JPS63144682A - Image processor - Google Patents

Abstract

PURPOSE:To easily and exactly erase noise from a still picture by making the color difference values of plural picture elements below the specified value continuously the same by a value obtained by multiplying the brightness of each picture element neighboring the vertical direction of the screen and the color differences by a prescribed coefficient, respectively, then adding them together. CONSTITUTION:Video signals are recorded in a Y memory 21, an R-Y memory 22, and a B-Y memory 23. A CPU 40 fetches data in the unit of one picture element in the order of printing, and the data are converted respectively to a C, M, Y signal by a conversion circuit 50. These signals are masked in a masking circuit 60, and printed by a head driven by a head driver 70. In the case, the CPU 40 generates differentials in illuminance Y between a fetched picture element and ones neighboring it in the vertical direction along with color differences(R-Y, B-Y) between the same, multiplies the results with a prescribed value, then adds them with each other respectively. In case resulting values are less than the prescribed value sequentially, said picture elements are regarded as belonging to the same part of an image, and the color difference values of said picture elements are made same. As a result, noises caused by the difference between color difference due to jitters, etc., are eliminated.

Description

[Detailed description of the invention] [Industrial application field] The present invention relates to an image processing device.

[Prior Art] In recent years, image output devices such as video printers that take color hard copies of moving images from TVs, VTRs, video discs, etc. as still images have been released, and provide images with image quality close to the original. In addition, the user can use TV, VTR,
People use this method to watch moving images on video discs, etc., and save the scenes they particularly like as hard copies for enjoyment.

At this time, the most problematic point is that noise such as jitter, which is not noticeable in a moving image, becomes very noticeable when the image is still, resulting in a still image with poor image quality.

In addition, the memory for one image is not a frame memory (in many cases it is a field memory. This is because the memory capacity for a field image is only half that of a frame image, so it is inexpensive, and it is also possible to use a field memory for fast-moving images. The reason is that if you record with , there will be no pre-recording and you can keep it stationary.

When this field memory is used, there is a problem that image quality is degraded due to interpolation, and in particular, noise such as jitter is amplified by interpolation, resulting in a still image with even worse image quality.

[Object of the Invention] The present invention has been made by focusing on the problems of the above-mentioned conventional devices, and it is an object of the present invention to provide an image processing device that can easily and reliably eliminate noise such as jitter in still images. This is the purpose.

[Embodiments of the invention] First, the principle of the present invention will be explained.

In natural images, the boundary between one content and another (for example, the outline of a face and its background) has a large change in brightness. This can be easily understood if a color image is understood as a black and white image with only brightness. This is understandable.

From this, it can be seen that areas where a certain pixel in an image has a small difference in brightness and a small difference in color difference values between a certain pixel and its neighboring pixels are areas that belong to the same content (for example, a skin-colored part of a face).
On the other hand, a portion where the difference in luminance is large or the difference in one color difference value is thick can be considered to be a boundary portion between mutually different contents.

Next, the following can be said about the properties of jitter.

Jitter appears in the direction shown by the dotted line in FIG. 2 (that is, in the left-right direction). And when this jitter occurs, the luminance (
The change in color difference (R-Y, B-Y) is larger than the change in Y).

In an embodiment of the present invention, a pixel is accessed in the direction shown by the solid line in FIG.
, + , R-Yt,j, B-Yi, J).

The next pixel vertically adjacent to this pixel (Yi−+, j
, R-Yi, +, + , B-Yi, +, t). Here, Dir= aX I Yi,;-Yi-+, j1+
bX l (R-Yi, ) ) −
(R-Yi-+, j) 1+ cX l (B-Y
t, J) -CB-Yi-+, 4) 1.

Portions where the difference is less than a certain value are considered to belong to the same content in the image.

Perform this operation sequentially in the vertical direction.

In this case, for example, if pixels (Ln, J) to (Ik, J) shown in FIG. 3 have the same content.

The values of the color difference signals RY and BY of each pixel in this area are averaged.

The value of RY of pixel (f, J) is expressed as RYIn,J, the value of BY of that pixel is expressed as BYIn,J, and RY'=(
RYln, J+ RYll, 1. l + ・-+ R
Yn,, +)/(k-n+1) BY'=(BYIn, J + BY+., Hiroshi, J ten...÷ BY+b, J)/
(k-n+1), RY'' and BY' are the average color differences from pixel (In, J) to pixel (Ik, J), respectively.

