Machine Vision Basics 06 - Camera Selection

Summary

This video script delves into the intricacies of machine vision for quality control, emphasizing the importance of camera and lighting selection. It covers the factors influencing camera choice, such as product size, defect detection, accuracy thresholds, and process speed. The script introduces two types of camera systems: smart cameras with integrated tools and algorithms, and vision system processors that handle image processing separately. It also highlights the capabilities of these systems in performing OCR and barcode reading, showcasing their versatility in industrial applications.

Takeaways

• 📐 Camera Selection: The process of selecting a camera for machine vision applications involves considering product size, defect size, accuracy requirements, and process speed.
• 🔍 Accuracy Threshold: It's important to determine whether the inspection process needs to be 100% accurate or if a 90% accuracy is sufficient for the application.
• 🔄 Process Speed: The number of objects inspected per minute is a critical factor in camera selection, which can range from 30 to 300 or more.
• 🌈 Color vs Grayscale: The need for a color camera versus a grayscale camera is determined by the type of inspection and application requirements.
• 🛠️ Smart Camera vs Vision System Processor: There are two types of camera systems - smart cameras with built-in tools and algorithms, and vision system processors that handle the processing while cameras relay images.
• 🔌 Smart Camera Setup: Smart cameras have power and Ethernet cords and come with different lens sizes, filters, and built-in lighting.
• 🖼️ Vision System Processor Layout: Vision system processors manage multiple cameras, with each camera connected via Ethernet and powered over Ethernet, acting as 'dummy' cameras.
• 📊 Real-time Monitoring: Data Logic's machine vision software offers real-time monitoring and statistics, including pass and fail rates, through a display module or software interface.
• 🔤 OCR Capabilities: The machine vision cameras are capable of performing optical character recognition (OCR), useful for reading barcodes and characters on objects.
• 🔢 Barcode Reading: While barcode readers are preferred for barcode-only tasks, machine vision cameras are useful for simultaneous object inspection and barcode reading.
• 📈 Data Presentation: The script mentions the use of pie charts and other statistics to represent the total counts, passes, and fails in the inspection process.

Q & A

What are the common factors to consider when selecting a machine vision camera?

When selecting a machine vision camera, one should consider the product size, minimum defect size, required accuracy, threshold for passing, process speed, and whether a color or grayscale camera is needed.

What is the importance of accuracy in machine vision applications?

Accuracy in machine vision is crucial as it determines the reliability of the inspection process. It can vary from needing to be 100% correct to being acceptable at 90% accuracy, depending on the application requirements.

How does the inspection speed affect the choice of machine vision system?

The inspection speed, measured in objects inspected per minute, influences the choice of system. Higher speeds may require more advanced or multiple cameras to maintain accuracy and efficiency.

What is the difference between a smart camera and a vision system processor?

A smart camera has all the tools and algorithms built-in, with power and Ethernet coming directly to the camera. A vision system processor, on the other hand, handles the processing while the cameras act as 'dumb' relays of images back to the processor.

Why would one choose a smart camera over a vision system processor?

A smart camera might be chosen for its simplicity and self-contained operation, which can be beneficial for applications where a compact and integrated solution is preferred.

What are the capabilities of the new vision processor mentioned in the script?

The new vision processor can support up to eight different cameras, allowing for more complex and multifaceted machine vision applications to be managed from a single processing unit.

How does a machine vision system with a processor display and monitor the inspection process?

The system can display the inspection process on a floor module or through machine vision software, showing real-time monitoring, statistics, and historical image data for analysis.

What is OCR and how are machine vision cameras capable of performing it?

OCR stands for Optical Character Recognition. Machine vision cameras capable of OCR can read and interpret text within images, which is useful for applications that require reading barcodes or other characters on objects.

Why would a barcode reader be preferred over a machine vision camera for reading barcodes?

A barcode reader is typically preferred for just reading barcodes due to its simplicity and efficiency. Machine vision cameras are more versatile for applications where both object inspection and barcode reading are required simultaneously.

