HALCON 13 major release for MVTec

For one thing, the new version provides significantly increased performance for shape-based matching. In addition, damaged, occluded, or incomplete bar codes, as well as deformed data codes can now be recognised and read more robustly.

The HALCON 13 also features a particularly easy-to-use texture inspection. The new release also offers an enhanced, deep learning based Optical Character Recognition (OCR), as well as the automatic, robust identification of dot print fonts. With HALCON 13, which is to be released on 1^st November, MVTec continues to promote standardisation in the area of machine vision as well as M2M communication.

Significant speed improvement of shape-based matching

Shape-based matching is one of the most important basic technologies for the identification and 2D position determination of objects. HALCON 13 helps speed up this processing step on machines with AVX2-compatible processors by as much as 300%. The software also significantly increases the speed of all related technologies, such as shape-based 3D matching, local and perspective deformable matching, as well as component-based matching.

Texture inspection made easy

Texture inspection is another challenging machine vision task. Structured objects such as textiles, leather, and carpets frequently have different characteristics regarding patterns or brightness. HALCON 13 offers an easy-to-use texture inspection, in which potential texture defects can be automatically identified by using an image-based comparison with flawless materials.

This new feature allows the classification of a wide range of different textures with only a few parameters, making surface inspections of corresponding textures much easier.

3D matching enhancements for more accurate position determination

HALCON 13 also provides a further improved surface-based 3D matching. By analysing the edges in the model and the 3D point cloud, recognition especially of flat surfaces is now more robust, which allows the position of objects to be determined more precisely. This new method is particularly suitable for finding the exact position of boxes in the 3D space.

The announced software release also includes a new technology for the multi-view reconstruction of 3D objects with surface fusion. Using the image information from multiple cameras produces more robust results than provided by pairwise reconstruction methods. During this process, HALCON transfers available colour information of the input images to the reconstructed objects, which improves visualisation and further processing (e.g., with a 3D printer).

Lastly, HALCON 13 also offers an improved OCR. By implementing new, deep learning based pre-trained fonts and the corresponding classifier, the character recognition rate is greatly improved.

User-driven development adds value

"We are delighted to have addressed the many challenges that our customers face in machine vision tasks even better with this new release. Our software development is largely driven by user feedback. As a result, HALCON 13 is guaranteed to deliver considerable added value to our customers’ imaging applications," explained Johannes Hiltner, Product Manager HALCON at MVTec.