Course Attendees

Still no participant

Course Reviews

Still no reviews

 

INTRODUCTION

Machine vision systems are an integral part of Industry 4.0 because imaging is now a crucial component of automated inspection systems and robotic systems. The design of high-performance machine vision systems requires much more than applying the latest deep convolutional neural network algorithms or installing the highest specification camera. There are several key modules of a machine vision system namely the image sensor, optical assembly, electronic acquisition, computation and storage systems, preprocessing and classification algorithms and the imaging and software strategy. Optimum system design requires that all modules be jointly optimized and not unnecessarily overdesigned. Research literature rarely focuses on systems-level perspectives and tradeoffs while conventional image processing courses tend to assume ideal data acquisition scenarios and do not consider how individual machine vision modules can interact with or limit the performance of other modules. In systems design, poor specifications in one module often leads to performance-limitations that invalidate the high-performance specifications of other modules. Oftentimes, by choosing the correct specifications for one module, one can significantly simplify the design and specifications of other modules at significant cost savings with no degradation and sometimes with improvements in performance. This course focuses on systems-level design concepts to meet the competency gap. We first introduce important considerations in the specifications of key modules in a complete machine vision system. With a good understanding of individual modules, we go on to discuss how to assess and identify the performance-limiting modules, how modules interact and affect the performance of other modules and how to balance and tradeoff the specifications between modules to achieve cost or performance improvements. Participants of this course will learn systems-level thinking for the design and selection of specifications in a complete machine vision system.

OBJECTIVES

Upon completion of this course, participants will be able to:

1. Evaluate and select specifications for all components of a machine vision system namely the image sensors, optical system, acquisition electronics, image preprocessing, machine vision algorithm and software platform
2. Develop system level strategies to integrate or assemble complete machine vision systems for Industry 4.0 applications.

COURSE CONTENT

Day 1

• Machine vision systems components & applications in Industry 4.0 - case studies
• Image sensors specifications  & Evaluating image sensor performance
• Optical system specifications & Limitations of optical lenses – aberrations, aperture, mounting
• Data acquisition, storage specifications & Electronics systems for real-time vision processing
• Imaging strategies & Image features and invariants
• Object versus Imaging system scale and resolution
• Perspective versus orthographic views
• Single imager, multi or moving imager and light-field systems

Day 2

• Image preprocessing
• Non-linear filters and denoising approaches
• Computational imaging and super-resolution
• Machine vision algorithms
• Complex feature descriptors (edge and blob features)
• Support vector machines classifiers
• Deep convolutional neural networks
• Guidelines for selecting machine vision algorithms

Day 3

• System-level trade-offs and optimizations
• System SNR – balancing sensors, lenses, lighting
• System responsivity - balancing acquisition hardware, algorithm complexity, computational threads
• System accuracy – balancing sensor resolution, size, algorithms, speed
• System cost – balancing hardware, software and knowing your performance
• System robustness – balancing algorithms and acquisition systems
• System reliability – points of failure and variations
• Systems design for machine vision metrology

TRAINER

Dr Eric Ho Tatt Wei received his MS and PhD degrees in Electrical Engineering from Stanford University in Silicon Valley, USA specializing in computer hardware and VLSI systems, He designed and built several high performance imaging systems including low-light high-resolution microscopes for fluorescence and multiphoton imaging, miniature portable brightfield microscopes, low-cost benchtop machine vision systems and a fully automated robotic system using real-time machine vision to prepare fruit flies for brain imaging studies, His research work has been patented and featured in the New York Times. He is recently pursuing research in applications of deep neural network technology to brain network analysis and MRI brain imaging. (https://www.nytimes.com/2015/05/28/science/a-robot-that-can-perform-brain-surgery-on-a-fruit-fly.html) h>2  

Download the Course Brochure: Practical High-Performance Machine Vision Systems for Industry 4.0

REGISTRATION IS NOW OPEN2 To register, download the CAPE Professional Short Course Registration Form and email the completed form to cape@utp.edu.my.  

Our Main Teachers