Facsimile and static presentation processing – Facsimile – Picture signal generator
Reexamination Certificate
2000-10-04
2004-08-03
Lee, Cheukfan (Department: 2622)
Facsimile and static presentation processing
Facsimile
Picture signal generator
C358S488000, C358S497000, C358S450000, C358S909100, C348S218100
Reexamination Certificate
active
06771396
ABSTRACT:
FIELD OF THE INVENTION
The present invention relates to the use of an electronic camera in a platenless document imaging system in which the document image is a composite image formed from a mosaic of overlapping images captured by the camera.
BACKGROUND OF THE INVENTION
In recent years, document scanners have become commonplace. Although these work well and are relatively inexpensive, a flatbed or platen-based document scanner occupies a significant amount of scarce desk space.
The use of a camera to take a photograph of a document consisting of text and/or images offers one way of dealing with the problem of wasted desk space. An electronic camera would need to have a detector with about 40 megapixel resolution in order to have a resolution over an A4 sized document comparable with that of the resolution of a typical document scanner, typically about 24 dots/mm (600 dpi). Such high-resolution detectors cost much more than the total cost of a desktop scanner.
As a result, it has been proposed to use an electronic camera with an actuator to scan the field of view of the camera over a document, and so form a composite image of the document from a number of overlapping image tiles. This permits less expensive lower resolution detector arrays to be used to build up an image of a document with a resolution comparable with that of a conventional document scanner. See, for example, patent document U.S. Pat. No. 5,515,181.
A problem with this approach is the fact that the image tiles must have some overlap, because it is impractical to use an actuator which moves the camera so precisely that tiles will fit together with no overlap. The conventional approach to fitting together overlapping tiles involves identifying features in the image of one tile in an overlap region and matching this against a corresponding feature in an adjacent tile's overlap region.
This feature matching approach suffers from various difficulties. First, computational algorithms to identify and match features are relatively slow compared with the process of gathering the images, which limits the throughput of a scanning camera document imaging system. Second, many documents have significant areas of blank space, for which it is not possible to match features. This necessitates the use of larger overlap areas to increase the likelihood that there will be suitable matching features in the overlap areas, with the result that more images must be captured. Third, it is possible that features will be incorrectly matched, particularly for text based documents in which common letters repeat frequently.
Another problem is that an image from an inexpensive camera will have some image distortion, particularly towards the edges of the field of view. The distortion is therefore strongest in the overlap region between tiles, which makes it more difficult to achieve a good overlap simply by matching features. As a result, it may be necessary to match several features over the extent of the overlap area to get a good fit between adjacent tiles.
As a result of problems such as these, scanning camera-based document imaging systems cannot yet compete with flatbed or platen-based document scanning systems.
The present invention addresses these and other problems.
SUMMARY OF THE INVENTION
In one embodiment disclosed herein, a method of calibrating an image capture system and subsequently capturing an image of a document comprises providing an electronic camera for imaging a portion of a document, an actuator for moving the camera over a document support surface and for cooperating with the camera to capture a plurality of overlapping image tiles of the document at different locations over the support surface and with a predetermined degree of overlap, and electronic processing means for joining the plurality of image tiles into a composite image of the document by generating from tile data points associated with each tile a corrected array of tile data points that correct for the expected distortion and predetermined overlap of neighboring image tiles in accordance with transform data for each image tile related to the predetermined overlap and to expected distortion for each image tile; providing a two-dimensional registration array within the field of view of the camera across an area corresponding with the document to be imaged, the registration array having a plurality of individually identifiable location identification features with a predetermined orientation and spacing amongst the features; using the camera to capture a plurality of overlapping image tiles of the registration array at predetermined locations and predetermined overlap, the locations and overlap corresponding to those to be used with the document to be imaged, each image tile having an array of tile data points that cover a plurality of location features; identifying for each image tile a plurality of individual location identification features, associating with the features particular tile data points and from the predetermined orientation and spacing of the specific features determining from the tile data points if there is any image distortion in that image tile; and generating the transform data for each image tile from the identity of the location identification features and the determined distortion; and wherein capturing the image subsequently comprises placing a document on the document support surface; positioning the camera above the document support surface; capturing a plurality of image tiles of the document at a plurality of different predetermined locations over the support surface with a predetermined degree of overlap between neighboring image tiles, each image tile having an array of tile data points and each predetermined location having a predetermined camera field of view; and causing the electronic processing means to generate from the transform data and the tile data points associated with each tile a corrected array of tile data points that correct for the expected distortion and predetermined overlap of neighboring image tiles, and to join the plurality of image tiles into a composite image of the document.
Using the camera to capture the plurality of overlapping image tiles may involve using the actuator to move the camera between image tiles in the same order as for the document to be imaged. The camera may further include a focus mechanism, the transform data may include separate data for different focus settings, and the method may further comprise focusing the camera on the document and selecting the transform data according to the focus setting.
Accordingly, the invention provides an image capture system, comprising: an electronic camera with an electronic detector and a lens with a field of view for imaging on the detector a portion of a document; an actuator for moving the camera field of view over a document support surface, the camera and the actuator co-operating so that a plurality of overlapping image tiles of a document can be captured at different locations over the support surface, each image tile having an array of tile data points and being subject to some expected perspective and/or camera distortion relative to the support surface; and electronic processing means by which the plurality of image tiles may be joined into a composite image of the document; characterised in that:
i) for each image tile the camera field of view relative to the support surface and the degree of overlap between neighbouring image tiles are predetermined;
ii) the electronic processing means includes a memory which stores transform data for each image tile, the transform data relating both to the expected distortion and to the predetermined overlap between image tiles; and
iii) the processing means is adapted to use the transform data to generate from the tile data points a corrected array of tile data points with said distortion corrected and with the corrected image tiles correctly overlapped with respect to neighbouring corrected image tiles to form a composite image of the document.
The image capture system may include a support
Cheatle Stephen Philip
Grosvenor David Arthur
LandOfFree
Document imaging system does not yet have a rating. At this time, there are no reviews or comments for this patent.
If you have personal experience with Document imaging system, we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Document imaging system will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-3350070