The new version 11 was released by Halcon in June 2012, with which objects can be identified without printed additions such as bar codes and data codes. In addition, Version 11 shows what the software is capable of in the field of 3D image processing.

Industrial products are subject to stringent quality criteria. Because of this it must be ensured that the shape of manufactured objects complies precisely with the specifications, even in the production process. In Halcon 11 comparison of the specified and actual shapes of three-dimensional objects is now possible without contact. For this, the model, i.e. the sample of the object must be known to the system. This can for example take the form of a CAD model, which is read in directly. An ideal sample can also be measured and generated in three dimensions with Halcon. For the actual measurement of the object which is to be examined, a 3D cloud of points can be generated with a laser sensor. This image is then compared with the model. On the basis of this data, a statement can be made as to whether the surface of the object complies with that of the model. The form of the object is therefore checked for errors.
3D object processing provides many improvements and innovations in the field of 3D vision. A common method of recording three-dimensional objects is to generate clouds of points with an appropriate sensor. However, as certain objects cannot be entirely depicted with a single image, several images from different directions are necessary. The corresponding points are then found with the subsequent matching. By registering the pint clouds, the individual images can be transferred to a uniform 3D model. However, it is only the interlinking of the point clouds which enables the creation of a spatial relationship between the points. With Halcon 11, this interlinking (triangulation) provides information about may characteristics of an object, so that characteristics of a 3D object such as shape and volume can be calculated.
If an object is recorded via one or more point clouds, a background is always recorded. However, in most cases, this background causes interference and is therefore not wanted.
With Halcon 11, such a background can be quickly and simply eliminated by calculation, whereby the software intersects the model with an imaginary plane. This also enables the object to be isolated from adjacent objects. For example, with this method a cross-section of an object can be generated. Sections of objects can not only be made with planes, but also with regions. The user can therefore separate an entire area from a 3D object model.
An important field of application for industrial image processing is surface inspection. Problems always occur, if there are unwanted indentations and other recesses in the surface which are difficult to identify. A method with a three-dimensional approach can offer a solution to this. The photometric stereo which has been further developed in Version 11 provides a remedy. 3D surface inspection detects such recesses in surfaces by greater observation of shadows. This enables faults to be quickly and simply discovered. For this, several images are made with different directions of illumination. This method can also be used for difficult textures, such as leather.

Identification without printing

Identification is one of the primary tasks of industrial image processing. Two-dimensional data codes have long-since become established in the field of dispatch and transport. 2D codes have also entered the daily life of smartphone users. Would it not be nice if a machine could identify an object without the need for such a code?
Halcon 11 now provides a new technology which is based on the direct recognition of a sample: sample-based identification. This new technology is able to identify a certain product or object which is known to the system, from a pile of different objects. This functions via the specific characteristics of the object such as colour or texture, and not via a printed code such as a bar or data code. This is therefore non-printed identification of a known object. For example, with sample-based identification, a certain product can be reliably identified from a series of other products.

Vision 2012 Halle 1D74