Training Workshop on

MACHINE VISION (IMAGE PROCESSING FUNDAMENTALS)

[Total number of engineers trained to date: 30]

(Please contact trainer for in-house training)

HRD Corp claimable (SBL Khas 10001259284)

​

Synopsis:

​

The pre-processing of digital images and extraction of useful features from the images for the purpose of decision making, sorting and classification are crucial for the successful implementation of machine vision solutions in the industry. In this two-day workshop, participants will be guided into understanding how images are processed for the purpose of enhancement, segmentation, feature extraction, defect identification and localization. Basic methods of contrast enhancement, gamma correction, histogram equalization, noise removal by various types of filtering operations, image sharpening, binary and grayscale morphological operations, edge detection using gradient and Laplacian operators, line detection, as well as image segmentation will be demonstrated using Python and OpenCV though hands-on guided activities​.

 

The training workshop is divided into activity-based theory (70%) to strengthen the fundamental knowledge of the participants, and practical activities based on digital processing of captured images (30%). The activities will be based on practical applications of image processing, such as detection and localization of stamping defects in a lead frame, detection of soldering defects, wire bond defects, wafer imprinting defects, blister pack inspection etc. The practical part will include detection of missing components on a printed circuit board, and blister pack inspection for missing capsules.

​

Course outcomes:

​

• Identify the five main approaches in digital image processing

• Distinguish between point, global, neighbourhood, geometric and temporal operations

• Compare various contrast enhancement methods

• Apply various types of filters to remove noise in image

• Apply various types of edge operators for edge detection

• Apply binary and grayscale morphological dilation and erosion operations

• Extract object properties from image for identification and classification

• Write Python-OpenCV codes to solve typical machine vision problems

​

Course content:

​

Day 1 - Part 1: Image Enhancement​

• Basics of digital image processing

• Digital convolution and its application 

• Image brightness modification and negation

• Contrast enhancement and gamma correction

• Histogram equalization

• Image filtering operations (average, Gaussian, median, k-nearest neighbour, sigma)

• Image sharpening using unsharp masking

• Activity 1: Determine mapping function for contrast enhancement by histogram stretching (see activity sheet)

• Activity 2: Effect of gamma correction on image (see activity sheet)

• Activity 3: Determine gray value distribution in output image after histogram equalization (see activity sheet)

• Activity 4: Compare contrast stretching, gamma correction and histogram equalization operations on image (see activity sheet)

• Activity 5: Threshold image using programming and OpenCV command (see activity sheet)

• Activity 6: Apply median filter to remove impulse noise in image due to dead/stuck pixels (see activity sheet)

• Activity 7: Determine output pixel value for various image filtering operations (see activity sheet)

• Activity 8: Sharpen image using unsharp masking (see activity sheet)

• Activity 9: Apply Gaussian filter to remove background noise

​

​

Day 1 - Part 2:  Edge Detection & Segmentation

• Basics of edge detection

• First and second order gradient operators

• Common edge operators (Roberts, Prewitt, Sobel, Canny, Laplacian)

• Basics of line detection using Hough transform

• Image segmentation by thresholding (multi-level global and adaptive)

• Activity 10: Compare effects of Prewitt, Sobel and Canny edge operators (see activity sheet)

• Activity 11: Detect lines using Hough transform (see activity sheet) (download code)

• Activity 12: Segment image and measure dimension of gap spacing in leadframe (see activity sheet)

• Activity 13:  Segmentation using simple thresholding, Otsu’s binarization and adaptive thresholding (mean and Gaussian) (see activity sheet)

​

Day 2 - Part 3: Morphological Operations, Feature Extraction & Classification

• ​​Erosion and dilation operations

• Opening and closing operations

• Extraction of region properties

• Rule-based classification

• Activity 14: Apply morphological operations on leadframe to find centroids of die attach pads (see activity sheet)

• Activity 15: Effect of grayscale morphology on printed characters on a rubber foam (see activity sheet)

• Activity 16: Extract basic properties of standard objects (see activity sheet)

​

Day 2 - Part 4: Practical Applications of Image Processing

• Activity 17: Detect wafer imprinting defects (see activity sheet)

• Activity 18: Detect defect on a PCB (see activity sheet)

• Activity 19 (Practical - open-ended group activity): Count value of Malaysian coins (see activity sheet)

• Activity 20 (Practical - open-ended group activity): Inspect blister pack for missing tablets/capsules (see activity sheet)

​

Course duration: 14 hours

 

Course fee:  Please contact the trainer for quotation

​

Who should attend: Junior engineers or anyone new to developing machine vision algorithms for product inspection and quality control, particularly those who have never taken an undergraduate course on machine vision. This training is also suitable for anyone who wishes to enhance their basic knowledge in image processing. You may assess yourself whether you need to take this training by answering the questions in this simple quiz.

 

Maximum number of participants: Limited to only 10 participants per session

​

See sample slides: Slide 1, Slide 2, Slide 3, Slide 4, Slide 5

​

​

Download sample slides (handouts)

​