JP3011739B2 - Television signal interpolation method - Google Patents

Description

Description: BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a television signal interpolation method that involves a motion correction process using a motion vector.

[Prior Art] The technique of performing motion compensation using a motion vector is used to improve the inter-frame coding efficiency in high-efficiency coding of a television signal, or to perform motion compensation by changing the number of fields in television format conversion. Used to reduce continuity.

The detection of the motion vector is performed by converting the television signal into m pixels ×
It is divided into blocks of n lines (m and n are integers), and the detection method is a pattern matching method (Japanese Patent Application Laid-Open No. 55-162683 and 162684) and a repetitive gradient method (Japanese Patent Application Laid-Open
No. 60-158786) is well known.

Next, as an example of an interpolation process using a motion vector, a case where a television system is converted will be described.

For example, in the television system conversion between the PAL system and the NTSC system, mutual conversion between 60 fields and 50 fields is required, and this conversion process is performed using a field memory in units of conversion between 6 fields and 5 fields. In the conversion for increasing the field, some field television signals (hereinafter simply referred to as field signals)
In the conversion for reducing the field, a part of the field signal is skipped. For this reason,
Field discontinuity occurs at repetitions and jump points, which results in discontinuity of the moving image. In order to correct this, a signal of two fields before conversion is generally subjected to a weight addition (linear interpolation) process to form a field signal after conversion.

FIG. 2 shows a linear interpolation process. The signals ca and CB separated by one field are added to the field interpolation ratio α,
(1-α) is added after the load to obtain an interpolation signal cc.
Here, the field interpolation ratio α is a constant determined from the positional relationship on the time axis between the pre-conversion signals ca and cb and the post-conversion signal cc.

However, this linear interpolation process has a problem with moving images. FIG. 3 shows linear interpolation for a moving image (movement of a circle), and similarly shows a case where an interpolation signal PC is obtained from pre-conversion signals PA and PB separated by one field. In addition, while FIG. 2 shows a pixel,
FIG. 3 shows an entire one-field image.

The area of the sloping part in the interpolation signal PC is obtained by adding the weight of the uncorresponding background part and the moving image part, and the level changes with the field interpolation ratio, which is the cause of jerkiness. Has become.

The interpolation method using a motion vector is for reducing the occurrence of this jerkiness, and FIG. 4 corresponding to FIG. 3 shows the basic concept. Before conversion P
A and PB are used to detect a motion vector of a circle as a moving object, and a motion correction for the moving object is performed by using a value αV obtained by multiplying the motion vector V by a field interpolation ratio α to generate an interpolation signal PD. It has gained.

However, it is difficult for the detected motion vector to capture 100% of the motion of all the moving images accurately, and in some cases, an error may occur. The following may not be performed. For this reason, both signals do not safely overlap in the interpolation processing through the motion correction using the value αV, and in the practical motion interpolation processing, the load addition processing using the field interpolation ratio is further performed on the motion-corrected signal. Has been given.

FIG. 5 shows the configuration of a motion interpolation section in a conventional television system conversion apparatus which performs a linear value interpolation process after performing motion correction on each field signal (19).
1989 National Convention of the Institute of Television Engineers of Japan, Proceedings 20-5, "Motion vector detection and motion interpolation of TV system converter", and JP-A-1-309597).

In FIG. 5, a current field signal S1 and a previous field signal S2 are supplied to a motion vector detection circuit 1. The motion vector detection circuit 1 uses these signals S1 and S2 to detect a motion vector for each block obtained by dividing one field into a plurality. The motion vector S3 (V) is input to the motion vector correction circuit 2 and the adaptive motion interpolation switching control circuit 3.

The motion vector correction circuit 2 corrects the motion vector V with the field interpolation ratio α, and stores the corrected motion vectors S4 (αV) and S5 {(1-α) V} in the corresponding motion correction memories 4 Give 5

Each of the motion correction memories 4 and 5 stores a correction motion vector α
V, the signal S obtained by shifting the coordinates of the target block of the previous field signal S2 or the current field signal S1 by (1-α) V.
6. Form and output S7. The difference between these signals S6 and S7 is obtained by a subtraction circuit 6 for each pixel, and the difference is converted into an absolute value by an absolute value conversion circuit 10, and then the above-described adaptive motion interpolation switching as a motion correction inter-field difference signal S8 is performed. It is provided to the control circuit 3.

