【作者】 汪传忠；

【导师】 廖振强；

【作者基本信息】 南京理工大学 ， 机械工程， 2007， 博士

【摘要】 在国家某重点课题的支撑下，本文对自动武器射击特征的图像检测和分析方法进行了较为系统深入地研究，对自动武器射击时的膛口火焰、射频、枪械后坐运动、枪管后坐运动、枪管组的旋转运动和机枪架座的后坐运动、射弹弹孔的坐标检测、射弹密集度分析以及射弹散布的规律等都提出了相应的图像检测处理和分析方法，建立了自动武器发射特征图像检测与分析的基本框架，并结合实际课题进行了实验验证。实现了自动武器多参数无接触同步图像检测和分析。此外，本文对自动武器射击性能检测试验建立了随机微分方程描述模型，并应用于小样本定型试验中，论文完成了以下几个方面工作：(1)提出基于视觉特性的枪口火焰图像的分割和强度度量方法，图像分割方法是利用简化Mumfond-Shah活动轮廓线图像分割方法对火焰图像的颜色距离图像进行分割，该方法是水平集方法在枪口火焰图像分割中的应用，可以实现枪口火焰图像的自动分割，具有一致性；基于视觉特性的强度度量方法同时考虑了火焰的强度和火焰的面积。使用本文提出的枪口火焰图像分割和度量方法可以初步实现基于视觉特性的枪口火焰定量检测和评估。(2)利用自动武器膛口火焰周期与射弹周期一致的特点，提出了基于火焰特征的自动武器序列图像射频检测方法，包括小波极值时间尺度检测方法、基于时频分析的序列图像射频估计方法和基于序列图像的转管机枪首发射弹枪管转速检测方法，并将之用于双联转管机枪射频估计，进而分析因转管特征引起的射弹散布。(3)提出了基于序列图像的自动武器射击多参数同步检测方法，可以同时检测自动武器的运动、射频及火焰等射击特征的定量信息。在运动检测方面，给出了自动武器序列图像位移检测方法和序列图像时变边界运动参数检测方法，利用该方法完成了某转管机枪枪口后坐和全枪后坐的二维运动检测；后坐位移检测结果验证了转管机枪的动力学建模，为动力学建模验证提供了一个新的途径。(4)在自动武器固定架密集度试验中，特别是小口径弹药(如5.8枪弹)的弹孔不完全重合时，弹孔坐标的可分离检测是自动武器立靶密集度自动检测首要解决的关键问题之一，本文在解决了该问题之后，提出了基于图像的自动武器弹道坐标自动测量方法，包括基于序列图像的自动武器弹道坐标测量方法和一组累积射弹图像的弹道坐标测量方法，前者不仅可以实现高射速自动武器连发射弹坐标的检测，而且可以分辨出连发射弹的弹孔顺序号，为高射速自动武器连发射弹的散布规律分析和密集度试验分析提供了一种有效的试验检测方法。(5)提出了自动武器性能检测试验中的随机微分方程描述方法，包括基于概率积分测度的随机微分方程描述方法和基于概率测度的随机微分方程描述方法，在此基础上给出了融合小样本定型试验模型，并对自动武器小样本可靠性试验和步兵榴弹发火性能小样本试验进行了初步分析利用。更多还原

【Abstract】 The paper was supported by one pre-research of an important national researchproject. The systematic study on the image-based measurement and analysis of thedynamic characteristics of the automatic weapons has been conducted. At the same time,the corresponding solutions, such as image-based measurement and analysis on the muzzleflame, the fire rate, the movement of the recession of tubes, the circumgyration of the tubegroup of Gatling weapon, the ballistic trajectory coordinates, the density and distribution ofthe shot of Gatling weapon are provided in the paper. The basic theory of imagemeasurement and analysis of the dynamic characteristics has been established and tested inthe research project. The multi-parameter and synchronized measurement based on theimage sequences has been proposed and testified. Furthermore, the Stochastic differentialequation (SED) model has been established and applied to the small sample test in thegun’s approval. The achievement of the paper:(1) The quantitative measurement and image segmentation of the muzzle flameimage based on the humane vision was proposed. The Mumfond-Shah active contoursegment is used in this method to segment the color distance image of the flame image,which is the application of the level set methods on the CIEL~*a~*b~* color space. It istestified to be effective and consistent in the automatic segmentation of the muzzle flameimage. The quantitative measurement includes the consideration of the intensity and thearea of the flame in the image, which can describe the difference on humane visionbetween two images. It can be used to in the quantitative measurement and evaluation ofthe property of the muzzle flame.(2) With the consideration of the same periodicity of the flame and the gun’s firingrate, the measurement of the fire rate, based on the image sequences of the flame has beenprovided which included the time-frequent analysis and the time-scale measurement of thewavelet analyzed signal. The measurement of rotary speed of Gatling gun’s barrels at themoment of first shot based on the image sequences has been included in this measurement.This measurement was applied to the estimation of the fire rate of the connected gatlingweapon, and then used to analyze the distribution caused by the characteristic of rotarybarrels.(3) A multi-parameter and synchronized measurement of gun’s fire characteristics onthe series of image sequences has been provided, which can be used in measurement of themovement, the fire rate, the muzzle flame, and so on. At the same time, the parameter measurement on varying boundaries based on the level set methods has been studied. Themeasurement of the movement has been applied to test the bi-dimensional movement ofthe recession of muzzles and recession of guns. The result of the parameter measurementtestified the dynamic model of Gatling weapon, which is a new method to testify thedynamic model of Gatling weapon.(4) During the dispersion test of automatic weapon, especially when non-fullsuperposition of ballistic trajectory holes happens on the small caliber ammunition, theseparable measurement of ballistic trajectory coordinates is one of the key problems in theautomatic dispersion measurement. Following the solution of this problem, theimage-based automatic measurement of the ballistic trajectory coordinate of the automaticweapons was proposed, including sequential-image-based measurement of the ballistictrajectory coordinate of the automatic weapons, and the group-hole-image-basedmeasurement of the ballistic trajectory coordinate of accumulated shots. The former can beused to test the high fire rate ballistic trajectory coordinate, and recognize the order of arunning fire shots, which provides an effective measurement on the dispersion and densityof a running fire with high fire rate.(5) An SDE model has been established for the test of the properties of automaticweapons, including the SDE for integral probability measure and probability measure. Themodel for small sample of automatic weapon’s approval test has been provided which isinitial used in the small sample tests of the reliability of automatic weapons and grenade’sproperties.更多还原