JP3242164B2 - Scanning line number converter for image signals - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to an apparatus for converting the number of scanning lines of an image signal.

[0002]

SUMMARY OF THE INVENTION The present invention relates to a scanning line number conversion apparatus for converting a television image signal of, for example, 2: 1 interlaced scanning and outputting television image signals having different numbers of scanning lines. By converting the number of scanning lines after performing progressive scanning conversion of a pattern, unnecessary aliasing disturbance in a moving image is removed, and the converted image quality of the moving image is improved.

[0003]

2. Description of the Related Art As conventional scanning line number conversion devices, there are, for example, the following (1) and (2) in which the device of (1) is adapted to be motion adaptive.

(1) "Scanning line number conversion system"
No. 174325 (Applicant: Japan Broadcasting Corporation, Inventors: Tanaka, Nishizawa) (2) “Number of Scanning Lines / Sequential Scan Conversion Method”
No. 71225 (Applicant: Japan Broadcasting Corporation, Inventors: Tanaka, Omura, Kurita,
Ninomiya, Nishizawa)

[0005]

The conventional scanning line number conversion apparatus has a sufficient conversion image quality for a still image, but is based on 2: 1 interlaced scanning of an input signal, as will be described in detail later. The aliasing causes an unnatural aliasing signal component for the moving image.
Therefore, the conversion image quality for moving images has not always been sufficient.

FIG. 2 illustrates the principle of the conventional scanning line number conversion method. FIG. 2 shows an image scanning situation in a time (t) -vertical (y) coordinate system. White circles in FIG. 2 indicate positions where the scanning lines exist. .. Assigned to the scanning lines are the scanning line numbers in the input signal, and the numbers 1 ', 2', 3 ',... Are the scanning line numbers in the output signal. D. C. (= Don't C
are) indicates that the signal content of the scanning line is invalid. It is assumed that the conversion ratio of the number of scanning lines is 5: 3 (for example, conversion from 1125 lines to 675 lines).

The conversion of the number of scanning lines is performed as follows. First, a scanning line is interpolated from the input signal on the left side of FIG. 2 by a vertical interpolation filter to create a scanning line (1 ', 2', 3 ', 4') for an output signal. At this time, the filters for forming the scanning lines 1 ', 2', 3 'have the same frequency characteristics but different sampling phases.
Circuits to realize this are conventional examples (1), (2)
Is described in detail. In FIG. 2, the scanning itself is still the same as that of the input signal in the scanning lines 1 ', 2', 3 ', and 4' for the output signal. To create an output signal. At this time, as is clear from the literature (2), the above operation does not significantly change regardless of whether the output signal is 2: 1 interlace scanning or sequential scanning.

In the conventional conversion of the number of scanning lines, when performing the above-described interpolation, the moving image is performed from the scanning lines in the field (the scanning line in the field at t = 0 on the left side in FIG. 2). However, the input signal is a 2: 1 interlaced scan, and the information in the vertical direction is only half of the frame image.
For this reason, in the intra-field interpolation, interpolation with high vertical resolution cannot be performed. For a still image only, as shown by a dotted line in FIG. , The interpolation for the number of scanning lines was performed. However, this was limited to still images.

Accordingly, it is an object of the present invention to improve this point and to provide an image signal scanning line number conversion apparatus capable of obtaining a sufficient conversion image quality even for a moving image.

[0010]

In order to achieve the above object, the present invention provides a motion vector generating circuit for outputting a motion amount of an image in an input image signal of interlaced scanning as a motion vector, and a delay amount of the input image signal. A motion compensation field delay circuit that performs motion compensation according to the motion vector, a delay circuit that delays the input image signal, and a weight that performs a weighted average on an output signal of the delay circuit and an output signal of the motion compensation field delay circuit. An averaging circuit, and a circuit that determines a weight in the weighted averaging circuit based on a motion compensated inter-frame difference signal obtained by motion-compensating the delay amount according to the motion vector;
A circuit for converting the number of scanning lines from the weighted averaging circuit to extract an image signal.

[0011]

According to the present invention, the number of scanning lines is converted after virtually performing motion-compensated sequential scan conversion on an interlaced scan input image. As a result, unnecessary aliasing disturbance in the moving image can be removed, and the converted image quality of the moving image can be improved.

[0012]

Embodiments of the present invention will be described below in detail with reference to the drawings.

FIG. 1 shows the configuration of a scanning line number conversion apparatus according to an embodiment of the present invention.

In FIG. 1, an input signal of this apparatus is an image signal having 1125 scanning lines / field frequency 60 Hz / interlace ratio 2: 1, and an output signal is 525 lines / field frequency 60 Hz / interlace ratio 2: 1 is an image signal. That is, the present apparatus converts only the number of scanning lines among the basic scanning parameters of an image signal, and the conversion ratio is 1125: 525 = 15: 7.