The output signals S6 and S7 from the memories 4 and 5 are given to the corresponding multiplication circuits 8 and 9, and after the multiplication circuits 8 and 9 load the field interpolation ratio α and (1−α), The addition is performed by the addition circuit 10 to obtain a motion compensation field interpolation signal S9. The motion compensation field interpolation signal S9 is provided to the multiplication circuit 11.

The current field signal S1 and the previous field signal S2 are also provided to the subtraction circuit 12. The difference for each pixel between these field signals S1 and S2 is obtained by a subtraction circuit 12, and this difference is converted to an absolute value by an absolute value conversion circuit 13, and the adaptive motion interpolation switching described above as an uncorrected inter-field difference signal S10 is performed. It is provided to the control circuit 3.

Further, the previous field signal S2 and the current field signal S1 are supplied to the corresponding multiplication circuits 14 and 15, and the respective multiplication circuits 14 and 15
After that, the field interpolation ratio α, (1−α) is weighted, and then added by the addition circuit 16 to form a linear interpolation signal S11. The linear interpolation signal S11 is provided to the multiplication circuit 17.

The adaptive motion interpolation switching control circuit 3 outputs a motion compensation field interpolation signal S9 or a linear interpolation signal S1 as a motion interpolation signal.
This is a circuit for determining whether to output 1 or to output both weighted addition signals, and determines the magnitude of the motion vector V, the motion-compensated inter-field difference signal S8, and the uncorrected inter-field difference signal S10 as parameters. , And outputs the adaptive interpolation switching signals S12 (β) and S13 (1-β) to the corresponding multiplication circuits 11 and 17 in pixel units.

For example, the adaptive motion interpolation switching control circuit 3 determines that the motion-compensated inter-field difference signal S8 is larger than the inter-field difference signal S10 without motion compensation, the motion vector V is not 0, and the motion-compensated inter-field difference signal S8 is smaller than the threshold value. When it is smaller, one adaptive interpolation switching signal β is set to 1.

Thus, the multiplication circuit 11 outputs a signal obtained by multiplying the motion compensation field interpolation signal S9 by β, the multiplication circuit 17 outputs a signal obtained by multiplying the linear interpolation signal S11 by (1−β), and these signals are added to the addition circuit 18 Are added to form a final motion interpolation signal S14.

[Problems to be Solved by the Invention] However, in the above-described adaptive switching control in the conventional device, since the control is performed also using the motion-compensated inter-field difference signal S8 as a parameter, a switching error occurs in the following points.

(1) When the minimum unit of the motion correction using the motion vector detection is a pixel, for a signal having a sharp edge, even if the motion vector can be accurately detected and the motion correction is correctly performed,
A difference between fields occurs at the edge. Originally, when the motion vector accurately captures the motion and the motion correction is completely corrected, the motion-compensated inter-field difference signal should be zero. Therefore, in the edge portion, the linear interpolation signal S11 is
Is easy to select, so that image distortion occurs at the edge portion.

(2) When a moving image has jitter, it may not be possible to accurately detect a motion vector. However, even if the motion vector is still inaccurate from the linear interpolation signal S11, motion compensation using the motion vector is performed. In some cases, the field interpolation signal S9 has less image distortion. At this time, the motion compensation inter-field difference signal S8 in the flat portion of the image does not increase so much, but the difference increases in the steep edge portion, so that the linear interpolation signal S11 can be easily selected, and the image distortion in the edge portion of the image is reduced. May occur.

The present invention has been made in view of the above points,
An object of the present invention is to provide a television signal interpolation system capable of reducing the occurrence of an error in adaptive interpolation switching at an edge portion of an image and reducing image distortion.

[Means for Solving the Problems] In order to solve the problems, in the present invention, two or more field (or frame) signals of a digitized television signal are interpolated to form a new field (or frame). A frame (or frame) signal is generated by adaptively converting a motion-compensated field (or frame) interpolation signal obtained by motion compensation using a motion vector and a non-corrected field (or frame) interpolation signal obtained by linear interpolation. In addition, switching or weight addition is performed, and at least the difference signal between the motion-compensated fields (or between frames) and the difference between the motion-uncompensated fields (between frames) are output as parameters for the switching control. The following circuits are newly provided in a television signal interpolation system using signals.

That is, a first edge detection circuit that performs edge detection on one of the motion-compensated field signals, and a difference between motion-compensated fields (or between frames) according to the detection value of the first edge detection circuit. A first attenuating circuit for attenuating and outputting a signal, a second edge detecting circuit for performing edge detection on one of the field signals which have not been subjected to motion correction, and a detection value of the second edge detecting circuit. Attenuating a differential signal between fields (or between frames) on which motion compensation has not been performed, and outputting the signal based on the outputs from the first and second attenuating circuits. The motion compensation field (or frame) interpolation signal is output as the new field (or frame) signal, or the motion compensation field (or frame) Whether to output a signal, or provided a controller switch determines whether to output both weighted addition signal.

