JPS6342288A - Picture processing method for television picture in picture scanning recorder - Google Patents

Abstract

PURPOSE:To reduce noise or a dot interference in a still picture part and to eliminate a comb shape blurring or fading in a moving picture part by inputting picture element information of two frames, and deciding the moving picture or the still picture for every picture element to form recording information. CONSTITUTION:The respective corresponding picture elements of the successive two frames are compared, the moving picture or the still picture is judged, that decided to be the still picture, the average of the mutually corresponding picture elements of the two frames is taken and outputted as a recording picture element. When it is judged to be the moving picture, in the first frame (or the second frame), it is decided that the picture element part is the moving picture or not within the one frame. When it is decided to be the moving picture, if the picture element is in an odd number filed (or even number field), the picture element part is interpolated from the picture elements of the even number fields (or odde number fields) at both sides, an original picture element is removed and outputted at the recording picture element and when the picture element is in the even number field (or odd number field), it is directly outputted as the recording picture element.

Description

[Detailed Description of the Invention] [Field of Industrial Application] The present invention provides a method for recording each pixel while switching the information of 27]/- frame, 1 frame, and 1 field when copying a television image. The present invention relates to an image processing method.

[Background of the invention]

In the past (: L, TV images are moving images, so they are recorded as still images), 1 frame or interlace scanning is used to record 1n information from the 1 field in 15 formats. However, in the latter method, one frame consists of two fields, and
Because there is a time difference of 1/60 second between the fields:
:, 'kh's blank part - i,; + comb-shaped b1 between two fields.・is fJ7. It was

In this case, if there is movement in each pixel, 1.1- will not result in an accurate still image, so the brightness value (or density value) of each pixel will be compared to determine whether it is a moving image or a still image. Fu J7-
Image information of Mu! There is also a method of selecting the image information of the first image and recording it as an accurate still IP image.

On the other hand, TV broadcasting is based on a black and white image with a chroma signal superimposed on it.
The main interference could not be removed, and so-called dot interference was occurring.

As a result, colored areas, especially human skin, appear rough, degrading the image quality. An effective way to remove such dot interference is to take the average of continuous fume norms in a stationary scene, but 2. comb-like blurring or blurring due to movement occurs.

If we summarize the above-4-, we get the following table.

[Object of the invention] (2) The object of the present invention is to eliminate the drawbacks caused by the differences in the recording methods described above in 1, and to improve the still image portion.
: Images with less noise and dot interference, and video parts.
) allows you to record images without comb-like blur or blur.
- To provide an image processing method according to the present invention.
O [Structure of the Invention] To this end, the present invention provides an apparatus for obtaining image data for recording still images of television images.
Input the pixel information for 2 frames, judge whether each pixel is a video or still image, and based on the judgment result, select the information of 2 frames, 1 frame, and 1 field. By switching and recording information, 12 was configured to form an image.
1 [Example]
1 FIG. 1 is a flowchart 1 showing a method according to an embodiment of the present invention. In this embodiment, first, in step 1, corresponding pixels of two consecutive frames are compared to determine whether the image is a moving image or a still image.
This judgment is made as follows. First, the second
The data of 4F pixels in the vertical direction of the three adjacent i-lines (the even fields are shown as solid lines, and the upper odd fields are shown as broken lines) on the screen of the first f1 norm shown in Figure 111). 1, Xn s

Then, for pixel data X, , pixel data X n
−1ψ or X. 3. The chroma of 2 is inverted and the chroma of 2 is not added.
-1 view is Y... Add the inverted side of the chroma of 1 5o Then, if the following formula (1) is satisfied, it is determined to be a still image, otherwise it is a video! Two judgments. Is A a constant for comparison that is determined by taking into account the is margin? be.

1 (Xn+x, -, (or x,,,+) :l
-(Y, +Y,, -1 (or Y, 1=+) :l
l < A... (1) As a result of this judgment, it is determined to be a still image! -Machi no Q (5, step 1, in the next step 2, take the average of the ↑ corresponding pixels of 27[/-m) and output (7) as the recording pixel.

Also, if it is determined to be a video in step 1,
Proceeding to step 3, it is determined whether the pixel portion in the first frame (or second frame) is a moving image (comb-like blurring occurs) within that one frame (odd field and even field).