An input signal of the present apparatus is first input to a motion vector detection circuit 1. A motion amount of the image of the motion vector detection circuit 1, the input signal is detected as a motion vector V _0. For this detection, various motion vector detection techniques already proposed can be used as they are. When a motion vector is provided from outside, it may be received. V ₀ is input to the motion vector value correction circuit 2, the value of which is slightly corrected, and the motion vector V (V) used in the other circuits in FIG.
_x , V _y ). The content and meaning of the correction will be described later. Here, V is actually 2 of V _x and V _y
Consists of two signals. V _x represents the horizontal component of the motion, the unit is set to (pixels / field). V _y represents the vertical component of the motion, and the unit is ((in a frame) scan line / field). The present apparatus can operate regardless of whether the value of V is updated for each pixel, for each pixel block of a certain size, or for each field. These circuits 1 and 2 constitute a motion vector generating circuit 3 as a whole.

The input signals are also (565H + V),
The signals are input to the motion compensation memories 10, 11, and 12 having delay amounts of (564H + V) and (563H + V). Here, H represents one line in the field. V is the motion vector. For example, the delay amount of (563H + V) is represented by the number of pixels as follows. Let N be the number of pixels in one line

[0017]

(Equation 1)

## EQU1 ## The other delay amounts can be similarly obtained. Here, the polarities of V _x and V _y are set to V _x > 0 and V _y > 0 when the image moves to the upper right with time. The output of the memory 10 is a signal x _-2 , the output of the memory 11 is x ₀ , and the output of the memory 12 is x ₂ . These are input to the weighted average circuit 16.

The input signal also includes a 1H delay 13, 1
The signals are input to the cascade connection circuits 4 and 15. Here, the input signal of the apparatus is x _-3 , the output of the delay 13 is x _-1 , the output of the delay 14 is x ₁ , and the output of the delay 15 is x ₃ . These are also input to the weighted average circuit 16.

In the apparatus shown in FIG. 1, the image quality deterioration in a moving image at the time of virtual progressive scan conversion is solved by motion-compensating the field delay amount in the memories 10 to 12 according to the motion vector V. For example, when the vertical motion of the image is V _y = 2, the scanning lines a, b, and c in FIG. 3 are delayed by the motion compensation memories 10 to 12 in FIG. 1 to obtain x ₂ , x ₀ , and x _{−. 2} is output and the weighted average circuit 1
Sequentially inputting a scan signal x ₃ ~x _-3 to 6. The display method of FIG. 3 is the same as that of FIG. 2, and the operation after interpolation is the same as that of the conventional method (details will be described later). As described above, in the apparatus shown in FIG. 1, even when there is a motion in the image, the image can be converted into the sequential scanning without using the vertical filter, so that the converted image quality of the moving image is good. That is, high resolution and high image quality can be obtained. This situation is the same for the horizontal motion component. In addition,
Since V _x = V _y = 0 in a still image, the inter-field interpolation for a conventional still image is also included in the motion compensation interpolation operation.

FIG. 3 shows only the case where V _y is an even number. However, when V _y is an odd number, motion compensation interpolation cannot be performed because there is no scanning line corresponding to sequential scanning one field before. . Accordingly, the motion vector value correction circuit 2 in FIG. 1 corrects the vertical motion vector value by the following algorithm. Assuming that the vertical and horizontal motion vectors detected by the motion vector detection circuit are V _y0 and V _x0 respectively,

[0022]

(Equation 2)

[0023]

Equation 3] _{V x = V x0 ... (3} ) i.e., for V _y0 are if V _y0 is odd are rounded to a small even number of more absolute value V _y0. This is because when the motion compensation is not accurate, the image quality is generally better in the under-compensated state than in the over-compensated state. Since there is no special situation based on the interlaced scanning as described above for the horizontal motion component, it is not necessary to correct _Vx0 .

The weighted averaging circuit 16 forms a seventh-order interpolation filter, and the above-mentioned x _i (i is an integer and -3 ≦ i ≦
3) is multiplied by a weight (coefficient) α _i, and then added and output. Ie

[0025]

[Outside 1]

At this time, the value of the coefficient α _i changes depending on the signal K described later and the scanning line number in a frame or a field. The output of the weighted average circuit 16 is input to the buffer memory 17. Here, by reading / writing signals according to the scanning line numbers in FIG.
Conversion of vertical scanning in the image is performed, and 525/60 /
A 2: 1 output signal of the device is created.

