Machine Vision Archives | DMC, Inc.

Automated Vision Inspection for Aerospace

Sam Levin — Tue, 18 Mar 2025 00:00:00 +0000

DMC provided hardware consultation along with designing & developing custom manual test control software to allow Client’s R&D team to productively test and refine the system components. Upon collection & analysis of test data, an automated test sequence was added, along with image processing features, that improved the performance of the prototype and allowed detailed image analysis.

DMC worked with the client to identify potential areas where inspection speed could be increased by modifying image processing algorithms, hardware, or test sequence. With estimated cost and benefits of each option under consideration, the client was able to make informed decisions and improve the program as budget allowed.
The following image processing steps are currently included in the program utilizing NI vision tools:

Dot matric calibration for image distortion correction
Background image subtraction
Combining raw images to form a single composite image from each scan
Stitching of composite images of different areas of the part to allow analysis of larger parts
Brightness and contrast adjustment
Histogram and line profile of selected areas
Point-to-point measurements in real-world units
Color mapping to custom color scale

DMC engineers continue to work with the client to plan and implement new features for accommodating complex image processing challenges.

Learn more about our Test & Automation Solutions for Aerospace & Defense and contact us today for your next project.

The post Automated Vision Inspection for Aerospace appeared first on DMC, Inc..

PID Control of a Furnace With an Automated Vision System

Sam Levin — Wed, 16 Dec 2020 00:00:00 +0000

This client produced a crystallized material used in components of common electronics such as your phone screen. To produce these crystals, the granular form of the material is heated in a furnace over a seven-day cycle. During this cycle, the temperature needed to change in specific areas of the furnace because it was important to control the location of the solid-to-liquid interface.

Achieving such precision required PID control of the furnace. PID control describes the continuous adjustment and output variation to accurately control a process and mitigate deviation. In this case, PID control would increase the temperature in a specific location in the furnace from 2500°C to exactly 2600°C—not from 2500°C to 2700°C to 2600°C.

DMC implemented a vision system using a specialized camera that can withstand extremely high heats to monitor production. The camera recorded images at a user-specified time interval, communicated with a PLC, and then adjustments were made accordingly. Therefore, the client still maintains some manual control of the process, but it does not require around-the-clock monitoring by an employee.

The production of these crystals is lengthy, requires extreme precision, and constant monitoring. DMC developed a specialized vision system to automate this process, helping our client cut down on costs while maintaining quality standards. The client can also adjust the parameters of the system as needed, so this solution is long term and sustainable.

Read more about DMC’s Test and Measurement expertise and contact us to get started on your next project.

The post PID Control of a Furnace With an Automated Vision System appeared first on DMC, Inc..

Machine Vision System to Detect Defects

Sam Levin — Mon, 14 Dec 2020 00:00:00 +0000

The client wanted to improve their quality control, so they came to DMC to design a system that detects irregularities in their product: chrome bars. As a solution, DMC developed a system which used a line-scan camera to capture images of the chrome bars as they were rotated in front of the camera. This allows for the vision system to full inspect all sides of these 3D parts and identify all defects on all sides of the part. Due to the reflective nature of the parts to be inspected, careful consideration was taken to ensure that the selected camera and illumination would minimize impacts of glare which could appear as a false defect in the inspection.

In addition to performing the image inspection, the software is designed to communicate with PLC controlling the line, using digital inputs and outputs. The digital outputs are used to notify the PLC system when a defect was found. The software had to be able to handle multiple different lengths of chrome bars. Therefore, the system used digital inputs from the PLC to identify whether a bar was present for inspection or not. The software provided a user interface which provided the following key functionality:

Display of live images with defects highlighted
Configuration of defect sizes for small, medium, and large defects
Recent defective images were buffered in memory
Functionality to support saving images to disk

Learn more about DMC’s Test and Measurement Automation expertise and contact us with any project inquiries.

The post Machine Vision System to Detect Defects appeared first on DMC, Inc..

Customizable Vision-Based Inspections

Sam Levin — Thu, 09 Jul 2020 00:00:00 +0000

The high-level operation of this system involves a fastener feeding system that loads fasteners onto a dial. The system can run as fast as 1000 parts per minute for some of the parts. Each fastener’s position is tracked by the FPGA as the Kollmorgen Servo rotates the dial at speeds from 200-800 rpm, moving fasteners past several Basler cameras. The images from these Basler cameras are processed using NI VBAI based on specific profile guidelines configured by the client.

The cameras take side and top view of the fasteners. The images allow them to search by screw head to make sure it is the right size, shape, length, pitch, etc. from the side of the screw.

