Research On Modeling And Software Developing Of Military Posture Still Standing Assessment System

A National Day parade is a manifestation of a country's comprehensive strength,and its military posture shows an open and confident image of the PLA.At present,traditional manual training is still used in the assessment and training of military posture.This method has some problems,such as organization complex,real-time evaluation ability poor,efficiency low,etc.The assessment results are influenced by subjective factors.It is difficult for instructors to observe the details of an assessors' military posture in real-time and comprehensively during assessing or training,which would weak effectiveness of training and assessment.Therefore,it is of some practical significance and value to develop a set of scientific military posture assessment system to realize an electronization,informationization,and digitization of single-person and single-row military posture assessment.To fill a gap of no intelligent equipment in military posture assessing and training,computer vision technology was applied to military posture assessment for the first time to adapt to the development of modernization and information technology of military training in the new era by us.Image processing,electronic sensing,in-depth learning,and data analysis were used to develop a military posture assessment system(mainly assessment,supplemented by training)to provide real-time,accurate,and intelligent application services support for assessors and instructors.The main contents of this article are as follows:(1)Data collection: OpenPose model was used to detect key points of human bones,and three detection methods,an improved edge detection method and color feature value extraction method were used to identify he interested parts.A collected image data were calibrated according to a test conditions and hardware used.A gravity center detector was designed,to establish a gravity center position model,and data were collected and processed on the gravity center of a person.After testing,the measurement error and accuracy of data acquisition meet the requirements.(2)Military posture recognition modeling: Based on the theoretical analysis and requirement analysis,two kind methods of military posture recognition were proposed: color marking method and bone key point coordinate method.Based on these two kind methods,a combined model of military posture recognition was established,and all collected data is processed and fitted.After testing,the recognition accuracy of key locations is higher than 97.5%.(3)Evaluation system modeling: According to the scoring rules,scoring weight and the linear function relationship between the characteristic quantity of the assessment index and the score of the assessment index,an evaluation model was established,which realizes the real-time evaluation of each assessment index and outputs reasonable and objective assessment results.(4)Data management and software development: Data management is based on MYSQL database.System development and interface design are based on QT software.After the system function,performance and robustness had been tested one by one in the laboratory and on-site(Wuqing,Shijiazhuang,Dingxing,etc.),the system had been tried by the Joint Command.The system has achieved the required functions and has good anti-interference ability,robustness,and practicability.Figure 86,table 13,reference 61.