JPH0779459B2 - Motion compensation device - Google Patents

Description

Description: TECHNICAL FIELD The present invention relates to a motion correction device applied to field number conversion processing in television format conversion, for example.

[Outline of Invention]

The present invention detects a motion vector by statistically processing the difference value of the luminance values of consecutive fields of an input video signal, and forms a motion-corrected image by the motion vector. Values are aggregated for all screens, a plurality of motion vectors corresponding to a plurality of values are detected from the smallest summed difference value, and a plurality of motion correction images are formed by each of the detected plurality of motion vectors. However, by selecting the motion-correction image with the smallest correction error among the plurality of motion-correction images, a good motion-correction image can be formed even if there are multiple motions in the same screen. Is.

[Conventional technology]

The motion correction is used in a noise reduction device as disclosed in, for example, Japanese Patent Laid-Open No. 57-193179. Also every second
By converting motion when converting a 60-field television signal into a 50-field television signal per second, an output image with smooth motion on the screen can be obtained.

FIG. 6 shows an example of a conventional system converter. This equipment is a high definition television system of so-called HD system (1125 lines / 60 fields) and PAL of (625 lines / 50 fields).
Convert to method. In FIG. 6, the digitized luminance signal in the high definition television signal is supplied to the input terminal indicated by 51. This digital luminance signal is supplied to the line number conversion circuit 52. The number of lines conversion circuit 52 performs deinterlacing processing and the number of lines is 11
Converted from 25 to 625.

The output signal (625 lines / 60 frames) of the line number conversion circuit 52 is written in the field memory 53. A digital video signal of two continuous fields from the input side and the output side of the field memory 53 is a motion vector detection circuit.
Supplied to 54. The motion vector detection circuit 54 calculates the difference value between the brightness value of the representative point of the previous field and the brightness value of the pixel of the current field for each detection area formed by subdividing one screen, and totals the absolute value of this difference value. However, the minimum value of the aggregated difference data is detected as a motion vector.
This motion vector is supplied to the motion correction circuit 55.

Both the previous field digital video signal from the field memory 53 and the input current field digital video signal are supplied to the motion correction circuit 55 and the linear interpolation circuit 59. The motion compensation circuit 55 forms a motion compensation image by synthesizing the signal obtained by shifting the image of the previous field by the shift command based on the motion vector and the signal obtained by shifting the image of the current field by the shift command based on the motion vector.

The linear interpolation circuit 59 is an interpolation circuit that multiplies each of continuous two fields of video by a predetermined weighting coefficient and adds the multiplication outputs. Output signal of motion compensation circuit 55 and linear interpolation circuit
The output signal of 59 is supplied to the selection circuit 61, and one output signal is selected. A selection signal generation circuit 64 is provided to control the selection circuit 61. This selection signal generation circuit 64
A differential signal is supplied to each of the motion correction circuit 55 and the linear interpolation circuit 59. These difference signals are the difference signals of the signals after the motion correction of the previous field and the current field of the motion correction circuit 55, and the difference signals of the signals of the previous field and the current field, respectively. A selection signal for selecting a correction signal corresponding to a smaller level difference signal between the two difference signals in pixel units is formed.

The motion compensation output signal or the linear interpolation output signal selected by the selection circuit 61 is supplied to the frame number conversion circuit 67, and the output terminal 68 of the frame number conversion circuit 67 (625 lines / 50 fields) PAL digital luminance signal. Is obtained.

FIG. 7 shows an example of the calculated total difference data A (that is, the representative point difference data is converted into an absolute value and totalized over the entire screen) in the motion vector detection circuit 54. Actually, the aggregated difference data A is a component in the z-axis direction with respect to the two-dimensional detection area and is three-dimensional data. However, for simplification, FIG. 7 shows the aggregated difference data A by focusing on, for example, only the lateral direction (x direction) of the detection area. In the distribution of the aggregated difference data A, the minimum value is detected as the x component of v of the motion vector.

The conventional motion compensation device detects one motion vector for one screen, and cannot effectively perform motion compensation for an image having a plurality of motions in one screen. was there. For example, as shown in FIG. 8, when two moving objects O1 and O2 are moving in opposite directions in one screen and their motion vectors are v1 and v2, respectively, the motion vector v1 which is relatively easy to detect is It is detected, and motion compensation is performed by this motion vector v1. Therefore, a moving object
At O2, the output signal of the linear interpolation circuit 59 is selected, and as a result, the output image becomes unnatural with jerkiness.

