude and longitude coordinates in the Flickr image metadata of the Hani Terraces into Arc GIS for visualization,the spatial distribution of the image elements of the Hani Terraces is obtained.The proposed path is consistent with the concept of perception of the city,and the image elements are densely distributed in the key scenic spots of the Hani Terraces Scenic Spot,which proves that the Flickr image data is reasonable and accurate to extract image elements;the second part is the image Element cognitive dimension verification.Through ROST CM 6,the texts in the Hani Terraces inscription documents are analyzed for high-frequency words.The image elements of the Hani Terraces under the official cognition also belong to the five overall image elements proposed.The third part is the verification of the proportion of image elements.Using Nvivo 11 to encode 586 pictures in the Hani Terraces inscription documents,it is found that the proportion of the image elements of Hani Terraces between official cognition and public cognition is mainly different in natural landscapes and landmark buildings,and the public cognition is more important.For the perception of the natural landscape of terraced fields,the official cognition deepens the perception of the local landmark building of the mushroom house,and the composition ratio of the other three image elements is roughly the same.