QQ客服 官方微博 反馈留言

题名	基于机器视觉的弹底缺陷与枪弹批号检测技术研究
作者	杜海静
学位类别	硕士
答辩日期	2014-05-28
授予单位	中国科学院沈阳自动化研究所
导师	肖阳辉
关键词	机器视觉 弹底区域分割 批号识别 缺陷检测 特征提取
其他题名	Research on Bullet Bottom Defects and Batch Inspection Technology Based on Machine Vision
学位专业	模式识别与智能系统
中文摘要	枪弹作为轻武器常规弹药，是现代战争、军事演习和军事训练最主要的消耗品之一。枪弹质量检测对武器装备质量的提高起着非常重要的作用。其主要包括尺寸测量、侧表面缺陷检测和弹底表面质量检测三部分。但由于技术条件的限制，目前国内仍沿用人工检测的方式来实现枪弹表面的检测，这种方式不但检测精度低，且安全性差。因此，研究开发高效的枪弹质量自动检测系统已成为军工生产单位的迫切需求。机器视觉检测技术作为一种非接触测量技术，近年来已广泛应用到军事、农业、视觉导航等领域。将其用于枪弹质量检测中不仅可以实现枪弹质量的自动化检测，而且能大大提高检测的精度和效率。因此，基于机器视觉的枪弹表面质量检测技术的研究具有重要意义。本文以其中的弹底表面缺陷和批号为检测对象，对基于机器视觉弹底质量检测系统中的系统设计、弹底区域分割、批号识别和缺陷检测进行了深入的研究。文中首先分析了弹底表面结构和几何特性，并搭建实验平台，得到原始图像。基于原始弹底图像的特点，本文提出了一种基于HSV模型的弹底区域分割算法，该方法以H分量为基准图像，采用投影法实现了弹底区域的准确分割。根据弹底批号属于压印字符这一特征，本文设计了一套适用于弹底批号的字符识别系统。其中针对字符边缘断裂问题，提出了一种基于链码最小距离的边缘连接方法，能够在双边缘条件下实现字符的边缘连接。在特征提取阶段，设计了一种基于多特征的字符特征描述方法，该方法能够有效准确地对字符进行表示。此外，为提高整个识别系统的识别精度和效率，本文构造了一个三级组合分类器。针对不同类型缺陷的特点，提出三种缺陷检测方法，实现了弹底缺陷的检测。根据检测结果，提取多种缺陷几何特征和灰度特征，并在此基础上设计了缺陷分类器，实现缺陷的识别和分类。为验证本文所提出的检测与识别算法，将其在MATLAB平台下进行仿真，并进行工程化实现，运用于课题组研制的枪弹质量检测系统中。检测结果表明，本文所提出的批号识别和缺陷检测算法能够准确地实现弹底质量的检测，且基本满足检测精度和效率的要求。
索取号	TP242.62/D77/2014
英文摘要	As light weapons of conventional ammunition, bullet is a one of the main consumables in modern war, military exercises and military training. Safe quality inspection of bullets plays a very important role in a national military industry healthy development process, mainly includes three parts: the size measurement, side surface defect inspection and the surface quality detection of bullets’ bottom. But due to technical limitations, the domestic is still using artificial test way to realize the surface inspection of the bullet, it has low detection efficiency, poor security and the intensity of labor is big. Therefore, the research and development of high-quality automatic detection system of the bullets has become an urgent demand for military production units. Machine vision inspection technology as a kind of non-contact technique has been widely applied to the military, agriculture, visual navigation and other fields in recent years. Applying it to the quality detection system of bullet can not only realize automation testing, and can greatly improve the detection accuracy and efficiency. Therefore, the research of bullets surface detection based on machine vision technology is of great significance.Using bottom surface defects and batch number as object to be inspected, the paper did an in-depth research on the system design, bottom region extraction, batch number recognition and defects detection in the quality detection system of bullets’ bottom based on machine vision.This paper analyzes the bottom surface structure and geometric features firstly, then setting up the experimental platform, and achieving the original image.Based on the characteristics of bullets’ bottom image, the paper proposes a segmentation algorithm of bullets’ bottom area based on HSV space. It takes the H component as a benchmark image, then using the projection method to realize the accurate segmentation of bottom area.According to this feature that the batch number belongs to suppress character, the paper designs a character recognition system, which is suitable for batch number. In view of the character edge fracture problems, we put forward an edge connection method based on chain code minimum distance, which can implement edge connection of the characters under the condition of double edge. In feature extraction stage, we puts forward a description method of characters based on multiple features, the method can represent characters effectively and accurately. Besides, we construct a triple combination classifier, which is used to improve the accuracy and efficiency of the whole recognition system.According to the characteristics of different types of defects, the paper puts forward three methods of defect detection, and defect detection is realized. According to the detection results, we extract a variety of geometric and gray-scale features of detects, and design a classifier on the basis, then classification is realized.To verify the detection and recognition algorithm proposed in this paper, we put it in the MATLAB simulation platform, and carried out engineering implementation for it, applied it to the quality detection system of the bullet. Experimental results show that, with batch number identification and defect detection as the main body of the bottom surface quality detection algorithm can accurately achieve quality detection of bullet’s bottom, and can basically meet the accuracy and efficiency requirements.
语种	中文
产权排序	1
页码	75页
分类号	TP242.62
内容类型	学位论文
源URL	[http://ir.sia.ac.cn/handle/173321/14792]
专题	沈阳自动化研究所_光电信息技术研究室
推荐引用方式 GB/T 7714	杜海静. 基于机器视觉的弹底缺陷与枪弹批号检测技术研究[D]. 中国科学院沈阳自动化研究所. 2014.

个性服务

查看访问统计

相关权益政策

暂无数据

收藏/分享

除非特别说明，本系统中所有内容都受版权保护，并保留所有权利。