Electronics company Omron launched its new Shape Search III image- processing technology last year, which is ten times faster than its predecessors, says Omron vision manager Philip Potgieter.
He explains that the Shape Search III software is used on the FH/FZ5 series controller, which is the company’s flagship unit. “It is a fast industrial personal computer, which is dedicated and made only for, Omron’s high-speed Vision inspection applications, with the new industry Ethercat communication as a standard,” he adds.
Ultrafast new complementary metal oxide semiconductor and charge-coupled device cameras from Omron, together with the FH or FZ5 controller inspect the exterior features of packaging, and printed information on labels or packaging. This correctness and quality of print play a central role in the process. These are the first elements that consumers see and are often a decisive factor in their decision to buy, Potgieter says.
The software uses an intelligent approach that links the factors influencing the image, such as backgrounds or imperfections, and presents them to a user even if the scanned objects or parts are not clearly defined.
Potgieter explains that high consumer expectations regarding information on food products, pharmaceutical and nonfood products, as well as new legal requirements, have resulted in producers increasingly requiring inspection solutions that check whether information has been applied correctly, in line with industry standards.
In addition to final inspections, these producers are also increasingly using in-line inspections at critical points in the corresponding processing steps, he adds.
Increasing cost pressures in production means that faster processing speeds are being used to keep up with consumer demands. For many inspection systems, this represents a major challenge in terms of precisely detecting objects. The necessary computing power for inspection units is often high and is, therefore, compensated for by a reduction in processing speeds, says Potgieter.
Compensated processing speeds have been mitigated through incorporating new image processing algorithms such as edge-based sparse features and variation-absorbing templates, as well as ultrafast and parallel hardware architecture.
“Omron’s FH optical inspection system not only achieves detection speeds faster than conventional inspection systems, but also increases detection quality, owing to the new algorithms,” highlights Potgieter.
Since the 1980s, there has been a breakthrough in image processing algorithms for object detection in each decade, he says. In the 1980s, binary-image-based algorithms enabled relatively rapid object detection. This rapid algorithm was adapted to the very low computing power available, but factors such as noise, lighting changes, shadowing, and low contrasts affected results.
By the 1990s, hardware speeds were increasing rapidly, allowing for a more accurate analysis of the image gray-scale value while reducing the number of problems encountered at low contrasts.
In the 2000s, edge-based algorithms resulted in improvements on lighting changes and shadowing, though these algorithms still had disadvantages in terms of blurring and low contrasts. The new, sparse edge detection algorithm takes the information that is used and reduces it to clearly identifiable and representative points. This eliminates the possibility of errors while also achieving significant improvements in speed.
Potgieter points out that, in conventional inspection systems, minor deviations in the position of objects, owing to a vibrating conveyor belt, can inhibit error-free image formation or the rapid processing of this information.
“In the past, inspection software compensated for errors, which significantly reduced computing power, therefore, reducing processing speeds. Often, a compromise had to be reached between reliability and speed,” he adds.
The new variation-absorbing method predicts possible variations in the representative points of the tracked objects. These variations are summarised using an intelligent clustering process, notes Potgieter.
An analysis of these clusters reduces detection errors, while the processing speed remains high, owing to the low memory use of the unit. This ensures that high-speed image processing can be completed with ten times the level of precision.
The criteria for achieving an object image that is as clear, stable and simple as possible to process are complex. In the past, an improvement in this original image for processing by inspection systems was often judged through trial and error or by using expertise accumulated over many years.
Potgieter concludes that the Shape Search lll technology is the culmination of all the positive aspects of inspection systems, resulting in a more efficient system.