MVTec continue the evolution of their bar code reading capabilties with significant improvement to robustness under difficult conditions, and updates to their barcode grading feature in HALCON 12
While HALCON has always provided powerful bar code reading tools, certain image “issues” occasionally interfered with accurate location and reading of bar codes. Edges that appear close to and aligned with the bar code result in the potential candidate region containing a lot of clutter. The algorithm cannot accurately discern whether the edge of the paper is part of the code or not. In HALCON 12, this is no longer a problem and the bar code is quickly located and decoded. Additionally, the search for candidate regions is now more robust in the presence of noise or distortions within the bar code.
Improvements to the bar code reader in HALCON 12 (including new and more intuitive parameter settings) have increased read rate accuracy by 35% when tested on a set of 24,000 challenging test images containing various bar code types with rotation, heavy noise, element sizes of 1 to 4 pixels, simulated defocus or a combination of all of these.
MVTec has also evolved their barcode and 2D data code grading capabilities. In HALCON 12 the grading and decoding algorithms for several types of codes have been updated to conform to the latest ISO/IEC and AIM standard editions.
Finally, the training mode of the bar code reader required a maximum of one bar code per training image. Now, the training may be used on images with multiple bar codes, as well. All bar codes contribute to the training result equally. The speed of training bar codes with large input images has also been significantly increased.