[Operation] In the present invention, a difference signal between fields (or between frames) may be apparently increased at an edge portion of an image and may cause a switching error, regardless of whether or not motion correction is performed. The above-described circuits are provided in order to reduce the apparent increase. These circuits detect an edge portion of an image, and reduce a difference signal between fields (or between frames) according to the detection level. As a result, a switching error of the motion interpolation switching control signal can be reduced, and image distortion in adaptive motion interpolation can be reduced.

Example Overall Configuration Hereinafter, an example of the present invention will be described in detail with reference to the drawings.

FIG. 1 shows the configuration of this embodiment.
Corresponding parts in the drawings are denoted by the same reference numerals.

Also in this embodiment, as shown in FIG. 1, the current field signal S1 and the previous field signal S2 are supplied to the motion vector detection circuit 1, and the motion vector detection circuit 1 uses these field signals S1 and S2 to block each block. To detect a motion vector. The detected motion vector S3 (V) is
It is input to the motion vector correction circuit 2 and the adaptive motion interpolation switching control circuit 3.

The motion vector correction circuit 2 corrects the motion vector V with the input field interpolation ratio α, and
(ΑV) and S5 {(1−α) V} are given to the corresponding motion correction memories 4 and 5, respectively. Each motion compensation memory 4,
5 is a motion compensation field signal in which the coordinates of the target block of the input previous field signal S2 or current field signal S1 are shifted by the compensation motion vector αV and (1−α) V.
S6 and S7 are formed and output.

The difference between the field signals S6 and S7 after the motion correction for each pixel is obtained by the subtraction circuit 6, and the difference is converted into the absolute value by the absolute value conversion circuit 10, and then the variable attenuation is obtained as the motion correction inter-field difference signal S8. Provided to circuit 20. The variable attenuation circuit 20 is supplied with an attenuation control signal (edge detection signal) S20 from the edge detection circuit 21. The variable attenuation circuit 20 generates a motion compensation inter-field difference signal S8 in proportion to the attenuation control signal S20. And the attenuation signal S21 is given to the adaptive motion interpolation switching control circuit 3 described above.

That is, the adaptive motion interpolation switching control circuit 3 is supplied with a signal S21 having a reduced value at the edge portion of the difference signal S8 between the correction fields.

The current field signal S7 output from the motion compensation memory 5
Is supplied to the edge detection circuit 21. Edge detection circuit 21
Detects the edge component of the motion-corrected current field signal S7 and supplies the detection signal S20 to the above-described attenuation circuit 20 as an attenuation control signal.

As described above, the motion compensation inter-field difference signal S8 is not directly supplied to the adaptive motion interpolation switching control circuit 3, but is attenuated in accordance with the result detected by the edge detection circuit 21 and transmitted to the adaptive motion interpolation switching control circuit 3. The reason is that the difference value of the edge part is different from other parts even if the motion correction is appropriately performed, and the switching error due to this is avoided.

Also, output signals S6 and S7 from the respective motion compensation memories 4 and 5
Are given to the corresponding multiplication circuits 8 and 9, and each of the multiplication circuits 8,
9, the field interpolation ratio α, (1−α) is loaded, and then added by the addition circuit 10 to obtain a motion-compensated field interpolation signal S9. The motion compensation field interpolation signal S9 is provided to the multiplication circuit 11.

The above-described current field signal S1 and previous field signal S2
Is also supplied to the subtraction circuit 12. Each of these field signals
The difference between S1 and S2 is obtained by the subtraction circuit 12, and this difference is converted to an absolute value by the absolute value conversion circuit 13, and then the variable attenuation circuit 22 outputs the uncorrected inter-field difference signal S10.
Given to. The variable attenuation circuit 22 includes an edge detection circuit 23
The variable attenuation circuit 22 attenuates the uncorrected inter-field difference signal S10 in proportion to the attenuation control signal S22 to perform the adaptive motion interpolation described above. It is given to the switching control circuit 3. That is, a signal S23 in which the value of the edge portion of the uncorrected inter-field difference signal S10 is reduced is given to the adaptive motion interpolation switching control circuit 3.

The current field signal S1 is supplied to the edge detection circuit 23, and an edge component in the current field signal S1 is detected by the edge detection circuit 23, and the attenuation of the attenuation circuit 22 is controlled according to the detection result.