What additional insights can the provided script offer regarding machine vision applications?

The script provides insights into the decision-making process for selecting camera types, the importance of accuracy and speed in inspections, the capabilities of smart cameras and vision processors, and the role of OCR in machine vision systems.

Outlines

00:00

📷 Camera Selection Criteria for Machine Vision Applications

This paragraph discusses the common approach to selecting cameras and lighting for machine vision applications. It emphasizes the importance of considering product size, defect detection, accuracy thresholds, process speed, and whether color or grayscale cameras are needed. It also touches on the type of inspection and the desired outcomes from the vision system. The paragraph introduces two types of camera systems: smart cameras with built-in tools and algorithms, and vision system processors that handle the processing while cameras act as simple image relays. The smart camera setup includes various lens sizes and filters, while the vision system processor can support multiple cameras and is responsible for all motor and drive control.

Keywords

💡Machine Vision

Machine vision is a technology that enables machines to see and interpret the world through image processing and analysis. It plays a crucial role in the video's theme by being the central focus of the discussion. The script mentions how different applications of machine vision require careful selection of cameras and lighting to achieve desired outcomes, such as inspecting objects for defects or reading barcodes.

💡Camera Selection

Camera selection is a critical process in machine vision systems, where the right camera is chosen based on various factors like the size of the object, the minimum defect size, and the required accuracy. The script emphasizes the importance of considering these factors to ensure that the camera can accurately capture the necessary details for the inspection process.

💡Defect Size

Defect size refers to the smallest flaw or imperfection that a machine vision system needs to detect on an object. In the context of the video, understanding the minimum defect size is essential for selecting the appropriate camera resolution and sensitivity to ensure that no defects are missed during the inspection.

💡Accuracy

Accuracy in machine vision is the measure of how precisely the system can detect and interpret the features of an object. The script discusses the importance of determining the required level of accuracy, such as whether a system needs to be 100% correct or if a 90% accuracy is acceptable, depending on the application.

💡Process Speed

Process speed is the rate at which objects are inspected by the machine vision system. The script mentions how the number of objects inspected per minute can influence the choice of camera and system setup, as faster speeds may require more advanced cameras and processing capabilities.

💡Color Camera

A color camera is a type of camera that can capture images in full color, as opposed to grayscale cameras that only capture shades of gray. The script discusses the decision between using a color camera or a grayscale camera based on the type of inspection required, highlighting that color cameras might be necessary for certain applications where color information is crucial.

💡Grayscale Camera

A grayscale camera captures images in shades of gray, which can be sufficient for some machine vision applications where color information is not necessary. The script contrasts grayscale cameras with color cameras, emphasizing that the choice depends on the specific requirements of the inspection process.

💡Smart Camera

A smart camera is a self-contained device that integrates the camera, processing tools, and algorithms within a single unit. The script describes smart cameras as having all necessary components, such as power and Ethernet, built-in, making them a complete solution for certain machine vision applications.

💡Vision System Processor

A vision system processor is a device that handles the image processing and analysis for multiple cameras connected to it. The script explains that these processors can manage multiple cameras, with each camera acting as a 'dummy' that simply relays the image back to the processor for analysis.

💡Optical Character Recognition (OCR)

Optical Character Recognition (OCR) is a technology that allows machines to recognize and interpret text from images. The script mentions that machine vision cameras can perform OCR, which is useful for applications where both object inspection and text reading, such as barcodes, are required.

💡Barcode Reader

A barcode reader is a device specifically designed to read barcodes on objects. The script differentiates between barcode readers and machine vision cameras, noting that while cameras can read barcodes, dedicated barcode readers might be preferred for straightforward barcode scanning tasks.

Highlights

Camera selection involves considering product size, defect size, accuracy, and process speed.

Accuracy thresholds can vary, with some processes requiring 100% correctness, while others might accept 90% accuracy.

The choice between color and grayscale cameras depends on the inspection type and application requirements.

