JPH08163566A - Motion vector detector - Google Patents

Abstract

PURPOSE: To eliminate the waste of the output word length of a filter circuit for removing a component unrequired for movement detection and to reduce a circuit scale without lowering detection accuracy. CONSTITUTION: The filter circuit for removing the unrequired component from input digital video signals is constituted of the IIR type digital filter 20 of 12-bit output for instance composed of an addition circuit 22, a delay circuit 25 and a multiplying circuit 26 and a selector 24 for selecting 8 bits of a valid part from the filter output of 12 bits and outputting them. By selecting the selector by the luminance level of the input video signals, the valid part of 8 bits is selected and outputted.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a motion vector detection device suitable for predictive coding of images and motion compensation of video cameras.

[0002]

2. Description of the Related Art A motion vector is detected when efficiently performing predictive coding processing of an image or when performing image stabilization of a video camera by image processing. A representative point matching method is known as a method for detecting such a motion vector. The representative point matching method sets a representative point in the previous field, compares the pixel data of the search range of the current field with the pixel data of the representative point of the previous field, and selects the pixel data of the representative point of the pixel data of the search range. The motion vector is obtained from the closest pixel data.

That is, FIG. 7 shows a method of detecting a motion vector by the representative point matching method. In FIG. 7, 101 indicates the previous field and 102 indicates the current field. In this example, as shown in FIG.
Three detection areas 103A, 103B, 103C, 103D
Is set. In the front field 101, the representative points D, D,
, Are set, and search ranges S, S, S, ... Are set in the vicinity of the representative points D, D, D ,. Each search range S, S, S, of the current field
The pixel data in ... Are compared with the pixel data of the representative point in the previous field. Of the pixel data in each search range S, S, S, ..., the one closest to the representative point is detected,
Based on this, the motion vector is obtained.

As shown in FIG. 9, a motion vector detecting device for detecting a motion vector by such representative point matching has a filter circuit 111 for removing a frequency component unnecessary for motion detection and pixel data of a representative point. A representative point memory 112 for storing the following, a difference calculation circuit 113 for obtaining a residual by comparing the representative point of the previous field and pixel data in the search range of the current field, and a residual memory 114 for storing the obtained residual. And a motion vector detection circuit 115 that detects a motion vector based on the residual.

As described above, the motion vector detecting device is provided with the filter circuit 111 for removing frequency components unnecessary for motion vector detection. The filter circuit 111 has, for example, an IIR type configuration including a feedback loop as shown in FIG.

In FIG. 10, a digital signal is supplied to the input terminal 121. The signal from the input terminal 121 is supplied to the adding circuit 122. The output of the adder circuit 122 is derived from the output terminal 124 and supplied to the delay circuit 123. The output of the delay circuit 123 is supplied to the multiplication circuit 125. The coefficient of the multiplication circuit 125 is, for example, 1/2. The output of the multiplication circuit 125 is fed back to the addition circuit 122.

[0007]

In the filter circuit 111 including such a feedback loop, the output word length extends at most 1 bit in the MSB direction and infinitely in the LSB direction. When the number of bits of the output of the filter circuit 111 is insufficient, overflow occurs when the brightness level is high, or accuracy decreases when the brightness level is low.
Therefore, the input word length of the filter circuit 111 is, for example, 8
In the case of bits, the output word length needs to be about 12 bits.

However, increasing the number of output bits of the filter circuit 111 increases the hardware scale.

Therefore, an object of the present invention is to provide a filter device capable of reducing the circuit scale without reducing the accuracy.

[0010]

SUMMARY OF THE INVENTION The present invention comprises a filter means for removing unnecessary components from an input digital video signal, and a motion vector detecting means for detecting a motion vector from the digital video signal passed through the filter means. And a filter means for removing an unnecessary component from the input digital video signal, the filter section having an M-bit output,
The motion vector detecting device is provided with a selector unit for selecting and outputting N bits (M> N) of an effective portion from the output of M bits of the filter unit.

In the present invention, the filter means includes a feedback loop.

In the present invention, the selector section of the filter means selects an effective portion according to the brightness level of the input digital video signal.

[0013]

The filter circuit 2 is composed of the digital filter 20 and the selector 24. For example, of the output of the 12-bit digital filter 20, only the effective portion of 8 bits is extracted and output. As a result, the filter circuit 2
The number of output bits does not increase, and the circuit scale can be reduced.

[0014]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS An embodiment of the present invention will be described below with reference to the drawings. FIG. 1 shows an example of a motion vector detecting device to which the present invention is applied. In FIG. 1, reference numeral 1 is an input terminal for a digital video signal. The digital video signal from the input terminal 1 is filtered by the filter circuit 2
Is supplied to.

The filter circuit 2 removes frequency components unnecessary for motion detection. This filter circuit 2 is
It is configured as shown in FIG.

