毕业设计
题 目：
基于深度学习的布匹瑕疵识别系统的设计与实现
副 标 题：
学 生 姓 名：
陈婷婷
学 号：
1610503407
所 在 学 院：
计算机与信息工程学院
专 业、班 级：
软件工程/软件 1604
指 导 教 师：
宋伟
完 成 日 期：
2020 年 05 月 21 日
南通理工学院本科生毕业设计
I
摘 要
布匹的品质控制在纺织品行业中起着相当重要的作用，然而影响布匹品质控制的重
要因素是布匹瑕疵。传统的目标识别检测一般都是通过人工的方式，但是随着目标的增
加以及人员的疲惫，人工识别方法比较慢而且识别准确率也得不到保证。现在，布匹瑕
疵自动识别的相关探索虽然已经取得了一定的成果，然而因为收集图像受光线、噪声等
相关因素，以及瑕疵具有类别较多等问题，使得布匹瑕疵识别仍然是一个具有挑战性的
研究课题。
本系统使用 Python 语言开发，使用 DenseNet 模型，利用 Keras 框架，通过创建和
训练深度学习神经网络，然后选择识别图片，从而进行布匹瑕疵的识别，识别结果分为
扎纱、擦伤、擦毛、擦洞、明嵌线等多种结果。
本文实现了自动化识别布匹瑕疵的过程。可以将本系统应用到纺织品行业中，对有
瑕疵的布匹进行自动检测，自动识别瑕疵类型，使布匹检测更加高效，便捷。
关键词：瑕疵识别；Python；神经网络；DenseNet
南通理工学院本科生毕业设计
II
Abstract
The quality control of cloth plays an important role in the textile industry, but the cloth
defect is an important factor affecting the quality control of cloth. The traditional target
recognition and detection is generally carried out by manual means, but with the increase of
targets and the exhaustion of personnel, the manual recognition method is relatively slow and
the recognition accuracy cannot be guaranteed. At present, although some achievements have
been made in the exploration of automatic recognition of cloth defects, the image collection is
still a challenging research topic due to the light, noise and other related factors, as well as
various types of defects.
This system is developed in Python language, using DenseNet model and Keras
framework. By creating and training deep learning neural network, it can select recognition
pictures to identify cloth defects. The recognition results include such types as yarn binding,
scraping, yarn abrading, holes, and explicit inserting lines.
In this paper, the process of automatic identification of cloth defects is realized. This
system can be applied to the textile industry to automatically detect the defective cloth and
identify the types of defects, so as to make the cloth detection more efficient and convenient.
Key words: defect recognition, Python, neural network, DenseNet
南通理工学院本科生毕业设计
I
目 录
摘 要 .....................................................................................................................................I
Abstract.....................................................................................................................................II
1 绪论.........................................................................................................................................1
1.1 研究目的及意义 ...............................................................................................................1
1.2 国内外现况分析 ...............................................................................................................1
1.3 研究内容 ...........................................................................................................................2
1.4 本文结构安排 ...................................................................................................................3
1.5 本章小结 ...........................................................................................................................3
2 相关技术及主要开发工具.....................................................................................................4
2.1 卷积神经网络 ...................................................................................................................4
2.2 Python 简介......................................................................................................................5
2.3 图像识别简介 ..................................................................................................................6
2.4 评测方法 ..........................................................................................................................8
2.5 本章小结 ........................................................................................................................10
3 可行性分析与需求分析.......................................................................................................10
3.1 系统可行性分析 .............................................................................................................10
3.1.1 经济可行性...............................................................................................................10
3.1.2 技术可行性...............................................................................................................11
3.2 系统需求分析 .................................................................................................................11
3.2.1 系统主要目标...........................................................................................................11
3.2.2 功能需求分析...........................................................................................................11
3.2.3 性能需求...................................................................................................................12
3.3 开发环境 .........................................................................................................................13
3.4 本章小结 .........................................................................................................................13
4 布匹瑕疵识别系统设计.......................................................................................................14
4.1 布匹瑕疵识别系统功能设计 ........................................................................................14
4.1.1 功能结构设计..........................................................................................................14
4.1.2 业务流程设计..........................................................................................................14
4.2 布匹瑕疵识别系统功能详细设计 .................................................................................15
