Abstract: An image processing technique for selecting a text region from an image is described. Character and formatting information for each word in the image is used to determine an active region for each word in the image. For a preferred embodiment of the present invention, the character and formatting information is derived during optical character recognition (OCR). A first and last word within a selected text region is identified based on at least one active region associated with at least one word within the selected text region. Using the first and last words within the selected text region, all words within the selected text region are identified. An image of the selected text region may be displayed. Text contained within the selected text region may be copied to an application program.

Abstract: A document image capture (scanning) system and control method are described for scanning and processing document images received live from a camera. A motion detector detects image motion between two image frames. When the image is stationary, image processing (such as OCR) is carried out automatically and made available to the operator. In one form, when movement is detected, the image processing results are discarded until the image is newly stationary, whereupon new image processing is carried out on the new image. In another form, the degree of movement is evaluated; if the movement is small, then at least some of the previous image processing results are re-used by re-mapping on to the new image.

Abstract: A thresholding technique for generating a background image of a pixmap image is described. For one embodiment of the present invention, background image is used to generate a background threshold surface for thresholding a pixmap image. For another embodiment of the present invention, the background image is used to normalize the background of the pixmap image. In yet another embodiment of the present invention, the background image is used as input for other image processing algorithms such as extracting features from a grey-scale image for character recognition.

Abstract: A device and method are described for capturing an image of a document using a digital camera. The invention includes a resolution analyzer for analyzing the resolution of text characters in the image, to assess whether the resolution is suitable for OCR. In one aspect, if the resolution is too small or too large, an indicator is generated to guide the user as to the corrective movement of the camera to improve the resolution for OCR. In a second aspect, an electronic zoom control is controlled to alter the magnification of the captured image, to correct the resolution. In a third aspect, a method is described for analyzing the image to determine the resolution according to the height of text characters in the image. The method includes analyzing the run lengths of pixels in the image to identify a predominant length corresponding to the average character height (in pixels).

Abstract: Perspective distortion is estimated in a digital document image by detecting perspective pencils in two directions, one being parallel to text lines, and the other being parallel to the boundaries of formatted text columns. The pencils are detected by analyzing directional statistical characteristics of the image. To detect a pencil, a first statistical line transform is applied to transform the image into line space, and a second statistical score transform is applied to transform the image into pencil space. A dominant peak in pencil space identifies the perspective pencil. In addition, a computationally efficient line summing technique is used for effecting sums of pixels along inclined target lines (or curves) through an image. The technique includes pre-generating partial sums, and summing along step segments of a target line using the partial sums.

Abstract: A composite higher resolution image of an object can be obtained by combining or patching together multiple low-resolution camera images to form a composite image having a higher resolution. Each low-resolution camera image represents a subsection of an object (also referred to as “object subsection”) to be scanned. For example, each subsection may represent a quadrant of a document page. With the use of a periscope, the image visible to the camera can be shifted to capture images of the various object subsections. Mosaicing techniques are used to combine these camera images into a composite image of the object. This approach to scanning improves the resolution of documents scanned while using existing low cost video cameras.

Abstract: A document camera is described for capturing an electronic image of a document, for subsequent optical character recognition. In order to avoid the need for a user to look through a viewfinder of the camera, the camera comprises a projector for projecting an optical field of view indicator on to the document. The user can see the projected indicator on the document page, indicating the effective field of view of the camera. The camera may include auto-focus and zooming mechanisms. In one form, the camera processes a detected image to determine a capture zone corresponding to whole regions of the document falling entirely within the field of view. The projected indicator is controlled to indicate only the capture zone. When the image is to be captured, the camera zooms in to the capture zone to capture the image at maximum resolution.

Abstract: An electronic camera imaging system records a color filter array (CFA) image. The missing color values of the CFA image are reconstructed using a linear transformation. The coefficients of the linear transformation are estimated in one embodiment by computing an average value over a window of the CFA image by equalizing the mean and the variance of the color channels of the window. In another embodiment, the coefficients of the linear transformation are estimated by fitting a line to a set of the sums of column and rows of color channels of a window of the CFA image.

Abstract: Following scanning of a document image, and optical character recognition (OCR) processing, the outputted OCR text is processed to determine a text format (typeface and font size) to match the OCR text to the originally scanned image. The text format is identified by matching word sizes rather than individual character sizes. In particular, for each word and for each of a plurality of candidate typefaces, a scaling factor is calculated to match a typeface rendering of the word to the width of the word in the originally scanned image. After all of the scaling factors have been calculated, a cluster analysis is performed to identify close clusters of scaling factors for a typeface, indicative of a good typeface fit at a constant scaling factor (font size).

Abstract: A point spread function (PSF) of a digital camera is measured in two-dimensions. To measure the PSF in two dimensions a grid of dots is defined on a calibration chart. The calibration chart is recorded with the digital camera to define a calibration image. The grid of dots recorded in the calibration image has dots with three different sizes. The two of the largest sizes of dots are used to provide a coarse-to-fine registration of the calibration image with the calibration chart. Once aligned, the locations of the smallest of the dots in the calibration image are used to define a composite dot image. The level of noise in the composite dot image is minimized by averaging over many small dots in the grid of dots. In addition, the composite dot image is super-resolved in order to minimize the effects of aliasing. Once the PSF is measured for the digital camera, the PSF can be used in a restoration filter to remove blur in images recorded with the digital camera.