This determination is made as follows in the case of the first frame.

l Xn-+ Xn l <B, AND I
X,, Xn++ l < B... (2) or l X,, -X,, l≧B, AND I X, -
X, ,+l≧B (3) If the expression (2) or (3) is satisfied, the image is determined to be a still image, and if not, it is determined to be a moving image. B is also a comparison constant determined in consideration of the noise margin. If it is determined that the image is a still image, the pixels of that frame are output as they are as recording pixels in step 4.

If it is determined in step 3 that it is a moving image (presence exists), in step 5, if the pixel is in an odd field (or even field), interpolation is performed from the pixels in the even field (or odd field) on both sides. The original pixels are removed and output as recording pixels. Furthermore, if the pixel is in an even field (or an odd field), it is output as is as a recording pixel.

By selecting each pixel in this way, chroma interference can be significantly reduced and noise can be reduced to 1/21/2 since the average of two consecutive frames is taken for the still image. On the other hand, since the moving image portion is recorded in one frame or one field, a hard copy can be obtained with the loss of information kept to a minimum.

From the above, images created in this way are smooth, high-quality images with extremely low noise levels.

This processing reduces noise (including chroma interference), so even if edge enhancement is applied, noise remains emphasized, and the original video image is enhanced, making it extremely effective as a preprocessing for edge enhancement. .

If only the conventional method of detecting a comb-shaped part from one frame and determining it as a moving image was used, there would be many false positives and a considerable amount of information would be lost.However, in the case of the method of this embodiment, the Since the motion is detected in two frames, the number of false judgments is greatly reduced. In particular, if there is a stationary comb-shaped part in the image itself, conventional methods would judge it as a moving image, but the method of this embodiment definitely judges it as a still image.

FIG. 3 shows a determination circuit constructed from hard logic to implement the above-described method. 10 is an A/D converter that converts a BGR or luminance signal input from a video camera or the like and extracts a digital pixel signal; 11 is an A/D converter that converts a BGR or luminance signal input from a video camera or the like to extract a digital pixel signal; ), and 12 is a second frame memory that stores similar data of the next frame.

13 to 15 sequentially read line data in the vertical direction of the first frame read from the first frame memory 11 (Xn-1% The first to third launches are transferred and latched, and the latches 13 to 15 latch adjacent pixel data along the vertical direction of the first frame.

16 to 18 are fourth to sixth launches that sequentially transfer and latch the vertical line data of the second frame read from the second frame memory 12 every time a clock arrives;
These latches 16 to 18 latch adjacent pixel data along the vertical direction of the second frame.

19 is a first selector for selecting one of the output pixel data of the first latch 13 and the third latch 15; 20 is a second selector for selecting one of the output pixel data of the fourth latch 16 and the sixth latch 18; be.

21 is a first adder that adds the output pixel data of the first selector 19 and the output pixel data of the second latch 14; 22 is the output pixel data of the first selector 20 and the fifth latch 1;
This is a second adder that calculates tjn of the output pixel data of 7.

23 is an arithmetic unit that subtracts the data obtained by the -1 second adder 21 and 22; It is a comparator that compares the

26 is the output pixel data of the second latch 14 and the fifth latch 1
This is an arithmetic circuit that performs an averaging operation with the output pixel data from 7, and its output ii is output as a recording data via a buffer 27 whose gate is controlled by a Jz comparator 24. It looks like this.

On the other hand, 28 is a subtracter for calculating the difference between the output pixel data from the first latch 13 and the second latch 14, and 29 is a subtracter for calculating the difference between the output pixel data from the second latch 14 and the third latch 15. .

In addition, 30 and 31 are data B from the constant setter 32, respectively.
and a comparator that compares the data from the subtractor 28.29,
33 is to exclusive OR 7'y'-1 which calculates the exclusive OR of the data from both comparators 30 and 31.

34 is the above 1. , , ,'l-Comparator 24 outputs data and field signals (odd number - 11', even number - [o
J) manually operated NanF date, 35 is the second Nando)
This is an ant game 1 which inputs the output from the knee 134 and the output from the Juki data 1 and 33.

