Image processing for clustering related text objects

• US 6,975,761 B1
• Filed: 11/28/2000
• Issued: 12/13/2005
• Est. Priority Date: 06/30/2000
• Status: Expired due to Fees

• Current Assignee: Consignia Plc (International Distributions Services plc)
• Original Assignee: Consignia Plc (International Distributions Services plc)
• Inventors: Tan, Robert Wei Liang, Swann, Robert Edward Meredith
• Primary Examiner(s): Johns, Andrew W.
• Assistant Examiner(s): Alavi, Amir
• Application Number:
  US09/724,182
• Time in Patent Office: 1,841 Days
• Field of Search: 382/101, 382/102, 382/173, 382/176, 382/181, 382/190, 382/218, 382/224, 382/225, 382/227, 382/162
• US Class Current: 382/173
• CPC Class Codes: G06V 30/10 Character recognition
• CPC Class Codes: G06V 30/147 Determination of region of interest
• CPC Class Codes: G06V 30/414 Extracting the geometrical structure, e.g. layout tree; Block segmentation, e.g. bounding boxes for graphics or text

Abstract

An image processor for producing clusters of related objects in a multi-level digital representation of an image includes a part for identifying predetermined objects in the image and for supplying data defining their locations. It also includes a part for deriving segmentation data directly from the multi-level digital representation. A segmentation device receives both sets of data and clusters the predetermined objects into groups for each segmented region before providing clustering data for subsequent analysis.

Claims

1. An image processing system for producing clusters of related objects for subsequent analysis comprising, means for supplying a multi-level digital representation of the image, means for identifying predetermined objects in the image and supplying data defining their locations, means for deriving segmentation data from the multi-level digital representation wherein the means for deriving the segmentation data from the multi-level representation comprises means for comparing differences in background data to derive the segmentation data, means for combining the data defining the locations of objects with said segmentation data, means for clustering the predetermined objects into groups for each segmented region and means for supplying data relating to the groups for subsequent analysis.

2. An image processing system according to claim 1 in which the objects are text objects.

3. An image processing system according to claim 1 in which the background data comprises colour data.

4. An image processing system according to claim 1 in which the background data comprises greyscale data.

5. An image processing system for producing clusters of related objects for subsequent analysis comprising, means for supplying a multi-level digital representation of the image, means for identifying predetermined objects in the image and supplying data defining their locations, first segmentation means for deriving segmentation data from the multi-level digital representation, second segmentation means receiving the multi-level digital representation of the image and the data identifying the locations of the objects in the image for deriving segmentation data relating to differences in the data of the multi-level digital representation from the said objects, means for combining the data defining the locations of objects with said segmentation data, wherein the combining means receives the segmentation data and further segments the data defining locations of the object image in dependence on this segmentation data, and means for clustering the predetermined objects into groups for each segmented region and means supplying data relating to the groups for subsequent analysis.

6. An image processing system according to claim 5, in which the second segmentation means derives the segmentation data thereof from the colour of the objects.

7. An image processing system according to claim 5, in which the second segmentation means derives the segmentation data thereof from the greyscale level of the objects.

8. A method for processing images to produce clusters of related objects for subsequent analysis comprising the steps of supplying a multi-level digital representation of the image, identifying predetermined objects in the image, supplying data defining the locations of these predetermined objects, deriving segmentation data for the image from the multi-level digital representation which step further comprises comparing differences in background data to determine the segmentation data, combining the data defining the locations of the objects with the segmentation data, clustering the objects into groups for each segmented region, and supplying data relating to the groups for subsequent analysis.

9. A method according to claim 8, in which the objects are text objects.

10. A method according to claim 8 in which the background data comprises colour data.

11. A method according to claim 8 in which the background data comprises greyscale data.

12. A method for processing images to produce clusters of related objects for subsequent analysis comprising the steps of supplying a multi-level digital representation of the image, identifying predetermined objects in the image, supplying data defining the locations of these predetermined objects, deriving segmentation data for the image from the multi-level digital representation, receiving the multi-level digital representation of the image and the data identifying the location of the objects in the image and deriving additional segmentation data relating to differences in the data of the multi-level representation for said objects, combining the data defining the locations of the objects with the segmentation data, wherein the combining step further segments the data defining the locations of objects in dependence on the additional segmentation data and clustering the objects into groups for each segmented region, and supplying data relating to the groups for subsequent analysis.