In FIG. 2, the input terminal 21 has, for example, 8
A bit digital video signal is provided. The digital video signal from the input terminal 21 is supplied to the adding circuit 22. The output of the addition circuit 22 is fed back to the addition circuit 22 via the delay circuit 25 and the multiplication circuit 26. The coefficient of the multiplication circuit 26 is, for example, 1/2. By the adder circuit 22, the delay circuit 25, and the multiplier circuit 26, II
An R type digital filter 20 is configured.

An 8-bit digital signal, for example, is supplied to the input terminal 21, but since the digital filter 20 is an IIR type, the output word length of the digital filter 20 increases. As the word length of the output of the digital filter 20, for example, 12 bits are prepared.

The selector 24 has a 5-bit input word length of 5 bits.
The input data is switched. The output of the 12-bit digital filter 20 is the selector 2
4 input terminals 24A, 24B, 24C, 24D, 24E
Are sequentially bit-shifted and input. That is, in FIG. 3, the 12-bit digital filter 2
If the LSB to MSB of the output of 0 are bits b _{0 to} b ₁₁ , respectively, the terminal 24A of the selector 24 is supplied with 8 bits of bits b ₄ to b ₁₁ as shown in FIG. 3A. As shown in FIG. 3B, 8 bits of bit b ₃ to bit b ₁₀ are supplied to the terminal 24B of the selector 24.
At the terminal 24C of the selector 24, as shown in FIG. 3C,
Eight bits from bit b ₂ to bit b ₉ are supplied. As shown in FIG. 3D, 8 bits of bit b ₁ to bit b ₈ are supplied to the terminal 24D of the selector 24. As shown in FIG. 3E, the bit b is applied to the terminal 24E of the selector 24.
8 bit ₀ through bit b ₇ is supplied.

The selector 24 is supplied with a select signal from the select signal generating circuit 28. To the select signal generating circuit 28, the average value of the luminance level of each field of the video signal is supplied from the terminal 26. The select signal generating circuit 28 generates a select signal according to the average value of the brightness levels of the video signal.

For example, as shown in FIG. 4, the average value of the brightness level of the video signal is set to five levels L1 to L5. If the average value of the brightness level is L1, the terminal 24
If it is set to E and the average value of the brightness level is L2, it is set to the terminal 24D. If the average value of the brightness level is L3, it is set to the terminal 24C and the average value of the brightness level is L3. In the case where the average value of the brightness levels is L4, it is set in the terminal 24B, and when the average value of the brightness levels is L5, the terminal 24B is set.
Set to A.

Thus, for example, the effective word length of the output data of the digital filter 20 can be taken out by switching the selector 24 according to the average value of the brightness level of the input signal. The output of the selector 24 is taken out from the output terminal 27.

In FIG. 1, the filter circuit 2 outputs an 8-bit digital video signal with the effective portion selected, as described above. The output of the filter circuit 2 is supplied to the representative point memory 3 and the subtraction circuit 4. The representative point memory 3 is a memory controller 5
Controlled by. The pixel data of the representative point is written in the representative point memory 3 for the next motion vector detection, and the pixel data of the representative point of the previous field is read out. The pixel data of the representative point read from the representative point memory 3 is supplied to the latch circuit 6. The output of the latch circuit 6 is supplied to the subtraction circuit 4. In the subtraction circuit 4, the pixel data of the representative point of the previous field and the pixel data of the search range of the current field are subtracted. The output of the subtraction circuit 4 is supplied to the absolute value circuit 7. From the absolute value circuit 7,
The absolute value of the difference between the pixel data of the representative point of the previous field and the pixel data of the search range of the current field is output. The output of the absolute value circuit 7 is supplied to the residual memory 8.
Of the absolute values of the difference between the pixel data of the representative point and the pixel data of the search range in each search range, the ones corresponding to the corresponding positions are sorted, and these are accumulated to obtain the residual difference. It is stored in the residual memory 8.

That is, for example, as shown in FIG.
One detection area includes (6 × 8) search ranges S1 and S
2, S3, ..., S48 are set. Then, in each of the search ranges S1, S2, S3, ..., S48, FIG.
As shown in, there are (5 × 7) pixels. The pixel data h ₁ (3,4) _n , h ₂ (3,4) _n , ..., H at the center of each search range S1, S2, ..., S9, S10 ,.
_The representative points are ₉ (3,4) _n , h ₁₀ (3,4) _n, .... Note that the search range S9 is shown in FIG. 5B, and h ₉ (3, 4) is the representative point in the search range S9.