As a motion compensating device for solving such a problem, a device shown in FIG. 9 has been proposed which divides a screen into four equal parts and detects a motion vector for each part. The video signal of the previous field and the video signal of the current field formed by the field memory 53 are supplied to the motion vector detection circuit 54, and the motion vector v1, v2 of each part formed by dividing one screen into four equal parts. v3 and v4 are detected.

In addition, motion compensation circuits 55 and 5 that perform motion compensation for each part of the
6,57,58 are provided. The motion vector v1 is supplied to the motion correction circuit 55, and the motion vectors v2, v3, v4 are supplied to the other motion correction circuits 56, 57, 58, respectively. A switch circuit 62 for selecting one output signal among the output signals of the motion correction circuits 55 to 58 on a pixel-by-pixel basis and one of the output signal of the signal selected by the switch circuit 62 and the linear interpolation circuit 59. A switch circuit 63 for selecting is provided. A selection circuit 61 including the switch circuit 62 and the switch circuit 63.
A selection signal generating circuit 64 generates a selection signal for controlling the. One switch circuit 62 of the selection circuit 61 selects the signal with the smallest correction error among the four motion-corrected images. The other switch circuit 63 selects a signal having a smaller error between the motion-compensated signal and the output signal of the linear interpolation circuit 59.

In the motion compensation device described above, the motion vector detection circuit 54
Motion vector v1, v2, for each area of one screen detected by
Examples of v3 and v4 are shown in FIGS. 10A, 10B, 10C and 10D.
, Respectively. The motion correction apparatus shown in FIG. 9 can correct an image having up to four different motion vectors. However, even when there are a plurality of moving objects in the same area among the four areas, the motion correction is performed only for the dominant one moving object, and the motion correction is not performed for the other moving objects. The output image is unnatural.

Therefore, an object of the present invention is to provide a motion correction device that can perform good motion correction even on an input image having a plurality of different motions in one screen.

[Means for solving problems]

The present invention relates to a motion correction apparatus configured to detect a motion vector by statistically processing a difference value between luminance values of two consecutive frames of an input video signal, and form a motion corrected image by the motion vector, Difference values are aggregated for all screens, and multiple motion vectors v1, v2, v3, v4 corresponding to multiple values from the smallest aggregated difference value, respectively.
A motion vector detection circuit 4 for detecting the motion, motion correction circuits 5, 6, 7, 8 for forming a plurality of motion corrected images by the detected motion vectors v1, v2, v3, v4, respectively, and a plurality of motions. This is a motion correction device that includes a switch circuit 12 that selects a motion correction image with a minimum correction error among the correction images.

[Action]

The motion vector detection circuit 4 detects, for example, four motion vectors of a plurality of moving objects in one screen. Using each of these motion vectors v1, v2, v3, v4, a motion compensation circuit
5,6,7,8 form a motion compensated image. The switch circuit 12 selects the motion-corrected image with the minimum correction error.
Therefore, since each moving object is shifted by its own motion vector, good motion compensation is achieved.

〔Example〕

An embodiment of the present invention will be described below with reference to the drawings. The description of this embodiment will be made in the following order.

a. Overall configuration of motion compensation device b. Specific configuration of motion compensation circuit and selection signal generation circuit c. Motion compensation operation a. Overall configuration of motion compensation device This embodiment is (1125 lines / 60 fields). The present invention is applied to the conversion of the number of fields when converting the so-called HD system of the high definition television system of (2) to the PAL system of (625 lines / 50 fields). Figure 1 shows this (HD →
The overall structure of the PAL) method converter is shown.

In FIG. 1, a luminance signal in a high-definition television signal is converted into a digital signal and supplied to an input terminal indicated by 1. This digital video signal is supplied to the line number conversion circuit 2. The line number conversion circuit 2 performs line number conversion and non-interlacing processing. The number of lines is
Converted from 1125 to 625 using digital frequency conversion technology. A digital video signal of (625 lines / 60 frames) is obtained from the line number conversion circuit 2.

In practice, the line number conversion circuit 2 allows the number of lines from the video of one field of the high definition television signal to be 625 /
Two 1st field images and 2nd field images are simultaneously formed. The video of the first field and the video of the second field are offset by 0.5 line. Therefore, from the line number conversion circuit 2, (625/2
A digital video signal including only the first field of (line / 60 fields) and a digital video signal including only the second field of (625/2 lines / 60 fields) are obtained. The motion vector detection is performed using the video signal of the first field, and the conversion process of the number of fields is processed separately for each of the first field and the second field. However, in the following description, the processing of the first field and the second field is not particularly divided for simplification.