Then, the color difference values from pixels (In, J) to (Ik, J) are changed to the above RY', BY'.

Through the above processing, the color differences in portions of the same content in the image are averaged to the same value, thereby making it possible to eliminate noise caused by changes in color differences due to jitter and the like.

FIG. 1 is a block diagram of a color video printer that is an embodiment of the present invention.

In this embodiment, A/D conversion circuits 11, 12.13 and Y
Memory 21, RY memory 22, and B-Y memory 23
, D/A conversion circuits 31, 32, and 33, a CPU 40, a conversion circuit 5o, a masking circuit 60, and a head driver 70.

The CPU 40 determines, for the plurality of pixels that the sum of the difference in luminance and the difference in color difference between pixels adjacent to each other in the vertical direction with respect to the screen, each multiplied by a predetermined coefficient, is continuously less than or equal to a predetermined value. This is an example of means for making the color difference values of those pixels almost the same.

Next, the operation of the above embodiment will be explained.

First, the video signal is transferred to the Y memory 21 and the R-Y memory 22.
, are recorded in the B-Y memory 23, respectively. Then, according to the printing order, capture one pixel at a time, Y, R-Y, B.
-Y signals are converted into C, M, and Y signals by a conversion circuit 50, masked by a masking circuit 60, and a head driver 70 drives the head to print.

In this case, between recording the video signal in the image memory and starting printing, the CPU 40 and the image memory 21.
22 and 23, noise cancellation processing is performed based on the above principle.

FIG. 5 is a flowchart showing the operation of noise cancellation processing in the above embodiment.

Here, the Y memory 21, the RY memory 22, and the BY memory 23 each have a size of 640
Y(I,
J), RY (1, J), BY (I, J). Note that J and I indicate the horizontal and vertical coordinates of the screen, respectively.

First, initial settings are performed (51, 52). B and E are the starting coordinates of the screen areas that display the same content, respectively,
Indicates the ending coordinate.

In S3, DY=a X l Y(I,J)-Y(I+1.J)
l +h X l RY(I,J)-RY(I+1.J
) l +b x l bY(I, J) -BY(I
+1. J) Find the distance 1tDY between pixels by calculating l. Here, a and b are weighting coefficients for brightness and color difference, and when brightness and color difference have the same value range and b is 1, a is preferably about 1 to 2.

When the value Y is less than or equal to the set value TI (33), E is incremented by 1 (SS). Note that the set value TH is a parameter indicating whether DY is large or small, and in the above embodiment, the luminance, When each color difference takes a value of 0 to 255 and a=b=1, the appropriate setting value TH is about 10 to 20.

When the value DY is larger than the set value TH, the memory RY(
Find the average value from memory BY (B, J) to BY (E, J), find the average value from memory BY (B, J) to BY (E, J) (S l l), and convert this average value to RY ( B, J
) to RY(E, J), BY(B, J) to BY(E, J) (S12), and
This operation is performed for all 480 pixels in the vertical direction of the screen (S7, S8), and also for all 640 pixels in the horizontal direction (S9.510).In this way, the color difference values of areas with the same content are calculated. The same value and the noise is eliminated.

[Effects of the Invention] According to the present invention, it is possible to easily and reliably eliminate noise such as jitter in still images.

BRIEF DESCRIPTION OF THE DRAWINGS FIG. 1 is a block diagram showing an embodiment of the present invention. FIG. 2 is a diagram illustrating directions on the screen. FIG. 3 is a diagram illustrating coordinate positions on the screen. FIG. 4 is an explanatory diagram of each memory used in the above embodiment. FIG. 5 is a flowchart showing the operation in the above embodiment. 21...Y memory, 22...RY memory, 23...B-Y memory, 50...conversion memory, 60...masking circuit, 70...head driver, 40... CPU.

Claims (1)

[Claims] In an image processing device having an image memory for still recording moving images, the difference in luminance and the difference in color difference between pixels adjacent to each other in the vertical direction with respect to the screen are each multiplied by a predetermined coefficient and added. For the plurality of pixels whose values are continuously equal to or less than a predetermined value,
An image processing device characterized in that the color difference values of those pixels are made almost the same.