The input signal of this device is also (525H + 2
V) is input to the motion compensation memory 4 having the delay amount. The operation of No. 4 is the same as that of the memories 10 to 12 except that the motion compensation amount is twice the motion vector V, and the delay amount is obtained in the same manner as the equation (1). The subtracter 5 calculates the difference between the input signal of the present apparatus and the output of the memory 4 and outputs the result. The output of the subtracter 5 is a motion compensation frame difference signal. The absolute value circuit 6 outputs the absolute value of the output of the subtracter 5. The K determination circuit 7 determines the signal K based on the output of the absolute value circuit 6. Here, the value of K is, for example, 0 ≦ K ≦ 1. The number of possible values of K within this range depends on the conditions of the weighted average circuit 16 described above. For example, the coefficient α _i
However, if three types can be changed with respect to K, there are three possible values of K, for example, K is 0, 0.5, or 1. The content of the K decision circuit 7 may be simply a quantizer that simply increases K if the output value of the absolute value circuit 6 is large, or various methods used in the conventional image motion detection method. May be included. The output of the K determination circuit 7 via the 1H delay 8 for phase adjustment between x _i, is input to the weighted average circuit 16 as a signal K. Note that the phase of the signal K is adjusted to the signal x- ₁ . Each of these components 4 to 8 constitutes an interpolation selection determination circuit 9 as a whole.

The conversion of the number of scanning lines according to the present embodiment uses the memories 10 to 12, the 1H delays 13 to 15, the motion vector generating circuit 3, the interpolation selection judging circuit 9, the weighted averaging circuit 16, and the buffer memory 17 as described above. Since the motion compensation type progressive scan conversion is performed on the 2: 1 interlaced scan signal virtually and then interpolation for converting the number of scanning lines is performed, high resolution and high resolution can be applied to moving images. It is possible to perform high quality conversion.

Scanning line number conversion time including the embodiment of FIG. 1—
The vertical spectrum is shown in FIGS. 4 and 5, the horizontal axis represents the time frequency f (Hz), and the vertical axis represents the vertical spatial frequency ν (cph). FIG. 4 (a) is the spectrum of the image discussed below,

[0030]

[Outside 2]

[0031]

[Outside 3]

A carrier for spatiotemporal sampling by interlaced scanning. In the figure, the original spectrum and the folded spectrum of the time sampling by the field are shown by solid lines, and the folded spectrum by interlaced scanning is shown by a dotted line. FIG. 3B shows a signal spectrum when the same image is captured by a camera having 525 scanning lines. In FIG. 10A, V _y = 2 (1125 scanning lines / field), but in FIG.

[0033]

(Equation 4)

It can be considered that

FIG. 4C shows the interpolation characteristics of the conventional still image interpolation described in FIG. The hatched portion in FIG. 3C is a stop band of the interpolation filter, and the signal spectrum of this portion is cut. By subjecting the signal remaining in the pass band to scan conversion at 525/2: 1, the spectrum shown in FIG. Referring to FIG. 9D, in addition to the original spectrum (solid line) whose length is shortened (decreased in resolution) and the spectrum (dotted line) from the interlace carrier X, −60 ≦
It can be seen that an unnatural spectrum is generated in the region of f ≦ −30 and 0 ≦ f ≦ 30. This is caused by the folded spectrum of the interlace remaining in the pass band in FIG. Also, this spectrum is ν of the original image.
Caused by the mid-range components. Because of this spectrum, the conventional conversion not only reduces the vertical resolution of the image, but also causes unnatural image quality interference, such as unnatural and noticeable interline flicker at the horizontal edge of the image.

Similarly, FIG. 11E shows the conventional interpolation characteristics for moving images, and FIG. 10F shows the result of the conversion of the number of scanning lines based on the interpolation characteristics. Also in this case, in addition to the decrease in the resolution of the original spectrum, an unnatural folded spectrum is also generated. However, in this case, this unnatural spectrum is caused by the ν high frequency component of the original image. That is, the ν high-frequency component of the original image is output after the value of ν is converted. However, with respect to the image quality, unnatural interlace interference also occurred with the decrease in the resolution, and the image quality similarly deteriorated.

On the other hand, according to the apparatus shown in FIG.
(G), and an output signal spectrum equivalent to that shown in FIG. 4B directly captured by the scanning line 525 system can be obtained as shown in FIG. is there. Further, when the interlace interference is conspicuous in the images of FIGS. 4B and 5H, the interpolation characteristic is intentionally narrowed with respect to ν as shown in FIG. As shown in (j), it is also possible to obtain an image that is visually noticeable and has no interference with the low ν band. in this way,
According to this apparatus, in addition to obtaining a high-quality scan conversion output free from unnatural interference, an extremely high-quality image with less interference than a normal 525-line camera may be obtained depending on the selection of characteristics. It is also possible to get

The control of the coefficient α _i by the signal K in the apparatus shown in FIG. 1 is intended to cope with a case where an error occurs in motion compensation due to an erroneous detection of a motion vector or the like. = 1, K = 0.5, K = 0.5, K = 0.5
In the case of (1), the characteristic shown in FIG. 5 (g) and the characteristic shown in FIG. 4 (e) may be set to an intermediate characteristic (for example, an average characteristic of both).

