Height Analysis And Identification Research Based On Barefoot Footprint

Influenced by the bones of the feet and the living environment,the footprint is unique and unique.Footprints contain information such as walking habits and foot structure,and are easy to collect and conceal.Therefore,the application research based on footprints has attracted more and more researchers' attention,and the effection footprint features based on footprint data becomes the key to successful application.Based on the study of footprint feature extraction and its application at home and abroad,this thesis analyzes the barefoot footprint feature,study the relationship between the footprint shape feature and height,and identification based on the footprint feature,the morphological feature and the convolution feature.The specific research content is as follows:(1)This thesis analyzes the footprint shape feature,the morphological feature and the convolution feature of the footprint data.According to different types of footprint data and different application areas,the footprint length and footprint width as the footprint shape features?the footprint area and the pressure histogram as the morphological feature and the convolution feature based on LeNet-5 model are exteacted for the footprint pressure data;the the footprint length?footprint width and distance between footprint feature points as the foorprint shape feature are exteacted for the footprint optical image data.(2)By the linear regression method,this thesis analyzes the differences in the results measured by three methods for measuring the footprint length and the footprint width and study the relationship between the footprint shape feature and height.Through the scatter plot,Pearson correlation coefficient and the linear regression fittiong experiments,it show that three measurements of the method of measuring the footprint and the footprint width has no significant difference,after eliminating the effects of multicollinearity between the footprint shape feature,only the footprint length and height are strongly correlated,and features that are strongly correlated with the footprint length are fAC?fAE?fAF?fBE?fCM?fEM?fFM.(3)This thesis proposes a classification and recognition method for footprint pressure data.Firstly,this method selects the feature suset with the best recognition rate by calculating the recognition accuracy of each feature and its combination features in the SVM.Then,implemented fusion of the footprint featurea by calculating the weight coefficient of each selected feature in the classification and identification.