附录B外文原文
On the Application of Automated Machine Vision for Leather Defect Inspection and Grading: A Survey
MASOOD ASLAM , TARIQ M. KHAN , (Member, IEEE), SYED SAUD NAQVI , GEOFF HOLMES , AND RAFEA NAFFA
ABSTRACT
Reliably and effectively detecting and classifying leather surface defects is of great importance to tanneries and industries that use leather as a major raw material such as leather footwear and handbag manufacturers. This paper presents a detailed and methodical review of the leather surface defects, their effects on leather quality grading and automated visual inspection methods for leather defect inspection. A detailed review of inspection methods based on leather defect detection using image analysis methods is presented, which are usually classified as heuristic or basic machine learning based methods. Due to the recent success of deep learning methods in various related fields, various architectures of deep learning are discussed that are tailored to image classification, detection, and segmentation. In general, visual inspection applications, where recent CNN architectures are classified, compared, and a detailed review is subsequently presented on the role of deep learning methods in leather defect detection. Finally, research guidelines are presented to fellow researchers regarding data augmentation, leather quality quantification, and simultaneous defect inspection methods, which need to be investigated in the future to make progress in this crucial areaofresearch
INDEX TERMSLeather defects, segmentation, classification, machine learning, computer vision.
- INTRODUCTION
Millions of tons of hides and skins are generated as coproduct from the slaughtering of animals for their meat each year. They are mostly converted into leather, the most important economic by-product of the meat industry. In 2003, the global leather industry produced approximately 18 billion ft2 , with an estimated value of US$ 40 billion [1]. Developing countries now produce more than 60% of the leather requirements world-wide. New Zealand hides and skins, especially herd skins, make a major contribution to leather worldwide by providing raw skins for the tanning industry [2]. Skins are mostly sourced from sheep, cow, deer, and goat in New Zealand. In 2011, 75% of sheep and lambskin were exported, mainly to the garment industry [2].
Before leather can be exported as a manufactured product, it must undergo numerous processing steps. The general leather processing steps include preparation for tanning, tanning, and finishing. These processing steps have multiple stages that depend upon the type of material used and the kind of leather required as a product. The processed leather is then subjected to leather quality grading, which is the process of categorizing leather based on the surface defects found during the inspection. The high demands for quality assurance are driven by global customer requirements, and increasingly rejection costs. Accordingly, the inspection of skin surface defects is essential for objectivity and reliability. Currently, the process of surface defect inspection and grading is carried out by human inspectors. The large scale of leather production makes defect inspection a labor-intensive and time-consuming process, which can be a potential bottleneck in the production pipeline,
FIGURE 1.Overall pipeline for leather visual defect inspection - a guideline for machine vision systems.
thus motivating the design and development of automated visual inspection systems for surface defect detection and grading.
This decade has seen astounding progress in the application of intelligent systems to real-world problems in areas including but not limited to medicine, telecommunications, finance, medical diagnosis, transportation, information retrieval, energy and many more. The urge for automation has revolutionized the industry sector with expert and intelligent systems finding applications in almost all kinds of industrial processing, ranging from resource optimization to industrial inspection. Intelligent machine vision systems have been at the heart of industrial inspection and surveillance for the past two decades. Image analysis based methods proposed for industrial inspection include both heuristic and machine learning methods. Despite being an important subject in industrial inspection, leather defect inspection has not received much attention yet. The majority of methods that exist for visual defect inspection of leather are heuristic with only limited studies that explore machine learning options for robust performance. Leather quality grading based on defect inspection is an important area of research and rapid advancements in intelligent systems for automatic leather grading are expected in the near future.
Advancements in expert deep learning based systems and their ability to surpass human performance has increased their application in numerous fields including healthcare, automotive industry, telecommunications, industrial visual inspection and many more during recent years. Most convolutional neural networks (CNN) based methods proposed for various computer vision applications can be categorized by one of the following tasks: image classification, detection, semantic segmentation and multiclass classification. Owing to their success in the aforementioned tasks and their ability to match or surpass human level performance, we present a complete deep networks based machine vision pipeline for visual defect grading, which can be utilized by future researchers as a guideline. Figure 1 shows the recommended pipeline for leather defect inspection.
In the first stage defect detection and classification is performed in parallel with defect segmentation. Based on theinformatio
剩余内容已隐藏，支付完成后下载完整资料
英语译文共 31 页，剩余内容已隐藏，支付完成后下载完整资料
资料编号：[607091]，资料为PDF文档或Word文档，PDF文档可免费转换为Word
以上是毕业论文外文翻译，课题毕业论文、任务书、文献综述、开题报告、程序设计、图纸设计等资料可联系客服协助查找。
