Research On Objects In The Same Characteristics Matching And Positioning Based On Binocular Stereo Vision

With the continuous development and intense research of technology of therelevant field, as well as the strengthening of the modernization degree of weaponsand equipment, the research of military training assessment system should bedeepened. How to solve the problem of remote monitoring and automatic assessmentin salvo bomb mode for the shooting range automatic target evaluation system is oneof the issues that need to be addressed. Based on this background, this paper focuseson multi-targets in the same characteristics matching and positioning.This paper outlines the background and significance of the research, and sums upthe targets detection and location technology development at home and abroadcombined with domestic and foreign literature. There are many kinds of imagematching technology algorithm. Many widely used image processing algorithms areclassified by reading a lot of literature. The classical algorithms are listed in eachcategory and compared and the different application scopes of different algorithms areresearched.The traditional image matching algorithms do not apply because the subject hasthe characteristics of multi-targets and the same characteristics. A matching methodbased on time division and epipolar constraint is raised based on the actual problem.Epipolar geometry and epipolar constraint as the theoretical basis of the matchingmethod, are first introduced and at the same time to understand the system used in themeasurement model.The system algorithm is developed based on the shooting range remotemonitoring and automatic assessment system. The system algorithm is based on thedesign of the system structure itself, so the system design principle and composition isintroduction briefly first. Contrast to the single-shot bomb mode, a more detailedanalysis of the salvo bomb mode point of impact floor-to-ceiling as well as thecharacteristics of the image by acquisition system is applied which the algorithmresearch need to consider and resolve, but also is the technical difficulties.The system is divides into two parts of the front-end measurement station whichcompletes target detection and identification and higher-level monitoring and controlcomputer which completes target matching and positioning. The system algorithm is divided into two parts of the front-end measurement station processing algorithms andhigher-level monitoring and control computer processing algorithms based on thedesign of the system, including five algorithm module for real-time inter-framedifference threshold determines trigger, the connected domain statistical methodcounts target features, image sequence space-time correlation detection--successiveframes to determine, time series segmentation matching, matching based on epipolarconstraint to finish target detection, identification and matching, positioning function.The algorithm overall process and part of procedures detailed process are given.Range live-fire exercises acquisition sequence pictures, the external field simulationsimage analysis, data analysis from algorithm processing results prove that theproposed algorithm can solve partial matching and positioning of the samecharacteristics targets.