Classifications

• • B—PERFORMING OPERATIONS; TRANSPORTING
  • B65—CONVEYING; PACKING; STORING; HANDLING THIN OR FILAMENTARY MATERIAL
  • B65C—LABELLING OR TAGGING MACHINES, APPARATUS, OR PROCESSES
  • B65C9/00—Details of labelling machines or apparatus
  • B65C9/06—Devices for presenting articles in predetermined attitude or position at labelling station
  • B65C9/067—Devices for presenting articles in predetermined attitude or position at labelling station for orienting articles having irregularities, e.g. holes, spots or markings, e.g. labels or imprints, the irregularities or markings being detected

Definitions

• the present invention relates to a method and apparatus for aligning a container. Furthermore, the present invention relates to a computer program for carrying out the method according to the invention.
• the published patent application DE 10 2005 050 902 A1 shows a device for aligning containers with respect to at least one geometric container feature comprising a conveyor with container receptacles for a respective container and along a transport path formed by the conveyor cameras of a Jardinkennungs- system comprises, wherein the image recognition system by aligning the supplied by the cameras actual image data with stored in an evaluation and control electronics SoIl image data or characteristics causes the alignment of the container.
• the containers are aligned by means of a geometric feature, for example by means of an embossing on the surface of the container. From this geometric feature, a picture is taken in the desired position of the bottle according to the prior art and then calculates a respective distortion of this geometric feature for the cases in which the container is rotated in each case by a certain angle from this desired position , The images or image data thus generated are then stored and used as SoIl image data.
• an image of the container or of the bottle is then taken by means of a camera and the thus generated actual image is compared with the stored target image data by means of an algorithm. It is determined with which image of the target image data the highest match exists. Based on the rotation of the container or the bottle deposited for this image in the target image data, it can then be rotated by the corresponding angle into the desired position.
• a plurality of stages of such alignment operations may be provided to successively bring about the desired orientation or position of the container or bottle.
• a plurality of cameras record an image of the bottle at the same time, which is then compared with the target image data, the alignment process also being carried out based on the best match.
• the invention proposes a method for aligning a container with respect to at least one geometric container feature in a desired position, in which an alignment of the container is effected by an evaluation of generated by at least one image pickup device actual image data in a control electronics, wherein the Actual image data in a search area are evaluated by means of a pattern recognition, which as a result outputs an actual position of the container, if the geometric container feature is recognized, and indicates if the geometric container feature is not recognized.
• a neural network is used, which has been taught beforehand in a learning procedure, how the geometric container feature used to align a container looks in the various orientations of the container, and the - A -
• an alignment process controlling control electronics is able to classify any other geometric container features or images in the search area to the effect that they do not represent the search instance by these so-called exclude instances are taught as such in advance.
• control electronics in the context of the pattern identifier is able to reject certain recordings of the search area from the outset, since they do not contain the search instance.
• pattern recognition takes place in that the control electronics classifies the actual image data on the basis of target image data as to whether they contain the geometric container feature.
• the target image data in this case comprise the learned geometric container feature as a search instance and furthermore the exclude instances described above and likewise trained.
• the teaching of the desired image data is carried out by firstly searching for the sample image bottle.
• the geometric container feature, the search instance is recorded and stored.
• the geometric container feature is then identified by a user, in particular based on its contour.
• This training of the search authority can basically at the site of the process, d. H. with the actually constructed filling and labeling, or off it, for example, in a laboratory.
• Bottle rotated in front of an image pickup device on itself, wherein in certain steps, respective images of the container or the bottle are made.
• the multitude of degree steps depends in particular on how exactly the alignment should be performed during operation. If, for example, an alignment is to be done with a precision of 1 °, it is indicated that the images are also displayed in 1 ° increments, better in 0.5 °.
• the search instance is now cut out, if it is still found in the searched area, which can happen automatically, since the search instance has been trained before and the neural network is known.
• the actual image data recorded by the image recording device are compared with the instances stored in the target image data, and it is determined whether the search region contains an exclude entity or the search entity.
• search area contains only one exclude instance, this is displayed and the corresponding image is not used to calculate the actual position of the bottle.
• the actual position of the container or its orientation is calculated on the basis of a perspective distortion of the geometric container feature, a known distance of the image recording device to the container during the recording and a known diameter of the container - net. From the difference between the actual position of the container to the desired desired position of the container, the degree and the direction in which the container is to be rotated can now be determined.
• the invention further includes an apparatus for carrying out a method according to the invention.
• the present description also covers a computer program with program code in order to carry out all the steps of a method according to the invention when the program is executed on a computer or a corresponding computing unit.
• the computer program may be stored on a computer-readable medium.
• FIG. 1 shows a schematic view of an embodiment of a device according to the invention, on which the method according to the invention can be carried out.
• FIG. 1 shows a device 1 on which the method according to the invention can be carried out.
• the device comprises a carousel 5, in the
• Container 2 transported via an inlet 3 and transported beyond a spout 4. From the labeling station 7, the container 2 in the desired
• the first stage comprises the image pickup devices 8 and 9, the second stage the image pickup device 10 and the third stage the image pickup device 11.
• background elements 13, 14, 15 is provided, which can also be illuminated.
• the alignment operation is controlled by a control electronics 12, which receives actual image data from the image sensing devices 8, 9, 10, 11 and can output control commands to container receptacles 6 on which the containers 2 are arranged.
• a learning process For commissioning and for the preparation of the alignment process, a learning process has to be carried out, in which first a search instance, ie a geometrical container feature used to align a container, is determined from a first image of a sample container identified a user and the search instance is so a neural network of the control electronics: 12 taught.
• the corresponding target image data set can be called up by an operator. This is followed by a slow inlet of a first container or a first bottle, wherein the image recording devices 8, 9, 10, 11 are aligned so that they detect the possible positions of the search instance as accurately as possible.
• a narrow optical evaluation range is a narrow optical evaluation range.
• the device according to the invention and the computer program according to the invention a simplified and thus accelerated and qualitatively improved alignment of containers, in particular bottles, can be provided.

Priority Applications (4)

Applications Claiming Priority (2)

Publications (1)

Family

ID=41682749

Family Applications (1)

Country Status (6)

Cited By (1)

Families Citing this family (4)

Citations (7)

Family Cites Families (2)

• 2008
  • 2008-11-20 DE DE102008059229A patent/DE102008059229A1/de not_active Withdrawn
• 2009
  • 2009-11-11 CN CN200980146264.5A patent/CN102216161B/zh active Active
  • 2009-11-11 EP EP09763831A patent/EP2358601B1/de active Active
  • 2009-11-11 WO PCT/EP2009/008025 patent/WO2010057592A1/de active Application Filing
  • 2009-11-11 PL PL09763831T patent/PL2358601T3/pl unknown
  • 2009-11-11 BR BRPI0916040A patent/BRPI0916040B1/pt active IP Right Grant

Patent Citations (8)

Also Published As

Similar Documents

Legal Events