The output signal of the line number conversion circuit 2 is written in the field memory 3. The digital video signal of the previous field is read from the field memory 3. The current field digital video signal and the previous field digital video signal are supplied to the motion vector detection circuit 4, the four motion correction circuits 5, 6, 7, 8 and the linear interpolation circuit 9, respectively.

The motion vector detection circuit 4 calculates the difference (field difference) between the brightness value of the representative point of the previous field and the brightness value of the pixel of the current field for each detection area obtained by subdividing one screen, and the field difference The configuration is such that four motion vectors respectively corresponding to the minimum value of the total difference data obtained by totaling the absolute values and the three smaller values from the minimum value are detected. This detected motion vector v1, v
2, v3 and v4 are supplied to the motion correction circuits 5, 6, 7 and 8, respectively.

The motion correction circuit 5 shifts the image of the previous field and the image of the current field according to the shift signal formed from the motion vector v1. Similarly, each of the motion correction circuits 6, 7, and 8 shifts the image of the previous field and the image of the current field according to the motion vectors v2, v3, and v4. The motion correction circuits 5, 6, 7, and 8 increase the motion (indicated by the motion vector) in the original image by 20% in order to convert 6 images into 5 images. Form a corrected image. In other words, when converting six input images F1 to F6 containing a moving object having a constant velocity motion that moves by v every (1/60) seconds into five images f1 to f5, The image obtained by shifting F1 by (0) v and the image obtained by shifting image F2 by (-1) v are added, and the addition output is halved to form image f1. Thereafter, similarly, an image f2 is formed from the image F2 shifted by (1/5) V and the image F3 shifted by (-4/5) V. Image f from image F3 shifted by (2/5) V and image F4 shifted by (-3/5) V
3 is formed. An image f4 is formed by shifting the image F4 by (3/5) V and shifting the image F5 by (-2/5) V. Image F5 (4/5) V and image F6 (-1 /
5) An image f5 is formed from the V-shifted image.

The output signals of the above-described motion correction circuits 5, 6, 7, and 8 are supplied to the selection circuit 11, and the switch circuit 12 of the selection circuit 11 selects the signal (luminance value) with the smallest correction error. In the case of panning or tilting, the motion correction can completely convert the number of fields if the motion vector can be accurately detected. However, in an actual image, there are portions that have various movements even within one screen, and there are portions that are stationary, and there are portions that cause problems due to shifting. In such a case, the output signal of the linear interpolation circuit 9 is selected by the switch circuit 13 of the selection circuit 11 instead of the output signal of the motion correction circuit 5, 6, 7 or 8.

The linear interpolation circuit 9 is configured as an interpolation circuit that multiplies each of continuous two field images by a predetermined weighting factor and adds the multiplication outputs. Similar to the above, when converting 6 consecutive images F1 to F6 every (1/60) seconds into 5 images f1 to f5, the image F1 is multiplied by the weighting factor of (1)
An image f1 is formed by adding F2 multiplied by a weighting coefficient (0). Further, the image f2 is formed by adding the image F2 multiplied by the weighting factor of (4/5) and the image F3 multiplied by the weighting factor of (1/5). Similarly, (3 / 5F3 + 2 / 5F4) forms image f3, (2 / 5F4 + 3 / 5F5) forms image f4, and 1 / 5F5
An image f5 is formed by + 4 / 5F6.

The linear interpolation circuit 9 has a problem that the moving part is doubled or unraveled, but the stationary part
Linear interpolation is better than motion compensation.

The switch circuits 12 and 13 of the selection circuit 11 are controlled by the selection signals generated at the output terminals 15 and 16 of the selection signal generation circuit 14, respectively, and the selection circuit 11 outputs the signal with the smallest correction error. The output signal of the selection circuit 11 is supplied to the frame number conversion circuit 17. This frame number conversion circuit 17
Alternately selects the digital video signal of the first field and the digital video signal of the second field converted to (625/2 lines / 50 fields). (625 lines /
A digital video signal of 50 fields) is taken out to the output terminal 18.

