Image analysis – Image enhancement or restoration – Artifact removal or suppression
Reexamination Certificate
1998-09-02
2003-02-25
Rogers, Scott (Department: 2624)
Image analysis
Image enhancement or restoration
Artifact removal or suppression
C348S276000, C348S241000, C358S513000, C358S463000
Reexamination Certificate
active
06526181
ABSTRACT:
FIELD OF THE INVENTION
The present invention relates to an imaging apparatus such as electronic or digital still cameras and digital video cameras. Particularly, the present invention relates to an apparatus and a method for eliminating the noise of image data output by a photoelectric transducer element in an imaging apparatus of such an apparatus.
BACKGROUND OF THE INVENTION
In digital still or video cameras, an optical image must be converted to an electronic image. A photoelectric transducer element such as a charge coupled device (CCD) is commonly used for such conversion.
In such a digital device, it is common for image data to be composed of separate color components such as red (R), green (G), blue (B), CMY, etc. A number of digital cameras use separate CCDs for each color, such as a CCD for red (R), a CCD for blue (B) and a CCD for green (G).
However, in order to reduce manufacturing costs and simplify equipment, many devices instead acquire image data for three primary colors of R, G and B using a single CCD. To acquire image data of the three primary colors using the one CCD, a method of attaching a mosaic optical filter to the single CCD is widely utilized. Such a filter is called a color filter array (CFA) and, when such a filter is attached to a CCD, data for each color is detected in a separate pixel. For example, data for red (R) may be detected in one pixel, and data for green (G) may be detected in the adjoining pixel.
The colors in a color filter may be arrayed in a variety of patterns with a Bayer color array being a common example of a typical array.
FIG. 1
shows how colors are arranged in a Bayer array color filter.
As shown in
FIG. 1
, there are twice as many pixels of G as R or B in the Bayer array. This is because the G information is more important for human vision.
In an image processing unit for acquiring image data using a CCD, the mismatch of gain in an even line and in an odd line may occur.
This type of mismatch is especially significant in devices where two output terminals are provided for an even line and one output terminal is provided to an odd line of a CCD. The mismatch becomes remarkable because, when two output terminals are provided, a separate output is provided for respective output terminals and differences between amplifier gains may be reflected in their respective output signals. That is, a difference between amplifier gains of the amplifiers appear as a difference in magnitude between output signals.
Such mismatches may also occur in CCDs provided with only one output terminal. Generally, when the data of a pixel is output, the value of a pixel being output is often influenced by the value of the previously output pixel because of the amplification characteristics of the amplifier provided to the one output terminal and a mismatch may result.
In this case, a mismatch may occur because, if the data of a pixel is influenced by that of the previous pixel as described above, the value of a green (G) pixel in an odd line is influenced by the value of the adjacent red (R) pixel because of the characteristics of a Bayer array. Similarly, the value of a green (G) pixel in an even line is influenced by the value of the adjacent blue (B) pixel. Therefore, if, for example, a red object is photographed, green in an odd line will be strongly influenced by the adjacent red pixel, however, green in an even line will be little influenced by the adjacent pixel. As a result, the values for green pixels in the odd and even lines will differ and, as a result, a mismatch will occur.
A method for adjusting an amplifier and an attenuator for adjusting a difference between gains in odd and even lines in cameras with a CCD with two output terminals, as well as a method of storing a calibration value and correcting gains in an odd line and in an even line in image processing are known or proposed.
However, the factors leading to a CCD gain mismatch are often variable and inconstant; for example, the effect of the previous pixel in the same line varies greatly with temperature. It is therefore in practice very difficult to execute any of the above described methods and their application remains, for the most part, theoretical.
Noise caused by the mismatch of gains in odd and even lines commonly causes lateral stripes in a final image. Further, interpolation, edge highlighting processing, or other processes applied to the image containing these lateral stripes may compound the negative effects. The quality of a final image may be severely deteriorated.
As such differences between gains in odd and even lines may be regarded as striped noise, the elimination of the striped noise by noise elimination processing in the final step of image processing may be also taken into consideration. However, there is then a problem that a considerable amount of time is required for image processing and a further problem that the edge of an image may blur because of the side effects of noise elimination.
SUMMARY OF THE INVENTION
The present invention is made to solve the problems outlined above and has an object of providing a noise elimination method and apparatus which can eliminate noise caused by the mismatch of odd and even line gains from image data acquired from a CCD provided with a color filter. The present invention is characterized in that, in order to solve the above problems, components as described below are included in a noise elimination apparatus in order to eliminate the noise of Bayer-type image data output by a photoelectric transducer element provided with Bayer-type color filter.
That is, the present invention is characterized in that the noise elimination apparatus includes high frequency component quantity detecting means for detecting the value of a high frequency component in each pixel of the Bayer-type image data, graduation quantity calculating means for calculating the quantity of graduation in each pixel of the above Bayer-type image data, and optimum graduation quantity determining means for changing the above quantity of graduation to an optimum value based upon the value of a high frequency component detected by the high frequency component quantity detecting means and then adding the changed quantity of graduation to each pixel.
By adding an optimum quantity of graduation based upon the value of a high frequency component of each pixel, the deterioration of image quality can be reduced.
The present invention may also be characterized in that the above high frequency component quantity detecting means may include a Laplacian filter to which the value of a target pixel and the values of the four pixels immediately adjacent to the target pixel are input. A signal output from the above Laplacian filter is output as the value of a high frequency component. The input of these pixel values enables the reduction of the effect of noise on each line.
The high frequency component quantity detecting means of the present invention may also comprise a Laplacian filter to which the value of a target pixel and of the four immediately adjacent pixels are input, upper and lower difference calculating means for outputting the absolute value of difference between the values of the upper and lower adjacent pixels to the target pixel, right and left difference calculating means for outputting the absolute value of difference between the values of the pixels to the right and left of the target pixel, and addition means for weighting and adding a signal output from the above Laplacian filter, a signal output from the above upper and lower difference calculating means and a signal output from the above right and left difference calculating means. A signal output from the above addition means is output as the value of a high frequency component. This configuration reduces the effect of noise on each line and thereby reduces, or eliminates, the resulting deterioration of image quality.
The graduation quantity calculating means of the present invention may include mean value calculating means for acquiring the weighted mean value of the value of a target pixel an
Miyano Toshiki
Omata Kyoichi
Smith Craig M.
Crocker Pamela R.
Rogers Scott
LandOfFree
Apparatus and method for eliminating imaging sensor line noise does not yet have a rating. At this time, there are no reviews or comments for this patent.
If you have personal experience with Apparatus and method for eliminating imaging sensor line noise, we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Apparatus and method for eliminating imaging sensor line noise will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-3172612