Image analysis – Image segmentation – Distinguishing text from other regions
Reexamination Certificate
2000-10-17
2004-08-31
Johns, Andrew W. (Department: 2621)
Image analysis
Image segmentation
Distinguishing text from other regions
C382S194000
Reexamination Certificate
active
06785416
ABSTRACT:
TECHNICAL FIELD
This invention relates generally to the processing of scanned electronic image data. More particularly, this invention relates to the processing of scanned image data with a higher resolution than conventionally scanned image data.
BACKGROUND OF THE INVENTION
A pixel (“picture element”) is the smallest logical unit of a graphical image. Each pixel corresponds to a single region of the image. The color of each pixel is encoded by a value. The number of bits used in the encoding is dictated by how many colors may be encoded with a full color encoding or how many shades of gray may be encoded with a grey scale encoding. For example, an eight bit value accommodates the encoding of 2
8
colors. For color pixels, the value for a pixel may be an intensity value, having three components corresponding to intensity values for red, green and blue, respectively. These three colors are used additively to create other colors. In the well known red, green, blue (RGB) encoding scheme, the other colors are encoded as combinations of red, green and blue components having associated intensities. The encoding is represented in the following format (red intensity, green intensity, blue intensity) as a triplet. For example, the triplet (0, 192, 192) represents the color cyan, while the triplet (255, 0, 0) represents the color red.
The number of pixels in an image depends upon the resolution of the image. Resolution is typically measured in terms of pixels or “dots” per inch (dpi). As technology has advanced, the industry standard for the resolution of electronic devices that display images has increased from 300 dpi to 600 dpi. Thus, the general trend has been towards increasing the resolution of display devices.
Scanners scan hard copies of images and digitize the images. The result of scanning an image is a digital representation of image where the image is represented as a sequence of pixels having associated color values. The processing of scanned image data requires differentiation among the different types of data that may be scanned. Varieties of data include halftone image regions, text image regions and photographic image regions in the data. Halftone image regions are produced by the process of halftoning, which creates new colors and shades by varying the patterns of dots/shapes making up the image.
Text image regions are regions that contain text, and photographic image regions are regions containing photographs. Halftone printing uses a grid of uniform squares with black or colored dots/shapes inside each square. By varying the amount of the square in the grid that is filled in by the dot/shape and therefore, the amount of the square that is left white, an illusion of continuous variations of gray or color tones is created. The effect occurs because the human eye averages features which are not clearly seen as distinct from surrounding objects. Images commonly found in newspapers are created using this method.
When a scanner scans a printed halftone image, the scanner scans the dot/shape patterns that were used to print the halftone image. The halftone dots/shapes that were scanned by the optical scanner may result in interference patterns known as “moiré” appearing in the scanned image. The moiré patterns result in a checkered, banded or dotted appearance. Consequently, when a halftone image is scanned, a process known as smoothing or descreening is used to reduce the transfer of moiré patterns into the resulting scanned image file. Text image regions correspond to places where text appeared in the original document and need “sharpening.” Sharpening accentuates the difference between an edge of text and the white space next to the text. The accentuation is done by making dark areas darker and white areas whiter. Photographic image regions need to be not modified or smoothed or sharpened. Electronic devices performing end processing of scanned image data, such as a high end copier, must be able to differentiate among different types of image regions. Failure to accurately distinguish among different types of image regions leads to inferior image quality due to the wrong type of processing being applied to an image region.
Segmentation classifies the scanned image data on a pixel by pixel basis (e.g. classifies as a halftone region, a text image region or a photographic image region. Segmentation has been developed to identify image regions for 300 dpi images. However, conventional segmentation techniques for 300 dpi images are not readily adaptable for 600 dpi images. The amount of data in the 600 dpi images is much greater than the amount of data found in 300 dpi images. For example, the 5×5 segmentation window currently used by copiers during the processing of 300 dpi images has to be increased to a 9×9 window to encompass the same amount of data included with the 600 dpi images. The associated computational costs of scaling the current algorithms in use for 300 dpi image processing to work with 600 dpi images is significant.
SUMMARY OF THE INVENTION
The present invention addresses the shortcomings of applying conventional 300 dpi image detection algorithms to scanned images with resolutions of 600 dpi or higher. The illustrated embodiment of the present invention uses image detection algorithms designed specifically for the amount of data contained in an image with a resolution of 600 dpi or greater. The present invention distinguishes between halftone image areas, large text image areas, small text image areas and photographic image areas and then applies appropriate image processing to each area.
In one embodiment of the present invention, scanned images with a resolution of at least 600 dpi are analyzed by an image detection algorithm designed to detect the probable presence of halftone image data and large text image data. A pixel window is used by the image detection algorithm to help perform calculations on the pixel values of each pixel in the image data. An individual pixel (the focus pixel) is analyzed in the context of the surrounding image area in order to determine if the image area fits a pattern displayed by halftone image areas or large text image areas. Appropriate image processing is then applied to the image based on the result.
In another embodiment of the present invention, scanned images with a resolution of at least 600 dpi are analyzed by an image detection algorithm designed to detect the probable presence of small text image data. A pixel window is used by the image detection algorithm to help perform calculations on the pixel values of each pixel in the image data. An individual pixel (the focus pixel) is analyzed in the context of the surrounding image area in order to determine if the image area fits a pattern displayed by small text image areas. Appropriate image processing is then applied to the image based on the result.
In yet another embodiment of the present invention, scanned images with a resolution of at least 600 dpi are analyzed by an image detection algorithm designed to detect the probable presence of the text characters ‘=’ and ‘×’. A pixel window is used by the image detection algorithm to help perform calculations on the pixel values of each pixel in the image data. An individual pixel (the focus pixel) is analyzed in the context of the surrounding image area in order to determine if the image area matches a pattern displayed by the text character ‘=’ or if the image area matches a pattern displayed by the text character ‘×’. Appropriate image processing is then applied to the image based on the result.
In yet another embodiment of the present invention, scanned images with a resolution of at least 600 dpi are analyzed by an image detection algorithm designed to detect the probable presence of the text character ‘≡’. A pixel window is used by the image detection algorithm to help perform calculations on the pixel values of each pixel in the image data. An individual pixel (the focus pixel) is analyzed in the context of the surrounding image area in order to determine if the image
Andree Fred
Pratt Steve
Yu Shenbo
Zhang Wei
Alavi Amir
Johns Andrew W.
Lahive & Cockfield LLP
Oak Technology, Inc.
LandOfFree
System and method for the processing of scanned image data... does not yet have a rating. At this time, there are no reviews or comments for this patent.
If you have personal experience with System and method for the processing of scanned image data..., we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and System and method for the processing of scanned image data... will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-3356218