This digital luminance signal is supplied to the PAL color encoder via a D / A converter and a low pass filter (not shown). Similarly to the luminance signal, the red color difference signal RY and the blue color difference signal BY subjected to the conversion of the number of lines and the conversion of the number of fields are supplied to the PAL color encoder. Therefore, a PAL system composite color television signal can be obtained at the output terminal of the PAL color encoder.

b. Specific Configuration of Motion Correction Circuit and Selection Signal Generation Circuit FIG. 2 shows a specific configuration of the motion correction circuit 5 and selection signal generation circuit 14. The digital video signal of the previous field from the field memory 3 is supplied to the shift circuit 21, and the digital video signal of the current field from the line number conversion circuit 2 is supplied to the shift circuit 22. A shift signal is supplied from the shift signal generation circuit 23 to the shift circuits 21 and 22. The shift signal is formed from the motion vector v1, and the signals shifted by the shift circuits 21 and 22 are added by the adder circuit 25. Shift circuit 21 and
As described above, 22 is for shifting the luminance value of the image F1 in the previous field by (1/5) v1 and shifting the corresponding luminance value of the image F2 in the current field by (-4/5) v1. . The output signal of the adder circuit 25 is halved by the shift register 26 and taken out to the output terminal 27.

In the motion correction circuit 5, the output signal of the shift circuit 22 is subtracted from the output signal of the shift circuit 21 by the subtraction circuit 28. The output signal of the subtraction circuit 28 represents the correction error when the motion correction is performed using the motion vector v1. This correction error signal is supplied to the low pass filter 29 of the selection signal generation circuit 14.

The output signal of the low-pass filter 29 is converted into an absolute value by the absolute value conversion circuit 30 and supplied to the minimum value detection circuit 40.
The motion correction circuits 6, 7 and 8 are configured similarly to the motion correction circuit 5, and each motion correction circuit forms a motion-corrected output signal and a correction error signal. This correction error signal is supplied to the low-pass filters 34, 35, 36 of the selection signal generation circuit 14 from the terminals 31, 32, 33. Low pass filter 34,35,
The respective output signals of 36 are supplied to the minimum value detection circuit 40 via the absolute value conversion circuits 37, 38, 39.

The minimum value detection circuit 40 detects the motion correction with the minimum correction error, and outputs a 2-bit output signal, for example. For example, the motion correction of the motion correction circuit 5 is (00), the motion correction of the motion correction circuit 6 is (01), the motion correction of the motion correction circuit 7 is (10), and the motion correction of the motion correction circuit 8 is (11). To be done. The output signal of this minimum value detection circuit 40 is an area filter.
It is taken out to the output terminal 15 via 41 as a selection signal.
The selection circuit controls the switch circuit 12 of the selection circuit 11 to select the motion correction signal with the minimum correction error.

Further, the correction error signals of the motion correction circuits 5 to 8 are supplied to the input terminals of the switch circuit 42, respectively. This switch circuit 42
Is controlled by the output signal of the minimum value detection circuit 40, and the minimum correction error signal is selected. The correction error signal selected by the switch circuit 42 is supplied to the comparison circuit 43.

The output signal of the previous field and the signal of the current field are supplied from the field memory 3 to the input terminals indicated by 44 and 45, respectively, and the subtraction circuit 46 obtains the difference signal between them. This difference signal is converted into an absolute value by the absolute value conversion circuit 47 and then supplied to the comparison circuit 43. The difference signal becomes small in the case of the brightness value of the static region. Therefore, the comparison circuit 43
By discriminating between the output signal of the switch circuit 42 and the output signal of the absolute value conversion circuit 47 on the side with a smaller level, a 1-bit selection signal for selecting one of the motion correction and the linear interpolation is output terminal 16. Occurs in. The switch signal 13 of the selection circuit 11 is controlled by this selection signal.

c. Motion Compensation Operation Motion compensation according to an embodiment of the present invention will be described by taking the input image shown in FIG. 3 as an example. This input image is a moving object O1, which has different motion vectors v1, v2, v3, v4.
Contains O2, O3, O4 (shown by hatching). Since the motion vector detection circuit collects the difference data over the entire screen of this input image, the horizontal direction (x direction)
Only, the aggregated difference data A represented simply is as shown in FIG.

The four poles in this aggregated difference data A are motion vectors v1, v
It is detected as 2, v3, v4 respectively. To find the pole
As shown in FIG. 5, three consecutive values A _i−1 , A _i and A _{i + 1} in the aggregated difference data A are taken out, and a value whose center point is always smaller than both ends is searched for, If there is data that satisfies this condition, the address and value of this data are stored in the memory. Four sets of memory for storage are prepared. When four or more data satisfying the above conditions can be found, the largest value among the data already stored in the memory is compared with the value of the newly found data. According to this comparison, when the value of the new data is smaller than the stored data, the stored data is replaced with the new data. By performing the above processing for all the aggregated difference data, four small values can be obtained from the minimum data. Such motion vector detection processing is
The same is done in the vertical direction.

