Font Size: a A A

Design Of Automatic Target Reporting System And Research On Key Technologies Based On Machine Vision

Posted on:2021-04-11Degree:MasterType:Thesis
Country:ChinaCandidate:Q B ZhangFull Text:PDF
GTID:2428330611952085Subject:Engineering, Electronics and Communication Engineering
Abstract/Summary:PDF Full Text Request
The application of image processing technology has involved all aspects of human life.Compared with analog image processing technology,digital image processing technology has the characteristics of high processing accuracy,rich content,and flexible methods.It has become a development trend to apply digital image processing technology to the target reporting link in military shooting training instead of manually reporting targets.This dissertation studies the aspect of automatic target reporting in military shooting,mainly introduces the typical solution ideas and effects in traditional way,and analyzes the key problems that need to be solved at present and the difficulties in the current methods.In the scheme design of the automatic target reporting system,a lowcost idea of the automatic target reporting system for light weapon shooting is adopted,and a hardware and software solution for the automatic target reporting system based on machine vision is designed.In this project,we have completed the hardware design,software algorithm design,including image acquisition,performance output and other aspects of the research.The main work of this article is summarized as follows:(1)In terms of hardware,a set of scheme of using wireless image transmission and wireless power management in hardware is designed.Only using a single camera can complete the image acquisition of multiple targets,which can avoid the transformation of the existing training ground in practical application,and improve the field compatibility of the system.(Corresponding to the second chapter of the dissertation).(2)In the aspect of image preprocessing,we propose a target segmentation method based on the color characteristics of the target paper,which can automatically recognize and cut out independent target images from a set of images containing multiple targets.Aiming at the problem that the target image is tilted,we design a method based on template matching to obtain the standard points automatically,which realizes the automatic processing of geometric correction of the target image.In view of the fact that there is an training subject under the condition of low illumination at night in the actual target shooting training,in order to achieve automatic target reporting in this scenario,we proposed a new non-realistic night image enhancement method,which can enhance the brightness and details of the image.The experimental results show that the method can effectively enhance the target image with a certain degree of darkness under low light conditions.(Corresponding to the third chapter of the dissertation).(3)As bullet hole recognition,we proposed a new bullet hole recognition method based on image region clustering.This method overcomes the shortcomings of the traditional image-based clustering method,which solved the problem that the traditional method based on image region clustering has poor effect on the character processing in the central ring area,and can obtain the centroid coordinates of all bullet holes on the target image.(Corresponding to the fourth chapter of the dissertation).(4)In terms of ring detection,a ring labeling method based on the area characteristics of the ring area has proposed.Each ring area in the chest ring target image is labeled as a numerical matrix consistent with the true ring value it represents,and the ring area is completed.Detection.(Corresponding to the fourth chapter of the dissertation).(5)For score judgment and precision analysis,the coordinate data of the bullet hole centroid is used to search for the ring value in the 8-ring neighborhood in the ring value labeling matrix to obtain the middle target score.Finally,the factors that affect the accuracy of automatic target reporting system are analyzed.(Corresponding to the fifth chapter of the dissertation).The advantage of the automatic target reporting system designed in this paper is that when the target is reported,only the last image after shooting is collected,which can meet the needs of all target reporting processes.Besides,considering the special case where the bullet hole just hits the loop line,the handling method of this case is explained accordingly.
Keywords/Search Tags:Automatic target reporting, machine vision, multiple target positions, bullet hole identification, ring zone marking
PDF Full Text Request
Related items