Another embodiment of the present invention is shown in FIG. The input / output signals are the same as in FIG. The circuit configuration differs from FIG. 1 in that the 547H delay 20, the motion compensation memories 21, 1H in FIG.
That is, delays 22 and 23 are used. The contents and operation of the other circuits are the same as those in FIG.

The input signal passes through the 547H delay 20,
Motion compensation memory 21 having a delay amount of (16H + V)
Is motion compensated. If | V | ≦ 16H, the operation of the motion compensation memory 12 in FIG. 1 is equivalent to the operation of the 547H delay 20 and the motion compensation memory 21 together. This is often | V | ≦ 16H in a normal image, so that the motion compensation memory 12 is provided with a 547H delay 20 and a motion compensation memory 21.
The purpose of the present invention is to reduce the delay size of the expensive motion compensation memory by dividing into two, and to economically realize the device. The output of the motion compensation memory 21 is 1H delay 2
It is input to 2, 23 cascade connections. Motion compensation memory 2
The output of the 1,1H delay 22 and 23 each x _-2, x _0,
the x _2. That is, the components 20 to 22 substitute the motion compensation memory 11 of FIG. 1 and the components 20 to 23 substitute the motion compensation memory 12. This is another way to save expensive motion compensation memory.

FIG. 7 shows an example of the operation of the apparatus shown in FIG. The notation of FIG. 7 is the same as that of FIG. FIG. 7 differs from FIG. 3 in that x ₀ and x ₂ are not created from the scanning lines b and c, but are created from x ₋₂ . This is equivalent to the operation of the apparatus of FIG. 1 when the motion vectors for interpolating x ₋₂ , x ₀ , and x ₂ are equal as shown in FIG. However, when a motion vector for a certain scanning line and a scanning line adjacent thereto are different, the operation of the apparatus in FIG. 6 is different from that in FIG. 1, and image quality may be deteriorated due to a motion compensation error. However, depending on the method of generating a motion vector and the type of image, it is highly possible that the apparatus of FIG. In such a case, the apparatus can be economically realized by the configuration of FIG.

The present invention can be applied to a system having a different number of scanning lines for the input and output, in addition to the examples described above.
Further, the present invention can be used not only for down conversion like 1125 → 525 in the present embodiment but also for up conversion like 525 → 1125. Further, the output signal may be a progressive scan. Further, the motion compensation may be performed using signals before and after one field of the signal of interest, and furthermore, it is possible to increase the number of motion vectors for which motion compensation such as an interpolated signal is performed in motion compensation interpolation.

[0043]

According to the present invention, by performing virtually motion-compensated progressive scanning conversion on an interlaced scanning input image signal and then converting the number of scanning lines, unnatural aliasing interference in a moving image is obtained. Can be prevented, and the conversion image quality of a moving image can be improved.

[Brief description of the drawings]

FIG. 1 is a block diagram showing an embodiment of the present invention.

FIG. 2 is a diagram illustrating an operation of a conventional scanning line number conversion device.

FIG. 3 is a diagram showing an operation example of the device of FIG. 1;

FIG. 4 is a diagram showing a time-vertical spectrum for explaining the operation of the apparatus of FIG. 1;

FIG. 5 is a diagram showing a time-vertical spectrum in the same manner.

FIG. 6 is a block diagram showing another embodiment of the present invention.

FIG. 7 is a diagram showing an operation example of the device of FIG. 6;

[Explanation of symbols]

 3 Motion vector generation circuit 9 Interpolation selection judgment circuit 10, 11, 12 Motion compensation memory 13, 14, 15 1H delay 16 Weighted average circuit 17 Buffer memory

Claims (1)

(57) [Claims] 1. A motion vector generation circuit for outputting a motion amount of an image in an input image signal of interlaced scanning as a motion vector, and a motion compensation field delay circuit for compensating a delay amount of the input image signal according to the motion vector A delay circuit for delaying the input image signal; a weighted average circuit for performing a weighted average on an output signal of the delay circuit and an output signal of the motion compensation field delay circuit; and a motion compensation for a delay amount according to the motion vector. A circuit for determining a weight in the weighted average circuit based on the obtained motion compensated inter-frame difference signal; and a circuit for converting the number of scanning lines from the weighted average circuit and extracting an image signal. Scanning line number converter for signals.