Method and arrangement for recognizing objects in mail item images, their position and reading their postal information

• US 7,480,394 B2
• Filed: 04/25/2005
• Issued: 01/20/2009
• Est. Priority Date: 04/14/2005
• Status: Active Grant

• Current Assignee: Korber Supply Chain Logistics GmbH
• Original Assignee: Siemens AG
• Inventors: Berlin, Bernhard, Delianski, Svetlozar, Kinnemann, Georg
• Primary Examiner(s): Tabatabai; Abolfazl
• Application Number:
  US11/113,315
• Publication Number:
  US 20080137907A1
• Time in Patent Office: 1,366 Days
• Field of Search: 382/100, 382/101, 382/181, 193/8, 229/92, 229/300, 229/301, 229/302, 229/305, 229/306, 229/921, 705/60, 705/401
• US Class Current: 382/101
• CPC Class Codes: B07C 3/14 using light-responsive detecting means for information carriers G06K7/14
• CPC Class Codes: G06V 10/147 Details of sensors, e.g. sensor lenses fingerprint or palmprint sensors G06V40/13; vascular sensors G06V40/145; eye sensors G06V40/19

Abstract

The invention relates to a method for recognizing objects in mail items and an arrangement for carrying out the method. In the method, the color image produced in the relevant color camera is converted into a gray level image having a resolution higher than that of the color camera and suitable for reading the postal information in the automatic reading devices; the color image produced in the relevant color camera is converted into a color image having a resolution lower than the camera resolution but sufficient far recognizing objects and their positions; the objects and their position present in the relevant lower resolution color image are determined and identified, the assignment of the objects to postal information categories being performed on the basis of structural rules established in an upstream teaching phase, and the required postal information in the gray level images of the relevant identified objects are automatically read.

Claims

1. A method for recognizing objects and their position as well as reading their postal information in color and gray level images of mail item surfaces using at least one automatic reading device, comprising the steps of:

capturing the mail item surface using at least one color camera,

converting the color image produced in the camera into a gray level image having a resolution higher than that of the camera and suitable for reading the postal information in the automatic reading devices,

converting the color image produced in the camera into a color image having a resolution lower than the camera resolution but sufficient for recognizing objects and their positions,

determining and identifying the objects in the relevant lower resolution color image including their position, the assignment of the objects to postal information categories being performed on the basis of structural rules established in an upstream teaching phase, and

reading the postal information in the gray level images of the relevant identified objects.

2. The method according to claim 1, wherein the automatic reading devices include a plurality of specialized OCR units assigned to defined objects, the OCR units arranged such that the postal information of the identified and defined objects in the associated gray level images are read by the OCR units.

3. The method according to claim 1, wherein the gray level images of the entire mail item surface are transmitted to the OCR units along with the object data and reading only takes place in the gray level images of the relevant objects.

4. The method according to claim 1, wherein the object data together with the gray level images of these objects only are transmitted to the OCR units.

5. The method according to claim 1, wherein the image data of the gray level image and low resolution color image of a mail item surface is transmitted synchronously to the reading device.

6. The method according to claim 1, wherein for the position of the objects, their orientation is also determined and the position and orientation data of a plurality of mail item sides is used to control an alignment device which brings the mail items into a uniform position.

7. An arrangement for recognizing objects and their position as well as reading their postal information in color and gray level images of mail item surfaces using at least one automatic reading device, comprising:

means for capturing the mail item surface using at least one color camera,

means for converting the color image produced in the camera into a gray level image having a resolution higher than that of the camera and suitable for reading the postal information in the automatic reading devices,

means for converting the color image produced in the camera into a color image having a resolution lower than the camera resolution but sufficient for recognizing objects and their positions,

means for determining and identifying the objects in the relevant lower resolution color image including their position, the assignment of the objects to postal information categories being performed on the basis of structural rules established in an upstream teaching phase, and

means for reading the postal information in the gray level images of the relevant identified objects.

8. The arrangement according to claim 7, wherein the means for capturing, means for converting and means for determining are executed in the processor of the color scanner.

9. The arrangement according to claim 7, wherein the means for determining and identifying the objects and the means for reading of the postal information takes place in at least one computer unit to which the image data of the gray level image and low resolution color image is synchronously transmitted via frame grabbers.

10. The arrangement according to claim 7, further comprising a plurality of OCR units assigned to defined objects and arranged to receive images of the assigned objects from at least one object identification arrangement which is connected to the color scanner.

11. The arrangement according to claim 7, wherein the color camera comprises an RGB CCD line sensor past which the mail items are transported.