JADAK VIDEO: Machine Vision Building Blocks
This video provides an introduction to a handful of tools and techniques that can be used for Machine Vision development. Using JADAK’s Clarity End-User Machine Vision Software as the platform to provide these examples, the video explains in basic terms a few of the common image evaluation processes.
*Want to learn more about JADAK’s Medical Machine Vision technologies? Register for a Tech Consultation Day from the experts at JADAK! Also, be sure to visit our RFID Products and Services, Barcode and Machine Vision pages.
Below is the video transcript.
Some Basic Building Blocks to Machine Vision…
How Machine Vision works is not necessarily the easiest technology to explain or understand. Think of the old saying that “a picture is worth a thousand words” and you start to realize how much information there really is in one image.
It’s something we often take for granted in our own sense of sight. Creating techniques that give machines the ability to recognize, assess and compile this visual information requires tremendous creativity to break down vision into specific processes.
This is exactly what Machine Vision software does.
Machine vision uses a combination of hardware that includes an almost unlimited range of cameras capable of “seeing” parts of the spectrum or events with capacity and accuracy that is simply impossible for humans to do.
For the purposes of this video we will focus on only a handful of the basic machine vision techniques.
Do not make the mistake to assume that these basic vision tools produce only basic results, a hammer, saw, wood plane and chisel when put into the right hands result in some of the most significant creations ever known.
JADAK is an industry leader in Medical Machine Vision with a staff of the most highly trained and expert engineers in the science of Machine Vision.
JADAK engineers provide our customers with ability to take on the most challenging problems specific to medical vision.
These same engineers have also created a Machine Vision Software package called Clarity that is intended to allow the end-user to create machine vision solutions on their own.
Clarity is based on Plug-In vision tools, executed according to a “scripted” job list to accomplish machine vision tasks. This video will highlight these plug-ins and the associated icons and visuals to help illustrate these Machine Vision Building Blocks.
There are 6 Machine Vision Plug-Ins…
- Pattern Match
Let’s get started…
We are going to start with Count…
Obviously everything is based on an image … and in this case it’s an image of 7 M&Ms. Count starts with establishing a ROI or “Region of Interest”.
This is accomplished by dragging an expandable dotted rectangle over the image to enclose the candies. As we expand this rectangle you see the M&Ms become filled with different colors and an object count is tracking the number of “blobs” as they are called.
Pass/Fail requirements can be defined that confirm the maximum and minimum number of objects required.
Getting in Shape…
The next Plug-In to look at is Pattern Match…
When using the Pattern Match Plug-In, you’re presented with “Pattern Training” This is where you have to teach Clarity what constitutes the pattern that has to be matched.
You can select either Pixel Matching or Edge Pattern Matching.
In this case we are going to use Edge Pattern Match, again we are establishing a region of interest and instead of finding a blob, Pattern Match identifies the outer edge of the shape.
This is articulated with the blue outer line and centered red dot. Once trained, Pattern match can identify other shapes and score those as compared to the source “Trained” shape. Additional options can by applied to the pattern to further refine the evaluation results.
Now You See It…
The plug-in that’s a little harder to understand and properly use is Presence/Absence…
Presence/Absence can detect a presence or absence of a feature in an image by looking at pixel thresholds, average brightness values or high contrast edge pixels within the set ROI.
You can use Presence/Absence in three manners:
- Threshold: This counts the number of cells above a certain threshold, marked blue in the ROI.
- Brightness : This calculates an average value of brightness in the image.
- EdgePixels : This counts the number of edge cells between a low and a high threshold value.
Getting the Message… Barcode Reading
Depending on how Clarity has been set-up, there’s an optional Decode Plug-in. If you have an image that contains a 1D or 2D barcode, you can use this plug-in to decode this barcode.
Finding what you’re looking for… Shapes
If you select the (Shape) Finder Plug-in , a Step box is added in a red rectangle, called Finder, underneath the green Acquire Image-step box. The Finder plug-in can be used to search for (vertical) lines, corners and circles in an image.
You’ll see a small red dotted square inside your image. This is the Finder plug-in’s Region of Interest (ROI).
Within this ROI, the Finder plug-in will look to find a line, corner or circle.
If you drag the ROI to a different part in the image, where a shape can be found, and depending on the settings of the Finder plug-in, the ROI may become green, indicating a successful ‘shape-find’.
Going The Distance…Measure
Finally, there’s the Measure Plug-in. The function of this plug-in is to measure the distance in pixels between two or more previous Plug-In steps in your job list.
As an example, it will measure the distance between an identified Pattern Match and an identified Shape Find. The real power of this Plug-in is that it can utilize and leverage identified components to verify or measure complex relationships within the image.
This video has only skimmed the surface of Machine Vision, how it functions and what it is capable of. But as was said before, good tools in the hands of creative, knowledgeable experts can produce amazing results. And Machine Vision never gets tired, blinks or takes it’s eyes off of the task at hand.
This is Mark Waterman, Director of Design at JADAK saying, thanks for watching.[/expand]