The reason why the uncorrected inter-field difference signal S10 is supplied to the above-described adaptive motion interpolation switching control circuit 3 via the attenuation circuit 22 is that the balance with the formation processing unit of the motion-compensated inter-field difference signal S8 is considered. It is. Also, even when a still image is targeted, the difference may be large in an edge portion due to jitter or the like, and this is for reducing errors in interpolation switching control due to the difference.

Further, the previous field signal S2 and the current field signal S1 are supplied to the corresponding multiplication circuits 14 and 15, and the respective multiplication circuits 14 and 15
Loads the field interpolation ratio α, (1−α),
After that, the linear interpolation signal S1
It becomes 1. The linear interpolation signal S11 is provided to the multiplication circuit 17.

The adaptive motion interpolation switching control circuit 3 outputs the motion compensation field interpolation signal S9 or the linear interpolation signal S11 as the motion interpolation signal, or outputs the weighted addition signal of both. The magnitude of the motion vector S3 (V), the motion-compensated inter-field difference signal S21 via the attenuation circuit 20, and the uncorrected inter-field difference signal S23 via the attenuation circuit 22 are determined as parameters.
Adaptive interpolation switching signal S12 (β), S13 (1-β) for each pixel
Is output to the corresponding multiplication circuits 11 and 17.

Thus, the multiplication circuit 11 outputs a signal obtained by multiplying the motion compensation field interpolation signal S9 by β, the multiplication circuit 17 outputs a signal obtained by multiplying the linear interpolation signal S11 by (1−β), and these signals are added to the addition circuit 18 Are added to form a final motion interpolation signal S14.

For example, when the motion-compensated inter-field difference signal S21 is larger than the uncompensated inter-field difference signal S23, the motion vector S3 is not 0, and the motion-compensated inter-field difference signal 21 is smaller than a threshold, one of the adaptive interpolation switching signals β is set to 1.

Details of Edge Detection Circuit In the case of this embodiment, the motion-corrected current field signal is used.
The edge detection circuit 21 for S7 and the edge detection circuit 22 for the input current field signal S1 both have the detailed configuration shown in FIG. FIG. 6 shows the configuration for detecting the edge component in the horizontal direction, but the configuration for the detection in the vertical direction is almost the same except that the processing is performed in line units.

The circuit example shown in FIG. 6 has a bandpass filter configuration using a digital circuit.

In FIG. 6, the input signal (current field signal)
The signal is directly supplied to the adder circuit 34, and is also supplied to the adder circuit 34 through four-stage latch circuits 30 to 33 that perform a latch operation based on a clock signal. Thus, the addition circuit 34 to 4
A signal obtained by adding the two pixel signals separated by the pixel is output, and this signal is delayed by one pixel via the latch circuit 35 and is supplied to the subtraction circuit 36. The pixel signal from the third-stage latch circuit 32 from the input terminal side is supplied to the subtraction circuit 36. That is, the addition signal and an intermediate pixel signal between the two pixels subjected to the addition processing are supplied to the subtraction circuit 36. Subtraction circuit 36
Subtracts the addition signal from twice the value of the intermediate pixel signal to obtain an edge detection signal. In the case of this embodiment, this edge detection signal is converted into a level suitable for the variable attenuating circuits (20, 22) by the level converting circuit 37 having a ROM configuration and output. The level conversion circuit 37 also performs absolute value conversion if necessary.

As described above, the attenuation circuit attenuates the inter-field difference signal level in accordance with the output signal from the edge detection circuit.

Correction of Inter-Field Difference Signal FIG. 7 shows a correction example of the motion-compensated inter-field difference signal S8.

The input current field signal S1 shown in FIG. 7 (E) is corrected based on the motion vector as shown in FIG. 7 (A), and the input previous field signal 2 shown in FIG. It is assumed that the correction is made based on the motion vector as shown in FIG. 7 (B). The corrected current field signal S7
And the previous field signal S6 is motion-corrected well, but the edge timing is slightly different.

In this case, as shown in FIG. 7C, a motion-compensated inter-field difference signal S8 having a higher level at the edge portion than at the other portions is output. FIG. 7 (C) and FIG. 7 (G), which will be described later, show a difference signal that has not been subjected to absolute value conversion in order to facilitate visual understanding. On the other hand, the edge detection circuit 21 outputs an edge detection signal S20 having a higher level in the edge portion than in other portions. Therefore, the motion-compensated inter-field difference signal S21 attenuated by the edge detection signal S20 is a signal in which the level of the edge portion is also reduced as shown in FIG. 7 (H). Therefore, the motion compensation inter-field difference signal S21
And the interpolation switching control based on the uncorrected inter-field difference signal S23 shown in FIG. 7 (G) (accurately, the edge attenuating process is also executed) S23 in FIG. 7 (C). Unlike the case of using the motion compensation inter-field difference signal S8 shown in FIG.

