Image analysis – Histogram processing – With pattern recognition or classification
Reexamination Certificate
2000-01-03
2001-01-30
Johns, Andrew W. (Department: 2721)
Image analysis
Histogram processing
With pattern recognition or classification
C382S305000, C707S793000
Reexamination Certificate
active
06181818
ABSTRACT:
BACKGROUND OF THE INVENTION
The present invention relates to a content-based image retrieval method and apparatus for designating an image or figure and retrieving image data similar to the designated image or figure in, e.g., an image database or an image file system.
As an image retrieval method for searching a plurality of images stored in an image file format to find a desired image, the following methods are conventionally used.
(1) In a so-called template matching method, when a designated image as a search template or search key is input, the designated image is sequentially superposed on image data stored in an image file to be searched, a sum total of differences between the pixels of the designated image and those of each image data in the image file is calculated, and the minimum value of the sum totals is determined as a difference between the designated image and the image data in the image file, thereby selecting image data having a difference equal to or smaller than a predetermined value as similar images.
(2) In another method, image data stored in an image file are segmented into some regions having similar pixel values, the correspondences between data that describe the image regions and a designated image or figure are calculated using characteristics such as an area, secondary moment, or the like, and image data having the largest similarly is selected as a similar image.
However, in the conventional method (1), the designated image must be moved with respect to image data to be searched, and a difference in the number of corresponding pixels must be calculated in each movement. For this reason, the search processing requires much time. Upon search, since an accurate search result cannot be obtained unless the designated image is compared while being moved in units of sufficiently small distances in a two-dimensional region of image data, a very long calculation time is required.
In the conventional method (2), a normal search operation is disturbed when one region of image data in the image file to be searched does not have a one-to-one correspondence with one region of an image designated as a search key, i.e., when an image in the image file is over-segmented into smaller regions than the corresponding regions in a designated image.
In a so-called segmentation technique for automatically segmenting image data into a plurality of regions, since a target area which is perceived by a person as one region is often divisionally extracted as a plurality of portions, such a limitation poses a serious problem in an application.
When an image retrieval operation is performed, calculations for checking matching between a designated image or figure to be retrieved (to be referred to as a designated image hereinafter) and each region of all images stored in the image file system are required. When there are a large number of stored images, the calculation amount becomes huge, and the time required for image retrieval also becomes very long. In order to solve this problem, a method for decreasing the number of images to be subjected to matching calculations by limiting the stored images using the color information of the stored images and the designated image to attain high-speed search processing has already been proposed.
However, in the above-mentioned search method, for example, when a stored image includes a large number of fine regions having various colors, a color similar to the color of a closed figure designated by the designated image is included in the stored image with high possibility independently of an image designated as the designated image. As a result, even when an image including such a region is not actually data to be searched, it is undesirably selected as data for which distance calculations are to be performed, and the effect of the limitation processing cannot be expected.
SUMMARY OF THE INVENTION
The present invention has been made in consideration of the above-mentioned prior arts, and has as its object to provide an image retrieval method and apparatus, which can efficiently retrieve image data even when a search region in image data is divisionally extracted as a plurality of regions.
It is still another object of the present invention to provide an image retrieval method and apparatus, which can efficiently retrieve an image by shortening the time required for similarity calculations of images.
It is still another object of the present invention to provide an image retrieval method and apparatus, which can efficiently retrieve an image by obtaining movable regions in which each region of an image to be searched and a designated image have a specific phase relationship therebetween, and discriminating the presence/absence of a crossing region of the movable regions.
It is still another object of the present invention to provide an image retrieve method and apparatus, which can quickly and accurately search for a desired image in such a manner that edge information of a region of an image or figure pattern or color information of a surrounding region is input as image description information, the number of image data to be accessed is decreased using various kinds of description information of image data on the basis of the input image description information, and a limited number of images are accessed to retrieve the image.
It is still another object of the present invention to provide an image retrieval method and apparatus, which can search for a desired image by inputting only the edge position and its surrounding color of an image.
It is still another object of the present invention to provide a high-speed image retrieval method and apparatus, which allow a reliable operation even when a designated image is a portion of a target image or an image to be retrieved, and can sufficiently limit the number of images to be tested.
It is still another object of the present invention to provide an image retrieval method and apparatus, which can realize high-speed image search processing by utilizing information associated with the area of a closed region of a designated image to be retrieved for.
It is still another object of the present invention to provide an image retrieval method and apparatus, which can realize high-speed image search processing by limiting the number of images to be tested and, thereafter, discriminating the similarities of only the limited images.
Other features and advantages of the present invention will be apparent from the following description taken in conjunction with the accompanying drawings, in which like reference characters designate the same or similar parts throughout the figures thereof.
REFERENCES:
patent: 5129011 (1992-07-01), Nishikawa et al.
patent: 5576950 (1996-11-01), Tonomura et al.
patent: 5586197 (1996-12-01), Tsujimura et al.
patent: 5644765 (1997-07-01), Shimura et al.
patent: 5652881 (1997-07-01), Takahashi et al.
patent: 5734893 (1998-03-01), Li et al.
patent: 5802361 (1998-09-01), Wang et al.
NEC Research and Development, Oct. 1994, Japan, vol. 35, No. 4, ISSN 0547-051X, pp. 410-420, Hirata K. et al. ‘The concept of Media-Based Navigation And Its Implementation On Hypermedia system “Miyabi”’.
NEC Research and Development, vol. 34, No. 2, Apr. 1, 1993, pp. 263-273, XP 000322020, Kyoji Hirata et al. ‘Rough Sketch-Based Image Information Retrieval’.
Proceedings of the International Conference on Multimedia Computing and Systems (Cat. No. 94TH0631-2), Proceedings of IEEE International Conference on Multimedia Computing and Systems, Boston, MA, USA, May 15-19, 1994, ISBN 0-8186-5530-5, 1994, Los Alamitos, CA, USA, IEEE Comput. Soc. Pres, USA, pp. 121-130, Yihong Gong et al. ‘An Image Database System With Content Capturing and Fast Image Indexing Abilities’.
Proceedings ICIP-94 (Cat. No. 94CH34708), Proceedings of 1st International Conference on Image Processing, Austin, TX, USA, Nov. 13-16, 1994, ISBN 0-8186-6952-7, 1994, Los Alamitos, CA USA, IEEE Comput. Soc. Press, USA pp. 76-80 vol. 2, Lee D et al. Query By Image Content Using Multiple Objects and Multiple Features: User Interface Issu
Bannai Yuichi
Okazaki Hiroshi
Sato Hiroaki
Tsujimura Kazuko
Yoshitani Akihiro
Canon Kabushiki Kaisha
Fitzpatrick ,Cella, Harper & Scinto
Johns Andrew W.
LandOfFree
Image retrieval method and apparatus does not yet have a rating. At this time, there are no reviews or comments for this patent.
If you have personal experience with Image retrieval method and apparatus, we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Image retrieval method and apparatus will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-2497430