Image analysis – Image transformation or preprocessing – Combining image portions
Reexamination Certificate
1999-01-29
2003-07-15
Mehta, Bhavesh (Department: 2625)
Image analysis
Image transformation or preprocessing
Combining image portions
C382S276000, C382S305000, C358S450000
Reexamination Certificate
active
06594403
ABSTRACT:
BACKGROUND OF THE INVENTION
1. Field of Invention
This invention relates to image processing. More particularly, this invention is directed to registering partial image portions obtained from a scanner into a single larger image.
2. Description of Related Art
Image scanning has assumed increased importance for applications such as desktop publishing, image processing and World Wide Web publication. Currently, image scanning is generally done using either high-end drum scanners or low-end flatbed scanners. Drum scanners have higher pixel resolutions than flatbed scanners. Drum scanners are also able to scan large-sized original documents. However, drum scanners are more expensive than flatbed scanners, particularly as inexpensive CCD imagers, and in the future, light-sensitive CMOS sensing elements, are incorporated into flatbed scanners.
Moreover, drum scanners are not able to scan large-size original documents that are too rigid to adhere to the curved drum. Flatbed scanners are compact, but typically have a relatively small platen size. Scanners in general, and flatbed scanners in particular, must therefore register partially scanned images to each other to assemble and complete large images.
Different techniques are available in the art usable to establish correspondence between partial image portions, depending on the application and the type of deformation between image portions. “A Survey Of Image Registration Techniques,” L. G. Brown, ACM Computing Surveys, vol. 24, no. 4, December 1992, pp. 325-376, (hereinafter “Brown”) incorporated herein by reference in its entirety, provides a general overview of image-registration imaging. “A Contour Based Approach To Multi-Sensor Image Registration,” H. Li, et al., (hereinafter “Li”) IEEE Trans. Image Processing Vol. 4, March 1995, pp. 320-334; “MRI-PET Registration with Automated Algorithm,” R. P. Woods, et al. (hereinafter “Woods”), Journal of Computer Assisted Topography, vol. 17, 1993, pp. 536-546, “Video Mosaics for Virtual Environments,” R. Szeliski, (hereinafter “Szeliski”) IEEE Computer Graphics and Applications, March 1996, pp. 20-30, and “Salient Video Stills: Content and Context Preserved,” by L. Teodosia et al. (hereinafter “Teodosia”), ACM Multimedia, 1993, pp. 39-46, each of which is incorporated herein by reference, in its entirety, discuss telereality applications and still video representation.
SUMMARY OF THE INVENTION
In image registration, a spatial transform between images under consideration generally needs to be found to allow the misalignment between partial images caused during scanning to be removed. The misalignment represents the translation and rotation between images (global transformation) and local deformations within each partial image (local transformation). Matched filtering techniques can be used to find the spatial transformation. Several matched filtering techniques are described in Brown. In these methods, the spatial transform is considered to be a pure translation. The major disadvantage of these techniques is their computational complexity.
Another known method for image registration is to use wavelet analysis to model the spatial transformation. This is proposed, for example, in “A Nonlinear Variational Problem for Image Matching,” Y. Amit, SIAM Journal Sci. Comput., vol. 15, no. 1, January 1994, pp. 207-224, incorporated herein by reference in its entirety. Yet another approach is to estimate the spatial transformation from the phase of a wavelet decomposition of images, as described in “Motion Estimation Using Complex Wavelets,” J. Magarey et al., Technical Report, Cambridge University, Dept. of Engineering, 1995, incorporated herein by reference in its entirety. The basic disadvantage of these techniques is that they model local transformations instead of global ones, which are the dominant transformation for scanning purposes.
“Accurate Mosaiking of Scanned Maps, or How to Generate a Virtual A0 Scanner,” J. van Katwijk et al., Proc. Of the First Annual Conf. of the Advanced School for Computing and Imaging, The Netherlands, May 1995, pp. 353-359, incorporated herein by reference it its entirety, proposes another image registration technique for registering partial images of scanned documents. With this method, scanned utility maps can be converted to structural descriptions. As a deformation model, affine transformation, which is robust for estimating global transformations, is used. However, the regions on which the transformation is estimated are pre-drawn on the images, requiring manual intervention.
This invention provides systems and methods for estimating the global transformation between scanned partial images of an original document to reconstruct a scanned image of the original document.
This invention separately provides systems and methods that determine the global transformation using a transformation.
This invention separately provides systems and methods that determine the global transformation using an affine transformation.
This invention separately provides systems and methods that divide each partial image into non-overlapping blocks and that estimate the global transformation for each block.
This invention separately provides systems and methods that cluster the estimated block parameters to find the actual transformation for the overall partial images.
These and other features and advantages of this invention are described in or are apparent form the following detailed description of the preferred embodiments.
REFERENCES:
patent: 4393410 (1983-07-01), Ridge et al.
patent: 4945502 (1990-07-01), Kwon et al.
patent: 5339170 (1994-08-01), Fan
patent: 5557684 (1996-09-01), Wang et al.
patent: 5706416 (1998-01-01), Mann et al.
patent: 5784115 (1998-07-01), Bozgadi
patent: 5819258 (1998-10-01), Vaithyanathan et al.
patent: 5838837 (1998-11-01), Hirosawa et al.
patent: 5917962 (1999-06-01), Chen et al.
patent: 5926578 (1999-07-01), Fukase et al.
patent: 5999927 (1999-12-01), Tukey et al.
patent: 6002492 (1999-12-01), Kamon et al.
patent: 6011430 (2000-01-01), Kim
“A Survey of Image Registration Techniques”, L. G. Brown, ACM Computing Surveys, vol. 24, No. 4, Dec. 1992.
“Video Mosaics for Virtual Environments”, R. Szeliski, IEEE Computer Graphics and Applications, pp. 22-30, Mar. 1996.
“Salient Video Stills: Content and Context Preserved”, L. Teodosio et al., ACM Multimedia, pp 39-46, 1993.
“A Nonlinear Variational Problem for Image Matching”, Y. Amit, SIAM J. Sci. Comput., vol. 15, No. 1 pp. 207-224, Jan. 1994.
Bozdagi Gozde
Harrington Steven J.
Mehta Bhavesh
Oliff & Berridg,e PLC
Patel Kanji
Xerox Corporation
LandOfFree
Systems and methods for registering scanned documents does not yet have a rating. At this time, there are no reviews or comments for this patent.
If you have personal experience with Systems and methods for registering scanned documents, we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Systems and methods for registering scanned documents will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-3048120