WORD-BASED DOCUMENT IMAGE COMPRESSION

• US 20110182513A1
• Filed: 01/26/2010
• Published: 07/28/2011
• Est. Priority Date: 01/26/2010
• Status: Active Grant

• Current Assignee: Hewlett-Packard Development Company, L.P. (HP Inc.)
• Original Assignee: Hewlett-Packard Development Company, L.P. (HP Inc.)
• Inventors: Forman, George, Reddy, Prakash, Eshghi, Kave
• Granted Patent: US 8,463,041 B2
• Time in Patent Office: Days
• US Class Current: 382/177
• CPC Class Codes: G06V 30/10 Character recognition
• CPC Class Codes: G06V 30/18143 Extracting features based on salient regional features, e.g. scale invariant feature transform [SIFT] keypoints
• CPC Class Codes: G06V 30/262 using context analysis, e.g. lexical, syntactic or semantic context
• CPC Class Codes: G06V 30/40 Document-oriented image-based pattern recognition
• CPC Class Codes: H04N 19/94 Vector quantisation

Abstract

Locations of word images corresponding to words in a document image are ascertained. The word images are grouped into clusters. For each of multiple of the clusters, a respective compressed word image cluster is determined based on a joint compression of respective ones of the word images that are grouped into the cluster. The positions of the word images in the document image are associated with the respective ones of the compressed word image clusters corresponding to the clusters respectively containing the word images.

Claims

1. A method, comprising:

by a computer ascertaining locations of word images corresponding to words in a document image;

by the computer grouping the word images into clusters;

for each of multiple of the clusters, determining by the computer a respective compressed word image cluster based on a joint compression of respective ones of the word images that are grouped into the cluster; and

by the computer associating the positions of the word images in the document image with the respective ones of the compressed word image clusters corresponding to the clusters respectively containing the word images.

2. The method of claim 1, wherein the grouping comprises:

for each of multiple of the word images, determining respective keypoints in the word images, ascertaining descriptor vectors describing local image content around the respective keypoints, and clustering the word images based on the descriptor vectors.

3. The method of claim 1, wherein the grouping comprises generating a respective set of hashes for each of multiple of the word images and clustering the word images based on comparisons of the sets of hashes.

4. The method of claim 3, wherein the generating comprises generating the respective set of hashes based on hashes of data vectors that comprise one or more respective descriptor vectors describing local image content around respective keypoints in the word images.

5. The method of claim 1, wherein the determining comprises:

for each of multiple of the clusters, compressing successive ones of the word images to produce the respective compressed word image cluster.

6. The method of claim 5, wherein the compressing comprises encoding differences between the successive word images in producing the respective compressed word image cluster.

7. The method of claim 5, wherein the compressing comprises applying a video compression process to the successive word images in producing the compressed word image cluster.

8. The method of claim 5, wherein the compressing comprises applying a run length encoding process to a bit stream derived from the successive word images in producing the compressed word image cluster.

9. The method of claim 1, wherein the associating comprises:

for each of the positions of the word images in the document image, associating the position with a respective one of the word images respectively contained in the compressed word image cluster associated with the position.

10. The method of claim 1, further comprising transmitting ones of the compressed word image clusters and a respective description of the associated positions of the corresponding ones of the word images in the document image.

11. The method of claim 10, further comprising:

in response to separate requests received from the remote network node for respective sets of the compressed word image clusters corresponding to different respective portions of the document image, the transmitting comprises transmitting the respective ones of the determined sets of compressed word image clusters to the remote network node in separate responses.

12. The method of claim 10, wherein the transmitting of the respective description is performed in response to a first request received from the remote network node, and the transmitting of the ones of the compressed word image clusters is performed in response to a second request received from the remote network node after the first request.

13. The method of claim 10, further comprising determining a set of the compressed word clusters based on which of the compressed word clusters previously were sent to the remote network node and the ones of the compressed word clusters associated with respective positions in a document image requested by the remote network node.

14. A method, comprising:

by a computer receiving a set of compressed word image clusters each comprising a joint compression of a respective cluster of word images;

by the computer receiving associations between respective ones of the word images in the respective joint compressions of the compressed word image clusters and positions of respective ones of the word images in a document image;

by the computer extracting the word images from respective ones of the compressed word image clusters;

rendering a version of the document image based on the extracted word images and the associations between the compressed word image clusters and the positions of the word images in the document image.

15. The method of claim 14, wherein the extracting comprises applying a video decompression process to the compressed word image cluster in producing the successive word images.

16. The method of claim 14, wherein the extracting comprises applying a run length decoding proces'"'"'s to the compressed word image cluster to produce a bit stream, and deriving the successive word images from the bit stream.

17. The method of claim 14, wherein the associations comprise, for each of the positions of the word images in the document image, associations between the position and a respective one of the word images respectively contained in the compressed word image cluster associated with the position;

and the rendering of the version of the document image is based on the associations between the positions of the word images in the document image and the respective word images respectively contained in the compressed word image clusters associated with the positions.

18. The method of claim 14, wherein the rendering comprises rendering a version of the document image on a display.

19. Apparatus, comprising:

a computer-readable medium storing computer-readable instructions; and

a data processor coupled to the computer-readable medium, operable to execute the instructions, and based at least in part on the execution of the instructions operable to perform operations comprising

ascertaining locations of word images corresponding to words in a document image;

grouping the word images into clusters;

for each of multiple of the clusters, determining a respective compressed word image cluster based on a joint compression of respective ones of the word images that are grouped into the cluster; and

associating the positions of the word images in the document image with the respective ones of the compressed word image clusters corresponding to the clusters respectively containing the word images.

20. At least one computer-readable medium having computer-readable program code embodied therein, the computer-readable program code adapted to be executed by a computer to implement a method comprising:

ascertaining locations of word images corresponding to words in a document image;

grouping the word images into clusters;

for each of multiple of the clusters, determining a respective compressed word image cluster based on a joint compression of respective ones of the word images that are grouped into the cluster; and

associating the positions of the word images in the document image with the respective ones of the compressed word image clusters corresponding to the clusters respectively containing the word images.