Introducing 

Prezi AI.

Your new presentation assistant.

Refine, enhance, and tailor your content, source relevant images, and edit visuals quicker than ever before.

Loading…
Transcript

Revisiting the approach

Ideal image (I1)

✳✱*

Test image (I2)

Image enhancement

  • Grayscale conversion

Ensures the amount of data to be processed gets reduced from a 3-way channel to 1-way channel.

  • Gamma correction

To contrast the edges or defects more clearly.

Image processing steps

Morphological Operation

DEMO

  • We are using the morphological closing technique on the resultant binary image.
  • Closing is dilation followed by erosion.
  • This helps in closing small holes inside the foreground objects, or small black points on the object.

Image comparison

The test image(I1) and the ideal image(I2) are compared and the absolute difference(|I1-I2|) of the images is evaluated.

What is the approach?

Image processing and Morphological techniques.

Pre-requisites:

  • An ideal image and test image are needed (in a loss-less image format like .bmp).

  • The camera orientation and lighting should be constant for all the images captured.

  • Eight images are captured for each Surface Pro device covering all the possible faces.

METHODOLOGY

Our timeline

Business impact

  • Digitalising Supply Chain
  • Enhanced accuracy which is promised by the carefully chosen sequence of algorithms.
  • Reduced number of faulty Surface Pro devices.
  • Minimal human labor requirement.
  • Time optimization.
  • $640k savings per site per year

"High value products are produced at high speed against rigorous quality standards"

Problem Statement

  • Ensuring that the finished products are free from defects like scratches, cracks, discoloration and dents for better customer service is achieved by a manual approach in the industry.

  • This is prone to potential human error and defect escapes and many other disadvantages.

What can be done to improve this?

Image Segmentation and marking defects

Automate the process of defect detection in the industry by using Image processing methods and techniques.

  • The image is segmented into different regions.
  • Contours are continuous lines or curves that bound the full boundary of an object in an image. Such parts are extracted and marked as defects.

Automating Cosmetic Inspection of Surface Pro devices using Image Analysis

Current Industry Scenario

  • Automated defect detection in ceramic tile industry.

  • Wafer industry checks for varied patterns in their wafers.

  • The glass industry detects the defects in the glass sheets mainly using image processing techniques because of their higher precision and speed

Aditya Tiwari, Haridhakshini S.A.

Scope

A huge thanks to the

SC-MAKE Services team!

Technology used

  • Build the data repository of images to draw better insights.

  • Incorporate the features of Image classification which can bring out marking and categorizing the defects such as scratches, dents, dust or color variation.

Used Emgu CV which is a cross platform .Net wrapper to the OpenCV image processing library.

Edge Detection

  • It is a feature extraction technique which brings out any defect in the form of a scratch or a color variation or a dent or any deformity in the image.

  • We use the most efficient Canny Edge detection algorithm with threshold and linking threshold values as 25 and 5 respectively.
Learn more about creating dynamic, engaging presentations with Prezi