Image analysis – Pattern recognition – Template matching
Reexamination Certificate
2001-04-09
2004-02-10
Mehta, Bhavesh M. (Department: 2621)
Image analysis
Pattern recognition
Template matching
C358S515000, C358S520000, C358S521000, C358S532000, C382S164000, C382S165000, C382S173000, C382S195000, C382S225000, C382S274000
Reexamination Certificate
active
06690828
ABSTRACT:
CROSS-REFERENCE TO RELATED APPLICATIONS
None.
BACKGROUND OF THE INVENTION
This invention relates to a method for representing digital images in a manner that enables different images to be compared with each other. When digital images are stored in an image database, it is difficult to remember which images have been stored and where each one is located. When an individual adds a new image, it would be useful to be able to quickly compare the new image with the stored images and determine whether the database already includes the same image or one that is similar to it.
Similar images may arise from several different sources. For example, imaging software may be used to add text, crop, or increase the contrast of digital photographs. In addition, two images may display the same subject from a slightly different perspective or with different exposure settings. Events such as these result in images that are not identical but are similar to each other. It would be helpful to have a system that informs the user whether or not a similar image is already stored in the database, thereby saving both the user's time and storage space. Furthermore, by identifying related images, such a system assists the image management process by automatically organizing and grouping related images together.
Many image management applications allow the user to manually associate text with the images in the database, typically in a comment or description field. Users can then search for images based on this stored text (e.g., display all of the images that have a “roses” comment). Content-based image retrieval systems have been developed for identifying similar images through complex image processing techniques. These image comparison algorithms are often computationally intensive and take a considerable amount of time to return their results. The methods used in these systems frequently involve color, shape, and texture calculations, require extensive parameter adjustment by the user, and employ time-consuming data processing techniques. There is a need for a simpler and faster method for comparing digital images with each other in order to find similar images.
BRIEF SUMMARY OF THE INVENTION
A method for generating a representation of a digital image according to the present invention involves converting the original image into a square bitmap and then dividing that bitmap into a plurality of square cells. Each cell is further subdivided into a plurality of square regions. Both the region with the minimum intensity (brightness) and the region with the maximum intensity are found for each cell. The minimum intensity value, the maximum intensity value, and the relative location of the minimum intensity region to the maximum intensity region make up each cell's representation data. The method combines the representation data for each of the cells in the square bitmap in order to create a compact representation of the original digital image.
The key relationship in the method of this invention is that visually similar images generate similar cell representation data. This relatively small amount of information is used by the method to compare images and to search for similar images in an image library. The method of the present invention is less complex and more efficient than prior art techniques for representing and comparing digital images with each other. Images are accurately represented by a small amount of information and compared with each other in a relatively quick and robust manner.
REFERENCES:
patent: 5329379 (1994-07-01), Rodriguez et al.
patent: 5331442 (1994-07-01), Sorimachi
patent: 5745595 (1998-04-01), Yamada et al.
patent: 5784488 (1998-07-01), Kuwata
patent: 5926578 (1999-07-01), Fukase et al.
patent: 5982926 (1999-11-01), Kuo et al.
patent: 6061471 (2000-05-01), Coleman, Jr.
patent: 6111982 (2000-08-01), Adachi
patent: 6249603 (2001-06-01), Rucklidge
patent: 6320981 (2001-11-01), Yada
patent: 6324309 (2001-11-01), Tokuyama et al.
patent: 6343146 (2002-01-01), Tsuruoka et al.
patent: 6389169 (2002-05-01), Stark et al.
“Similarity-Based Image Segmentation,”Determination of Brain/Liquor Ratio by Alzheimer Dementia, Institut für Bildverarbeitung und angewandte Informatik e.V., Leipzig.
Colin C. Venters and Dr. Matthew Cooper; “A Review of Content-Based Image Retrieval Systems,” University of Manchester.
Christos Faloutsos; “Fast Searching of Multimedia Objects by Content,” Carnegie Mellon University, Pittsburgh, Pa.
“Computer Vision and Pattern Recognition Group,” Home Page, Universiy of Bonn/CSIII John E. Johnson, Jr.; “Microscopy Research and Technique,”Lucis.
“Attrasoft ImageFinder” Advertisement,Get Software.
P. Thévenaz, U.E. Ruttimann and M. Unser; “A Pyramid Approach to Subpixel Registration Based on Intensity,” IEEE Transactions on Image Processing, Jan. 1998, vol. 7, No. 1, pp. 27-41.
Theano Nikitas; “Imaging Goes Quicksilver,”Digital Output, The Business Guide for Electronic Publishers.
David Bradburn Aragon; “Recognition Technology and Pattern Analysis,” Data Analysis Page.
“Multimedia Indexing and Retrieval Bibliography,” Center for Intelligent Information Retrieval.
Desire Gregory M
Mehta Bhavesh M.
Moore, Hansen & Sumner
LandOfFree
Method for representing and comparing digital images does not yet have a rating. At this time, there are no reviews or comments for this patent.
If you have personal experience with Method for representing and comparing digital images, we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Method for representing and comparing digital images will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-3287877