For example, in the case where the condition for selecting the motion compensation field interpolation signal is that the difference signal between the motion compensation fields is smaller than the threshold, according to the motion compensation field difference signal S8 shown in FIG. The linear interpolation signal S11 is selected for the edge portion by the interpolation control, and image distortion occurs. However, according to the interfield difference signal S21 in which the edge is attenuated as shown in FIG. The motion compensation field interpolation signal S9 is selected by the adaptive interpolation switching, and the image distortion is eliminated.

Although FIG. 7 is for a moving image, even in the case of a still image, since the inter-field difference signal S10 without correction is attenuated, the motion-compensated field interpolation signal S9 is erroneously selected for an edge portion. Can be prevented.

As described above, according to the above-described embodiment, since the inter-field difference signal does not increase at the edge of the image,
It is easy to determine the quality of the motion-compensated signal, and the switching can be controlled appropriately, so that image distortion at the edge can be reduced.

Other Embodiments Although the present invention is particularly described with respect to a television system conversion apparatus, it can be sufficiently applied to other apparatuses that perform motion correction using a motion vector.

As an example of this, as a high-efficiency coding method, the present invention can also be applied to a case where fields are thinned out on the transmission side and fields thinned out using a motion vector are reproduced on the receiving side. In addition, there is also a system in which motion vector detection and motion interpolation are performed in high-quality television (high-definition television), which has been rapidly advancing in recent years.

As the edge detection circuits 21 and 23, various types can be applied, and are not limited to those shown in FIG.

The variable attenuation circuits 20 and 22 are connected to the absolute value conversion circuits 7 and 13 respectively.
May be provided in the preceding stage.

[Effects of the Invention] As described above, according to the present invention, an edge-corresponding portion of a difference signal used for determination of switching control of an interpolation signal is attenuated. This can reduce the image distortion that has conventionally occurred at the edge portion.

[Brief description of the drawings]

FIG. 1 is a block diagram showing an embodiment of a television interpolation system according to the present invention, FIGS. 2 and 3 are explanatory diagrams of a linear interpolation system, and FIG. 4 is a motion interpolation system with motion compensation. FIG. 5 is a block diagram of a conventional television signal interpolation system, FIG. 6 is a block diagram of a detailed configuration of an edge detection circuit of the above embodiment, and FIG. 7 is a signal waveform diagram of the above embodiment. is there. 1 ... Motion vector detection circuit, 2 ... Motion vector correction circuit, 3 ... Adaptive motion interpolation switching control circuit, 4,5 ... Motion correction memory, 6,12 ... Subtraction circuit, 7,13 ... Absolute value conversion circuit, 8, 9, 11, 14, 15, 17 ... Multiplication circuit, 1
0, 16, 18: Addition circuit, 20, 22: Variable attenuation circuit, 2
1, 23 …… Edge detection circuit.

Claims (1)

(57) [Claims] 1. A method of generating a new field (or frame) signal by interpolating two or more field (or frame) signals of a digitized television signal, using a motion vector. A motion-compensated field (or frame) interpolation signal with motion compensation and a non-compensated field (or frame) interpolation signal with linear interpolation are
The output is adaptively switched or over-added, and at least the difference signal between the motion-compensated fields (or between frames) and the motion-compensated field (between frames) are output as parameters for the switching control. In a television signal interpolation method using a differential signal of the following, a first edge detection circuit that performs edge detection on one of the motion-compensated field signals, and according to a detection value of the first edge detection circuit, A first attenuating circuit for attenuating and outputting a motion-compensated difference signal between fields (or between frames), and a second edge detecting circuit for performing edge detection on one of the field signals that have not been motion-compensated And a difference signal between fields (or between frames) on which no motion correction has been performed, according to the detection value of the second edge detection circuit. A second attenuating circuit for attenuating and outputting the signal, and interpolating the motion compensation field (or frame) as the new field (or frame) signal based on the outputs from the first and second attenuating circuits. A switching control unit for determining whether to output a signal, an uncorrected field (or frame) interpolation signal, or to output both weighted addition signals. Signal interpolation method.