As shown in FIG. 6, in the search range S1, the pixel data h ₁ (3, 3) of the representative point of the previous field is obtained.
4) _n and the pixel data h ₁ (1,1) _{n + 1} , h ₁ (1,2) _{n + 1} , ..., H of the search range S1 of the current field
₁ (2,1) _{n + 1} , h ₁ (2,2) _{n + 1} , ..., h
The absolute value of the difference from ₁ (3,1) _{n + 1} , h ₁ (3,2) _{n + 1} , ... Is obtained. _{| H 1 (1,1) n +} 1 -h 1 (3,4) n | | h 1 (1,2) n + 1 -h 1 (3,4) n | ... | h 1 (2,1 _{) n + 1 -h 1 (3,4} ) n | | h 1 (2,2) n + 1 -h 1 (3,4) n | ... | h 1 (3,1) n + 1 -h 1 (3,4) _n | | h ₁ (3,2) _{n + 1} −h ₁ (3,4) _n |

In the search range S2, the pixel data h ₂ (3,4) _n of the representative point of the previous field and the pixel data h ₂ (1,1) _{n + 1} of the search range S2 of the current field,
h ₂ (1,2) _{n + 1} , ..., h ₂ (2,1) _{n + 1} , h
₂ (2,2) _{n + 1} , ..., h ₂ (3,1) _{n + 1} , h
₂ The absolute value of the difference between (3, 2) _{n + 1} , ... Is obtained. _{| H 2 (1,1) n +} 1 -h 2 (3,4) n | | h 2 (1,2) n + 1 -h 2 (3,4) n | ... | h 2 (2,1 _{) n + 1 -h 2 (3,4} ) n | | h 2 (2,2) n + 1 -h 2 (3,4) n | ... | h 2 (3,1) n + 1 -h 2 (3,4) _n | | h ₂ (3,2) _{n + 1} −h ₂ (3,4) _n |

Similarly, in the search ranges S3, S4, ... Shown in FIG. 5, the absolute value of the difference between the pixel data of the representative point of the previous field and the pixel data of the search range of the current field is obtained.

When there are 48 search ranges from S1 to S48, the residual is obtained as follows. δ (1,1) = Σ _{j = 1 to 48} │h _j (1,1) _{n + 1} −h _j (3,4) _n │δ (1,2) = Σ _{j = 1 to 48} │h _{j (1,2) n + 1 -h j} (3,4) n | ... δ (2,1) = Σ j = 1 to 48 | h j (2,1) n + 1 -h j (3,4 ) _N | δ (2,2) = Σ _{j = 1 to 48} | h _j (2,2) _{n + 1} −h _j (3,4) _n │ ... δ (3,1) = Σ _{j = 1 to 48} | h _j (3,1) _{n + 1} −h _j (3,4) _n | δ (3,2) = Σ _{j = 1 to 48} | h _j (3,2) _{n + 1} −h _j ( 3, 4) _n |

In FIG. 1, the residuals δ (1,1), δ (1,2), ..., δ (2,1), thus obtained, are obtained.
δ (2,2), ..., δ (3,1), δ (3,2), ...
Are stored in the residual memory 8. Then, the motion vector detection circuit 9 detects the smallest of these residuals, and the motion vector is obtained. The obtained motion vector is output from the output terminal 10.

[0029]

According to the present invention, the filter circuit 2 is
Comprised of a digital filter 20 and a selector 24,
The selector 24 is switched depending on the brightness level, and for example, of the output of the 12-bit digital filter 20, only the effective 8 bits are extracted and output. For this reason,
The number of bits of the filter circuit 2 does not increase, and the circuit scale can be reduced.

[Brief description of drawings]

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

FIG. 2 is a block diagram of an example of a filter circuit according to an embodiment of the present invention.

FIG. 3 is a schematic diagram used to describe a filter circuit according to an embodiment of the present invention.

FIG. 4 is a schematic diagram used for explaining a filter circuit according to an embodiment of the present invention.

FIG. 5 is a schematic diagram used to describe an embodiment of the present invention.

FIG. 6 is a schematic diagram used to describe an embodiment of the present invention.

FIG. 7 is a schematic diagram used to describe a motion vector detection method by representative point matching.

FIG. 8 is a schematic diagram used for explaining a motion vector detection method by representative point matching.

FIG. 9 is a block diagram of an example of a conventional motion vector detection device.

FIG. 10 is a block diagram of an example of a filter circuit in a conventional motion vector detection device.

[Explanation of symbols]

 2 Filter Circuit 20 IIR Type Digital Filter 24 Selector 28 Select Signal Generation Circuit

Claims (3)

[Claims] 1. A filter means for removing an unnecessary component from an input digital video signal, and a motion vector detection means for detecting a motion vector from the digital video signal passed through the filter means. Is an M-bit output filter section for removing unnecessary components from an input digital video signal, and a selector section for selecting and outputting N bits (M> N) of an effective section from the M-bit output of the filter section. And a motion vector detecting device provided with. 2. The motion vector detection device according to claim 1, wherein the filter means includes a feedback loop. 3. The motion vector detecting apparatus according to claim 1, wherein the selector section of the filter means selects an effective portion according to the brightness level of the input digital video signal.