3G is a buffer for inputting pixel data from the second latch 14, and its gate is controlled by the data 1-35Q;-.

, 37 is an arithmetic unit that averages the output pixel data of the first latch 13 (!': the output pixel data of the third latch 15, and the pixel data from there (!') It is designed to be output as recording data via the Noah Gate 38.
and 1- input the output from the comparator 24.

Now, the vertical pixel data of the 1st] 1 norm is latched as shown in Figure 4 at the -11th to 3rd 1:1-15, and the vertical pixel data of the 1st frame is The data also
The fourth--sixth latches 16-18 are similarly launched.

The pixel data output from the first latch 13 and the second latch 15 whose chroma phase is inverted is selected by the first selector 19 and added to the pixel data output from the second latch 14 by the adder 21. Ru.

Furthermore, the pixel data output from the fourth latch 15 and the sixth latch 17 whose chroma phase is inverted is selected by the second selector 20 and is combined with the pixel data output from the fifth latch 18 by the adder 22. will be added.

Then, the difference between the data obtained by both the adders 21 and 22 is processed by the arithmetic unit 23, and this difference is compared with the value A set by the constant setter 25 by the comparator 24. As a result of the comparison, if the distribution is smaller than the set value A, the above formula (1) is satisfied, the 'IJ outputs the two image parts as still images, and the above difference is the set value A. If it is larger than , it is assumed that it is a moving image and 'OJ' is output.

'When IJ outputs (a still image between frames), the buffer 27 opens the gate, so the arithmetic unit 2 averages the pixel data obtained by the second latch 14 and the fourth latch 17.
The output of G, that is, the average pixel data of the first frame and the second frame is output as m-image data for recording.

'If OJ is output 1 or more (video between frames), subtracters 28, 29, comparator 30. ．． 31. The fields of the first frame are compared in the constant B setter 32 and the gate 33, and if the above-mentioned step (2) or (3) is followed, it is determined that it is a still image. '
When the buffer 3G is open, the pixel data of the second latch 14 (pixel data of the first frame) is output as recording data. This judgment is shown in the video (Day I.33
Even if it is determined that the output is "oj", the output of days 1-34 will be 'IJ' in the case of an even field, so the second lunch 140 pixel data will always be output as recording data.

Next, 1. If none of the above applies, that is, if the judgment between two frames is "movie", the field signal is an odd number, and the judgment between the fields of the first frame is "movie" (,1
1. The output of the gate 38 becomes 'IJ', and the output pixel data of -13 and 15 are averaged by the calculator 37, that is, 14 data are canceled, and the interpolated data j-2 is output as the pixel data t,2 by interpolating the pixel data of the 10-year old pixel data for both of the data.

〔Effect of the invention〕

Therefore, according to the present invention, the claw interference in the still image portion is significantly reduced. The noise can be reduced to l/21/2. In addition, since the video portion is recorded in 1, 1) [, -m or 1 field, the loss of information is minimized. '.

You can suppress it.

[Brief explanation of drawings]

Figure 1 is a flowchart 1-1 for explaining an embodiment of the present invention. Figure 2 is an explanatory diagram for still image/video determination.
The figure is a block diagram of an example for implementing the method of this embodiment, and FIG. 4 is an explanatory diagram of the launch operation in FIG. 3. Agent Patent Attorney Tsuneaki Nagao Figure 11, Figure 4

Claims (3)

[Claims] (1) A device that obtains image data for recording still images of television images inputs pixel information for two frames, determines whether each pixel is a moving image or a still image, and based on the result. An image processing method characterized by selectively switching information of two frames, one frame, and one field to form an image as recorded information. (2) If it is determined that it is a still image by comparing the pixel information of two frames, the average value of both frames will be recorded image information, and if it is determined that it is a moving image, the odd number of one frame,
Compare both pixels of an even field, and if the result of this comparison determines that it is a still image, the pixel information of the frame is used as recording information, and if the latter comparison result determines that it is a moving image, one field is recorded on both sides. 2. The image processing method according to claim 1, wherein pixel information interpolated from images of fields before and after is used as recorded information, or pixel information of the other field is used as recorded information as is. (3) The image processing method according to claim 2, wherein the comparison standard for the former comparison or the latter comparison takes into account a noise margin.