The results of these inspections are sent to the LabVIEW program where each screw is matched with its correlating pass/fail data from each camera. Each fastener that passes the inspection is blown off the dial with a puff of air and passed through a chute into boxes on a pack line which are automatically progressed as boxes become full. The inspection system is boxing up the parts, so they are ready for shipment and counted into appropriate quantities when they come off the line.

In this case, the client is generating the NI VBAI inspections for many reasons.

1. VBAI has an easy to use interface for generating inspections, which makes it simple for the client to specify inspections without in-depth programming knowledge.

2. The client has the most in-depth knowledge of the inspections that need to be performed, which makes them the ideal candidate for generating those inspections.

3. The client generates custom fasteners, which means that new products are always in production. Using VBAI eliminates the need for the client to come back to DMC to generate new inspections for each new product.

Learn more about DMC’s LabVIEW Vision Expertise.

The post Customizable Vision-Based Inspections appeared first on DMC, Inc..

3D Vision Inspection with SICK TriSpectorP and AppStudio

Sam Levin — Wed, 06 May 2020 00:00:00 +0000

DMC worked with one of the world’s leading suppliers of frozen food products to add a SICK TriSpectorP programmable 3D line scanner to improve their Quality Assurance process. The TriSpectorP was programmed in SICK AppStudio using the Lua programming language.

The customer’s existing system was only able to generate pass/fail results for a single product at a time. DMC’s support was requested because of our extensive experience with vision applications and image analysis routines. The TriSpectorP gave DMC the flexibility to measure and record actual data of entire batches to quantify the results of the Quality Assurance process.

SICK’s custom programming environment and built-in API suite allowed DMC to rapidly develop a program to extract actual measurements and generate batch reports.

The intuitive user interface allowed the customer to adjust the settings of the image acquisition process to maximize the performance of the system.

SICK also has a separate software package called AppManager which allowed the customer to store and deploy different programs to their TriSpectorP without having a development license.

The customer was so satisfied with the results of the analysis for a single image of the product that they requested additional features be added to the system. In order to increase the size of their sample batch, they requested we add the functionality to report results of multiple images. We were also able to easily add new analysis routines for other product types to the system because of the flexibility provided by the programming environment of SICK AppStudio.

Learn more about DMC’s work for the food and beverage industry and our vision inspection expertise. Contact us for project inquiries.

The post 3D Vision Inspection with SICK TriSpectorP and AppStudio appeared first on DMC, Inc..

Layer Imaging and Inspection Programming

Sam Levin — Wed, 03 Jul 2019 00:00:00 +0000

DMC started this project by collaborating with the client to develop a method to measure layer heights of stacked material quickly and accurately. Under natural lighting, it was difficult to distinguish the adhesive layer transitions from regular variation in layer color.

To overcome this, DMC developed a method utilizing multiple images under different lighting conditions and a bandpass filter on the camera. This method allowed DMC to accentuate the overall part dimensions in the first image to identify the top and bottom of the part. Then lighting and camera settings were modified to image the layer transitions using fluorescence for the second image.

Once the method was determined, DMC wrote a LabVIEW program to automate the process. This programming process involved turning on and off appropriate lighting and modifying camera settings before acquiring each image. The program also allows users to create and save recipes with expected layer heights for each part under test.

These recipes are used to determine regions of interest (ROIs) for the edge finding algorithms that identify the bottom, top, and layer transitions of the part. During operation, the user interface displays the lines on the acquired image as well as the measured layer heights in real-time. Measurements that are out of tolerance are flagged, and all measurements record to a file for part traceability.

A dot grid calibration and NI IMAQ tools were used to undistort images and to convert pixel measurements to real-world units. This method allows the user to quickly calibrate the system by taking images of a printed grid of dots placed in the same spot as the part under test.

Learn more about DMC’s Test and Measurement Automation Solutions and our partnership with National Instruments.

The post Layer Imaging and Inspection Programming appeared first on DMC, Inc..

Keyence Camera Integration with a Siemens S7-300 PLC

Sam Levin — Tue, 14 Mar 2017 00:00:00 +0000

DMC worked with a leader in the confectionery industry to add a Keyence CV-X series camera vision system to an existing packaging machine. DMC integrated the camera with a Siemens S7-300 series PLC via digital IO. The Keyence camera was programmed with the CV-X programming software, and the Siemens PLC was programmed with Step 7 Simatic Manager. A user interface was used to control the system created on a Siemens TP1500 Comfort Panel and utilized Siemens TIA Portal and device proxy for programming.

The integration was started by mounting and programming the CV-X100 camera with the Keyence programming software. The camera was taught to distinguish good product from bad product. Once the camera was reliably detecting bad product, a digital signal was sent to the PLC to inform the rest of the system to perform the reject logic.

