HALCON 13 - Preview

The new major version of MVTec's standard machine vision software, HALCON 13, will be released on November 1, 2016. It introduces some noteworthy innovations which we would like to acquaint you with.

Algorithm Improvements

Algorithm speed-ups have advanced consistently - achieving an average 14% improvement on all tools as compared with HALCON 12. Two outstanding examples of this are shape-based matching and 3D reconstruction with speed-ups of up to 3 and 4 times respectively.

Additionally, many functions of the algorithm have been added to or improved. Two new edge-preserving noise filters offer significant performance enhancement and can also be used for image stacking as in HDR applications.

The MSER algorithm, which has proven to be more robust than threshold and region-growing methods for segmentation, has been implemented in order to simplify the segmentation of sharp-bordered foreground objects against contrasting backgrounds.

Developments for 3D Applications

High quality 3D reconstructions can now be created using the combined image data from a calibrated multi-view camera system. In comparison with previous methods, outliers and mismatched point correspondences are reduced and multiple correspondences are adjusted to create a closed surface. With additional image data from a color camera, the surface can be texture mapped in order to produce a realistic 3D rendering. The resulting image data can also be directly uploaded to a 3D printer.

Surface-based matching has been expanded with 3D edge detection. This enables considerably more precise surface-based matching with fewer false matches and is especially useful for the matching of objects with large uniform surfaces and little surface detail.

Additionally, a new 3D point filter has been included which is particularly effective in eliminating outliers in edge areas.

Simplified and Enhanced Classification Processes

In 2D imaging, a new significantly simplified texture classification process for the training of surface features will be available. The new texture classification can return classified areas directly as regions.

For better control, Barcodes/2D code recognition can be taken from other threads. The QR-code reader features a more robust handling of distorted and low-contrast codes.

For text recognition there will be a new OCR classifier which is enhanced by deep-learning classification algorithms. HALCON 13 achieves recognition rates of up to 3 times higher than those from HALCON 12 without sacrificing the incredibly fast reading rates.

More Debugging Flexibility and Simplified Rendering

Soon, it will be possible to debug the HdevEngine directly in Visual Studio. This can also be carried out with multithread applications. The connection between application and script is accessed via an IP socket, thereby making it irrelevant whether the application is run locally or on another computer in the network.

Finally, the visualization function has been improved for simplified rendering. Text and other overlays can now be inserted as transparent or non-transparent and displayed flicker-free in live-mode. The image area can also be zoomed into without loosing the overlay.

Test HALCON 13 for Free

HALCON 13 will be available for purchase as either a new license or as upgrade starting November 1, 2016. You can also test HALCON 13 at no charge. Please contact our sales department if you are interested in an evaluation license.