Image analysis – Image compression or coding – Gray level to binary coding
Reexamination Certificate
1998-05-26
2001-10-23
Lee, Thomas D. (Department: 2724)
Image analysis
Image compression or coding
Gray level to binary coding
C358S451000, C358S451000
Reexamination Certificate
active
06307972
ABSTRACT:
FIELD OF THE INVENTION
The present invention relates to a method and apparatus for generating a bitmap which is representative of an original image such as a film separation.
DESCRIPTION OF THE PRIOR ART
Input scanners conventionally sample an original image (such as a continuous tone image or a film separation) and generate a bitmap suitable for output to a bi-level printing device.
A problem with known input scanners is that they can take a significant amount of time to scan the original image at the resolution required by the bi-level printing device.
SUMMARY OF THE INVENTION
In accordance with a first aspect of the present invention there is provided a method of generating a bitmap representative of an original image, the method comprising
(1) scanning the original image to generate a greyscale pixel map comprising a plurality of greyscale pixel values by;
(i) directing a light beam onto the original image whereby the light beam is modulated by the image to generate a modulated light beam;
(ii) causing relative scanning movement between the light beam and the original image;
(iii) detecting the modulated light beam to generate a picture signal; and
(iv) generating the plurality of greyscale pixel values from the picture signal;
(2) interpolating the greyscale pixel map to generate a higher resolution greyscale pixel map; and
(3) converting the higher resolution greyscale pixel map into the bitmap.
The method enables an original image to be scanned more quickly and reduces the amount of image data which needs to be processed.
The interpolation may comprise a linear 2-point interpolation. However preferably the interpolation comprises a 4-point interpolation which provides a small amount of sharpening of the image. In a further alternative, an interpolation algorithm having an order greater than 4 could be used, which in some cases may improve image quality.
Conventional input scanners may be used to scan film separations, ie. films which carry information on a particular colour component of an image. Typically four film separations need to be scanned for each image, each corresponding to one of the four conventional printing colours cyan, magenta yellow or black (CMYK). It is important for the scanned images to be in register. Conventionally this is achieved by manually rotating the film separations on the scanner cylinder. This is inaccurate and time consuming.
In accordance with a second aspect of the present invention there is provided a method of producing an output image from a plurality of colour separations comprising
(1) scanning each colour separation to generate a respective plurality of greyscale pixel maps;
(2) rotating one or more of the greyscale pixel maps to correct any misregistration between the colour separations;
(3) converting each greyscale pixel map into a respective bitmap; and
(4) producing the output image by superimposing the bitmaps.
The second aspect of the present invention enables the bitmaps to be accurately and easily registered with each other. It has been appreciated that if the bitmap (as opposed to the greyscale pixel map) is rotated this can result in systematic errors which are visible in the final image. By rotating the greyscale pixel map these errors are minimised.
In a preferable embodiment the methods of the first and second aspects of the present invention are combined. The interpolation and rotation may be carried out sequentially but preferably the greyscale pixel map is rotated and interpolated in a single composite transformation. This reduces the amount of computation required.
In a preferable embodiment the original image is a screened image such as a film separation. Scanning such an original image is known conventionally as “copy dot scanning” in which an original screened image comprising a number of (typically bi-level) dots is scanned. The film separations have typically been previously created by a conventional digital halftoning technique. That is, the separations are created by scanning an original continuous tone image to generate cyan, magenta, yellow and black greyscale representations of the image, converting the greyscale representations into binary form suitable for printing, and printing the bitmap on film. The film separations are conventionally known as binary screened images since they are binary images which have been created by sampling an original image using a screen.
A film separation can be modelled as a digital black and white bitmap. That is, film separations when examined at high magnification can be seen to be a regular grid of black dots on a clear background.
FIG. 13
illustrates a magnified portion of a film separation along with its associated screen grid. The magnified portion contains three half-tone dots
100
-
102
on a screen grid
104
. Ideally the grid
104
would be scanned at exactly the resolution at which it was produced, and in register, so that the sampled pixel locations are centred on the original pixel locations. However due to inaccuracies in generating the original, film stretch during processing or handling, and problems matching the input scanner screen grid with the screen grid
104
this ideal is impossible to achieve. This problem is illustrated in
FIG. 14
which illustrates the sampling of the dot
100
using an offset sampling grid
105
. The greyscale representation of the dot
100
is illustrated at
106
, where the numbers illustrate the percentage density at each sampling point. Conventional techniques then convert the greyscale representation of the dot using thresholding techniques. If the threshold is set at 50% then the bitmap representation of the dot is illustrated at
107
. If the threshold is set at 40%, then the bitmap representation will be as illustrated at
108
. It can be seen that neither
107
nor
108
are accurate representations of the dot
100
. This will result in tonal change—for instance where a 50% threshold is used, only four black pixels are generated and as a result the image will be lighter than required. If a 40% threshold is used, then seven black pixels are generated, which will result in the image being darker than required. Where there are few black pixels in the original there will tend to be fewer in the regenerated bitmap. Where nearly all the pixels in the original are black even more will be black in the regenerated bitmap. In addition to the tonal changes there are also problems with regenerating the original cluster shapes, with the edges tending to become jagged.
In accordance with a third aspect of the present invention there is provided a method of generating a bitmap representative of an original image, the bitmap comprising a plurality of white/black binary pixel values each having a respective pixel location, the method comprising
(1) scanning the original image to generate a greyscale pixel map, the greyscale pixel map comprising a plurality of greyscale pixel values each having a respective pixel location; and
(2) converting the greyscale pixel map into the bitmap by
(i) ranking each greyscale pixel value against the greyscale pixel values of a neighbourhood of adjacent pixel locations;
(ii) determining a desired number of black binary pixels (B_Pc) in the neighbourhood;
(iii) comparing the rank of the greyscale pixel with the desired number of black binary pixels; and
(iv) assigning a black binary pixel value to the pixel location when the comparison carried out in step (iii) satisfies a predetermined condition.
This method ensures that the number of black dots in the original image is accurately reproduced in the bitmap, and results in more accurate tonal reproduction. Typically a black binary pixel value is assigned to the pixel location when the rank of the greyscale pixel is greater than or equal to the desired number of black binary pixels.
It will be understood that the terms white/black are simply labels which refer to the two values (i.e. 1/0) which can be assigned to each binary pixel, and do not necessarily relate to the tonal or colour content of the image.
Typically step (ii) comprises summing the grey
Gouch Martin Philip
Riley Andrew James
Fujifilm Electronic Imaging Ltd
Lee Thomas D.
Sughrue Mion Zinn Macpeak & Seas, PLLC
LandOfFree
Method and apparatus for generating a bitmap does not yet have a rating. At this time, there are no reviews or comments for this patent.
If you have personal experience with Method and apparatus for generating a bitmap, we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Method and apparatus for generating a bitmap will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-2593231