To automate the detection and rejection process, the PLC was programmed to send a digital output pulse to the Keyence camera to trigger a picture to be taken and return the result to the PLC. This result was then placed into a bit shift register to track the bad product as it was transferred to the packaging area. If the camera detected bad product, the PLC would prevent the machine from pulling a package and would reject the bad product for rework.

The machine packages at rates close to 1Hz, but the camera system displayed capabilities to function at much faster rates. With the flexibility of the Keyence camera programmable detection, this system is easily expandable and modular.

The post Keyence Camera Integration with a Siemens S7-300 PLC appeared first on DMC, Inc..

Automated Vision Inspection of Chemical Droplets

Sam Levin — Mon, 14 Nov 2016 00:00:00 +0000

DMC’s customer was dispensing small drops of chemical reagent into a well on a clear plastic medical device. The precise location and volume of the drop is critical for the performance of the microfluidics medical device. To ensure each drop was dispensed properly, the customer asked DMC to implement an automated vision inspection that confirms the location and size of each droplet.

The droplets are clear and dispensed onto a clear plastic background. To provide contrast, the customer added fluorescein dye to the droplets. Fluorescein is an organic compound that fluoresces green when exposed to blue light. DMC exploited this fluorescent property by illuminating the droplets with a blue LED light and using a green filter in front of the camera. This caused the background to be dark with only the drops showing up brightly. The area of each droplet was determined using simple blob analysis tools.

To ensure accurate positioning of the droplets relative to the device, a fiducial measurement was needed. DMC added a white light inside the vision system. The white light was controlled by a digital output on the Basler camera. For each device under test, DMC’s software acquired two images: a fiducial image illuminated using the white light, and an inspection image illuminated only by the blue light. Using the fiducial image, DMC’s software found the position and rotation of the current device under test and then calculated the expected positions of the droplets. The inspection image was analyzed to confirm that the drops were actually contained by the acceptable position boundary.

Result data was logged to NI’s TDMS format for later analysis and traceability purposes. The software also provided the option to save each acquired image.

The post Automated Vision Inspection of Chemical Droplets appeared first on DMC, Inc..

LabVIEW Consulting and Training for Internal Development Team

Sam Levin — Mon, 07 Apr 2014 00:00:00 +0000

The client began with some prior LabVIEW experience but had not yet designed an entire application before. The client had written and tested individual VIs for system components but was unsure how to piece these functions together to form a cohesive test application. Apart from delivering final software, the client was interested in investing in one-on-one training with LabVIEW experts.

DMC met with the client software team, gathering system information and requirements. After reviewing all pieces and defining an architecture that fit their application, development tasks were divided among all programmers. DMC tackled more advanced topics, such as developing a custom TCP message handler between PC and PXI, while the client moved forward, modifying subVIs to work with a queue-based state machine. On other pieces, DMC worked alongside client staff, forming libraries and cleaning up previously written LabVIEW code. DMC and client developers met nearly every week to check on status, merge code, and test on the PXI station. DMC provided design, best-practice guidance, and inside NI know-how throughout the entire project. DMC also introduced good software project management skills, such as scheduling demos, setting release timelines and issue tracking as well as ramping up team on source code control.

Once integration and full debug were complete, DMC rolled both applications into stand-alone executables. DMC taught the client development staff best practices for maintaining and reusing source code internally for future projects. The client retained all libraries and source code to continue development as they see fit.

The post LabVIEW Consulting and Training for Internal Development Team appeared first on DMC, Inc..

High-Speed Pharmaceutical Vision Inspection

Sam Levin — Mon, 06 Jan 2014 00:00:00 +0000

DMC’s system utilized NI hardware to integrate with the existing Delta Tau Motion Controller. The NI cRIO was used to determine pill presence, track pill results from the EVS, and output results to the Delta Tau Motion Controller in order to reject bad tablets. The NI EVS triggers the cameras and lights, acquires images, runs inspection algorithms, and outputs a result to the cRIO. Each camera views two tablets, so pass-fail results are determined for 16 tablets nearly simultaneously. The first step of the inspection algorithm is to separate the pills from the background, then each individual pill is analyzed using three algorithms:

Check for bright spots within the interior of the tablet (the outside of the tablet is darker than the powdery inside of a broken tablet).
Compare the perimeter of the tablet to the ideal perimeter of a known, good tablet. The perimeter comparison allows for rotated and flipped tablets. This check will reject broken tablets that don’t have any white powder visible to the camera.
Check the gradient of pixel brightness a few pixels inside the perimeter of the tablet. This check finds hairline cracks that are not large enough to be detected as a contiguous blob of brightness.

The results are tracked in the cRIO until the pills reach the rejection point on the machine, at which point the results are sent to the Delta Tau controller and the bad pills are blown into a reject container.

Undamaged Pill

The post High-Speed Pharmaceutical Vision Inspection appeared first on DMC, Inc..