The detected motion vector is supplied to the motion correction circuits 5, 6, 7, and 8, respectively. Each motion correction circuit performs motion correction corresponding to one motion vector on the entire screen. The selection signals from the selection signal generation circuit 14 control the switch circuits 12 and 13 of the selection circuit 11 for each image. In the case of the input image shown in FIG. 3, the luminance values of the pixels of the moving objects O1 to O4 are motion-corrected by the motion vectors v1 to v4, and the output signal of the linear interpolation circuit 9 is selected in the stationary area. It

The number of motion correction circuits is not limited to four, but may be two or more.

〔The invention's effect〕

According to the present invention, a plurality of motions are different from the case where one motion vector is detected in one screen or one part obtained by dividing one screen and motion correction is performed using this detected motion vector. Since a plurality of motion corrections based on the vector are performed and the one having the smallest correction error is selected from the plurality of motion corrections, the motion correction can be favorably performed when a plurality of moving objects are included in the same screen.

[Brief description of drawings]

1 is a block diagram of an embodiment of the present invention, FIG. 2 is a block diagram of an example of a motion correction circuit and a selection signal generating circuit in an embodiment of the present invention, and FIG. 3 is a block diagram of an embodiment of the present invention. A schematic diagram of an example of an input image used for the explanation, FIGS. 4 and 5 are schematic diagrams used for explaining the detection of a motion vector in an embodiment of the present invention, and FIG. 6 is an example of a conventional motion correction apparatus. FIG. 7, FIG. 7 and FIG. 8 are schematic diagrams used for explaining the operation of the conventional motion compensation device, FIG. 9 is a block diagram of another example of the conventional motion compensation device, and FIG. It is an approximate line figure using explanation of another example of a amendment device. Description of main symbols in the drawing 1: Digital video signal input terminal, 3: Field memory, 4: Motion vector detection circuit, 5, 6, 7, 8: Motion correction circuit,
9: Linear interpolation circuit, 11: Selection circuit, 14: Selection signal generation circuit.

 ─────────────────────────────────────────────────── ─── Continuation of the front page (72) Inventor Kunio Matsumoto 6-735 Kita-Shinagawa, Shinagawa-ku, Tokyo Sony Corporation (72) Inventor Yutaka Tanaka 1-1-10 Kinuta, Setagaya-ku, Tokyo Japan Within the Broadcasting Corporation Broadcasting Technology Laboratory (72) Inventor Toshiro Omura 1-10-11 Kinuta, Setagaya-ku, Tokyo Inside the Broadcasting Technology Laboratory Japan Broadcasting Association (72) Inventor Taiichi Kurita 1-10-10 Kinuta, Setagaya-ku, Tokyo No. 11 Inside Broadcasting Technology Research Institute of Japan Broadcasting Corporation (72) Inventor Seiichi Koshi 1-10-11 Kinuta, Setagaya-ku, Tokyo Inside Broadcasting Research Institute of Japan Broadcasting Association (72) Inventor Yoshimichi Otsuka Kinuta Setagaya-ku, Tokyo 1-10-11 Broadcasting Technology Institute of Japan Broadcasting Corporation (72) Inventor Taiji Nishizawa 1-10-11 Kinuta, Setagaya-ku, Tokyo Broadcasting Engineering Laboratory of Japan Broadcasting Corporation (72) Invention Yuichi Ninomiya 1-10-11 Kinuta, Setagaya-ku, Tokyo Within the Institute of Broadcasting Technology, Japan Broadcasting Corporation (56) Reference JP-A-61-159880 (JP, A)

Claims (1)

[Claims] 1. A motion vector is detected by statistically processing the difference value of the luminance values of the n-th and (n + 1) th fields of the input video signal, which are continuous in time, and the motion-corrected image is detected by the motion vector. In the motion correction apparatus configured to form the above, the difference values of the brightness values of the n-th and (n + 1) th fields are totaled for all screens, and the minimum value of the totaled difference values and the smaller m in order from this minimum value. A motion vector detection circuit that detects (m + 1) motion vectors corresponding to the respective difference values; and a motion correction circuit that forms a plurality of motion corrected images from each of the plurality of detected motion vectors, And a selection circuit for selecting a motion correction image having a minimum correction error among the plurality of motion correction images.