Smart cameras contain tools and algorithms internally and are powered via Ethernet.

Vision system processors handle the processing for 'dummy' cameras, which relay images back to the processor.

Data Logic's new processor supports up to eight cameras, showcasing technological advancement in vision systems.

Vision system processors can output to display modules for real-time monitoring on the factory floor.

Machine vision software allows for detailed inspection, including previous images and real-time statistics.

Pie charts and statistics provide insights into total counts, pass, and fail rates in the inspection process.

OCR capabilities in machine vision cameras enable optical character recognition for applications requiring text reading.

Barcode reading is a common application for OCR, with the choice between barcode readers and vision cameras depending on inspection needs.

Machine vision systems can be used for simultaneous object inspection and barcode reading, useful for complex applications.

The transcript discusses the importance of selecting the right camera and lighting for machine vision applications.

Different lens sizes and built-in filters are available for smart cameras to suit various inspection needs.

The layout of a smart camera system includes power and Ethernet cords, with all processing done within the camera.

Vision system processors are connected to cameras via Ethernet, with the processor performing all the image processing tasks.

The transcript provides a sample mesh from a program, indicating the practical application of machine vision in real-world scenarios.

The importance of gray level and pixel considerations in machine vision is briefly mentioned towards the end of the transcript.

Transcripts

each machine vision application is

different but the way we go about

picking cameras and lighting kind of all

are the same so what goes into a camera

selection will take the products and the

size of it your minimum defect size that

you want to have on the object you're

intersecting required accuracy

what's your threshold what do you want

to be able to pass the does it need to

be a hundred percent correct every time

can it something be 90 percent accurate

and still pass your process speed for

example how many of these objects are

you inspecting per minute is it 30 is it

a hundred is it three hundred are you

required to have a color camera we're

going to lead with a grayscale camera

what is a color camera required what

type of inspection are you doing what

type of application is is where you're

looking for what do you want to

accomplish what are you looking to get

out of this machine vision camera and

then just your rules of vision and then

we'll go through all of those and then

we'll discuss our available cameras and

then we'll get your object and do some

testing and then just go from there with

DES logic we have two different types of

cameras we have a smart camera and then

we have a vision system processor or our

smart camera this is where all your

tools and all your algorithms are inside

of the smart camera you're gonna have a

power in a ethernet cord coming from the

camera everything is running through the

actual camera and you can kind of see

what the typical system of a smart

camera kind of what the layout is all

these cameras come with different lens

sizes different filters built-in

lighting and the next type of system we

have is a vision system processor this

is actually where all the motor and all

the drive and all the tools and all of

that good stuff is actually going to be

run through a vision processor the

cameras will have ethernet cords running

to them they'll be powered over Ethernet

but these cameras are actually what you

would call dummy cameras

they have nothing no type of horsepower

running on them what they're doing

they're just relaying the image back to

the vision system processor and the

processor is doing all the work

data logic actually we have a brand new

processor that came out several months

ago but it can actually we can actually

have eight different cameras running

from one vision processor and this is

kind of the typical system layout when

it comes to a vision system processor

this is kind of what you'll see you can

get this output on a display module on

your floor or if you pull up the data

logic machine vision software you can

take a look at it like this but you can

you see you have the object you're

inspecting you have previous images down

below and you also have real-time

monitoring and statistics over here on

the bottom left with the pie chart you

have the total counts you have your

passes and your fails we also do OCR

these machine vision cameras are capable

of performing an OCR optical character

recognition we do do some OCR

applications as well

these cameras are able to read barcodes

but typically if we're reading a barcode

just a barcode we want to stick with a

barcode reader typically reading

barcodes comes in if you want to inspect

an object that has a barcode on it and

you have to inspect the object and read

the barcode at the same time comes in

very useful then for the types of

applications like that

and here's the sample mesh from the

program I think we might win over this a

few times and fuelie up slides that up

we went over earlier regarding gray

level in pixels