NI Vision Tools Meet the Challenges of Real-Time Inspection

Author(s):
Patrick Lessnick - Averna Technologies, Inc.

Industry:
Machine Vision/Imaging, Automotive

Products:
Machine Vision, LabVIEW, PCI-1422

The Challenge:
Detecting 100 micrometer-size cracks on both sides of a fast-moving cast aluminum slab with a surface temperature of 400˚C and marking the defective slabs in real time.

The Solution:
Integrating image acquisition and process-control hardware with an efficient, reliable, PC-based application built with National Instruments LabVIEW and an NI image processing library to deliver a cost-effective, rapid-developing, reliable vision inspection system.

Implementing a Visual Inspection System

Recently, in a very demanding application, we used NI LabVIEW and image acquisition and processing modules combined with fast frame grabbers to implement a high-throughput real-time visual inspection system. Our customer needed to monitor and improve their production quality against surface cracks when casting a special-purpose aluminum alloy. The alloy is cast in the shape of slabs with a 2.8 m length and variable widths (0.3 m to 0.5 m) and heights (10 mm to 40 mm). The slabs move on a conveyer at a speed of 20 m/minute.

The inspection requirements included:

• Inspect both sides of the slab for cracks as small as 100 micrometers in width and a few mm in length, and mark the side of the defective slabs in real time.
• Detect and reject slabs with pinhole clusters.
• Alarm the operator about the crack occurrence, show and store images of the last cracks, and indicate cast process quality in real time.

Based on system performance and requirements, we used two line-scan cameras with 8,000 pixels each and a line acquisition rate of 5 kHz on either side of the slab. We used spatially randomized light lines to illuminate the slabs with high-intensity white light. A speed-encoder wheel and a pair of limit switches generated the TTL trigger envelope that was routed to the camera through an NI PCI-1422 frame grabber. The images were analyzed by dual 2.8 GHz XEON workstations running optimized software built with NI LabVIEW and image processing library. If a surface defect was detected, the PC actuated a rejection system to mark the slab on the side.

We needed a camera capable of detecting 100 m cracks (comparable to the width of a human hair) on the quickly moving slab. Because we needed to process a large amount of data in real time, we needed a cost-efficient and optimal design for both data transfer and image-processing algorithm. We met this challenge by running the application on commercially available desktop servers with dual 2.8GHz XEON architectures and building a CPU-efficient image acquisition and processing application using LabVIEW. The software performed the following periodic steps:

• Image acquisition and transfer
• Threshold and conversion of an 8-bit image to a binary image
• Particle closure to recover any missing pixels
• Particle analysis and detection of cracks exceeding the minimum size
• Pinhole detection
• Defect marking system actuation

We reduced the application development time by using interactive image processing and automatic code generation features of NI IMAQ Vision builder module. We further reduced the image acquisition and processing time by using the PCI-1422 frame grabber, which provided fast data transfer, triggering, and onboard image processing capabilities. A detailed system timing analysis and bench marking approach ensured that no bottlenecks were created in real time performance of the system.

Integration Challenge

One challenge that we had to consider was the slabs’ 400C temperature. The normally delicate vision inspection system needed to operate in a harsh casting process line environment. We housed all cameras and illumination components in NEMA enclosures with forced cooling and protection against environmental and mechanical hazards. We enclosed the light sources and computing hardware in a cooled cabinet and operated the system through an industrial touch screen front panel. Efforts of a multidisciplinary team of mechanical, electrical and software engineers were coordinated to design and implement the system in a very short duration of five months.

We developed a semiautomatic procedure to calibrate the system for light intensity, camera height and focus. The vision system acquired and analyzed reference pattern images and provided the operator guidance for adjustments as needed. Once calibrated, the system operated through a simple graphical user interface that provided a real-time process tuning feedback while showing the surface defects and their occurrence statistics. The system stored crack images on disk for future analysis.

Intelligent Algorithms

The reflecting properties of slab surfaces change depending on slab temperature, alloy composition, conveyor “cleanliness” and other metallurgic process parameters, making crack detection even more challenging. Intelligent algorithms were developed to take these variations into account, which made the system easier to operate for the end-user,  as well as more robust. The system also becomes independent of the light source intensity stability over time.

Creating a Cost-Effective System

We implemented a challenging and leading-edge inspection system using flexible and reliable NI vision components, which proved to be the determining factor in making this system cost effective and dramatically quick to deploy. We achieved seamless component integration and efficiency while focusing the engineering efforts on meeting the challenges of a high-throughput, real-time system operating in a harsh environment. From conception to deployment, we developed the system in a short period of five months, and saved more than $200,000.

Shahzad Sarwar  was also a co-author on this paper. 

Author Information:
For more information on this Case Study, contact:
Patrick Lessnick
Averna Technologies, Inc.
87 Prince - Suite #140
Montreal H3C 2M7
Canada
Tel: (514) 842-7577
Fax: (514) 842-7573
info@averna.com

Browse All Case Studies »