LIMITED OFFER
Save 50% on book bundles
Immediately download your ebook while waiting for your print delivery. No promo code needed.
Quantified Automated Optical Inspection
Fully Enabling Machine Learning for Enhanced Hardware Assurance
- 1st Edition - June 1, 2025
- Editor: Nathan Jessurun
- Language: English
- Paperback ISBN:9 7 8 - 0 - 4 4 3 - 2 4 7 1 0 - 1
- eBook ISBN:9 7 8 - 0 - 4 4 3 - 2 4 7 1 1 - 8
Quantified Automated Optical Inspection: Fully enabling machine learning for enhanced hardware assurance outlines various inspection modalities, the growing popularity of machin… Read more
Purchase options
Institutional subscription on ScienceDirect
Request a sales quoteQuantified Automated Optical Inspection: Fully enabling machine learning for enhanced hardware assurance outlines various inspection modalities, the growing popularity of machine learning (ML) techniques for enhanced AOI, critical bottlenecks in these approaches, and solutions that increase scalability, reliability, and accuracy of ML-based AOI techniques. As a result, readers will clearly see which trends must be updated to properly incorporate deep learning strategies given the constraints present in PCB analysis. These constraints include limited publicly available data, lack of rigor/consistency in ground truth collection, lack of result quantifiability, and limited diversity in fundamental ML approaches. Solution spaces will be explored such as enhanced generative models to quantify PCB component properties, verification against component datasheets, and codified recommendations for rigorous data aggregation and dissemination.
- Features a union between state-of-the-art optical inspection approaches and machine learning techniques - the first of its kind
- Includes conversations around ML for hardware assurance, including discussions of their limitations and technical considerations when applied to PCB images
- Serves as a handy reference for engineers wishing to correctly apply ML techniques without requiring a strong background in the subject
Quality assurance engineers, optical inspection engineers, hardware assurance specialists, hardware engineers, quality control managers, manufacturing managers, electronic hardware designers, Systems engineers, failure analysts, QA engineers, government program managers, hardware engineering faculty & students, and standards committees such as ISO and IEEE
1. Overview of imaging modalities Introduces physical inspection via optical imaging, including the benefits and tradeoffs between each modality. The discussion will include THz, IR, optical, and X-ray wavelengths.
2. History of AOI Discusses why AOI is the most popular inspection technique and its evolution over time.
3. Advent of Machine Learning in AOI Provides a survey of machine learning (ML) techniques for AOI over the last decade, especially in contrast to traditional computer vision (CV) approaches. This includes their methodology, and source domains from which they were adapted.
4. Quantifiability in current ML methods Explains the critical gap that exists between modern ML practices and what translates to assurance objectives. This chapter frames all future arguments and suggestions for quantifiable and explainable assurance workflows.
5. Addressing methodological bottlenecks Enumerates the common problems occurring when methods from other domains such as medical imaging and autonomous vehicle scene analysis are applied to PCB inspection. The focus is on fundamental differences between these domains and PCB images rather than specific technical implementations. This chapter will be useful for audiences without ML experience to make their own judgement calls about which future ML concepts are adaptable even after this book is published.
6. Addressing data bottlenecks One of the biggest shortcomings with applying AI to PCB inspection is the lack of available data. This chapter discusses resulting problems, proposed solutions, and calls to future researchers for mitigating this bottleneck in the future.
7. Looking forward: Newly enabled research Outlines which assurance objectives will be achievable once the above bottlenecks are addressed, and how these compare to traditional capabilities. Proof of concept implementations are proposed
2. History of AOI Discusses why AOI is the most popular inspection technique and its evolution over time.
3. Advent of Machine Learning in AOI Provides a survey of machine learning (ML) techniques for AOI over the last decade, especially in contrast to traditional computer vision (CV) approaches. This includes their methodology, and source domains from which they were adapted.
4. Quantifiability in current ML methods Explains the critical gap that exists between modern ML practices and what translates to assurance objectives. This chapter frames all future arguments and suggestions for quantifiable and explainable assurance workflows.
5. Addressing methodological bottlenecks Enumerates the common problems occurring when methods from other domains such as medical imaging and autonomous vehicle scene analysis are applied to PCB inspection. The focus is on fundamental differences between these domains and PCB images rather than specific technical implementations. This chapter will be useful for audiences without ML experience to make their own judgement calls about which future ML concepts are adaptable even after this book is published.
6. Addressing data bottlenecks One of the biggest shortcomings with applying AI to PCB inspection is the lack of available data. This chapter discusses resulting problems, proposed solutions, and calls to future researchers for mitigating this bottleneck in the future.
7. Looking forward: Newly enabled research Outlines which assurance objectives will be achievable once the above bottlenecks are addressed, and how these compare to traditional capabilities. Proof of concept implementations are proposed
- No. of pages: 250
- Language: English
- Edition: 1
- Published: June 1, 2025
- Imprint: Academic Press
- Paperback ISBN: 9780443247101
- eBook ISBN: 9780443247118