JPH1028267A - Video-encoding device - Google Patents

Abstract

PROBLEM TO BE SOLVED: To prevent a video from generating notches, or to prevent an edge from being partially blurred in the video pattern of a title image of the like after compression encoding. SOLUTION: An input video signal 101 is divided into plural blocks by a block divider 102, discreate cosine converted by a discrete cosine converter 103, quantized by a qiantizer 104 by using a using Q parameter (quantization width) selected by a switcher 117, variable length encoded by a variable length encoder 105, and outputted as a compression-encoded column. The switcher 115 selects a Q parameter 111 which is not modulated by activity as the Q parameter used when an image is a prescribed video pattern such as a title image, and selects a Q parameter 114 which is modulated by activity by a Q parameter modulator 113 when the image is an image except the prescribed video pattern.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to an improvement in a video encoding device for compressing and encoding a video signal.

[0002]

2. Description of the Related Art In recent years, as an apparatus for encoding a video signal, an image is divided into blocks each including a plurality of adjacent pixels.
A method using a transform coding method for performing an orthogonal transform such as a discrete cosine transform for each block is generally known. This encoding method is a method of quantizing a transform coefficient by a predetermined quantization width, and compressing and encoding using a variable length code such as a Huffman code.

Further, in coding of a moving image such as a television signal, inter-frame coding is performed using correlation between frames. This inter-frame coding is performed by predicting a target frame using a past frame as a reference frame when viewed from a frame to be coded, and predicting a target frame and coding an error signal thereof. There is bi-directional frame prediction encoding in which a target frame is predicted using a future frame as a reference frame, and the prediction error signal is encoded. Both types of encoding can be efficiently compression-coded.

In this case, intra-frame coding or forward frame predictive coding is performed for each predetermined frame, and frames between these frames are subjected to bidirectional frame predictive coding.
The reference frame of the forward frame predictive coding is the closest frame among the intra-frame coded frames or the forward frame predictive coded frames located in the past. In addition, the reference frame of the frame to be subjected to bidirectional frame predictive encoding is an adjacent frame among intra-frame encoded frames and forward frame predictive encoded frames located in the past and the future.

[0005] Prediction between frames is performed for each block composed of a plurality of pixels, and motion compensation is performed by detecting the motion amount of each block. Therefore, the motion amount of each block is transmitted or recorded together with the prediction error signal.

Conventionally, for example, Japanese Patent Laid-Open No. 2-184188
As disclosed in Japanese Unexamined Patent Publication, a video signal is stored as a plurality of continuous frame data, divided into three or two dimensions, and an encoding process is performed using signal correlation (activity). is there.

As a method of determining a quantization coefficient for image data, a report on MPEG2 video data “Test Model 3 Draft (26-Nov, 1992) ISO-IEC / J
TC1 / SC29 / WG11 pp.69-72 "describes the following three-dimensional quantization coefficient determination method.

In particular, in the second term “rate control” and the third term “adaptive quantization”, the quantization width mq
uant (j) is obtained as follows.

Here, d0 is the initial value of the virtual buffer of one frame image, B (j) is the amount of generation up to a predetermined area j in one frame image, MBCNT is the number of predetermined areas in the image,
T (j-1) is the predicted generated code amount immediately before the predetermined area.

Assuming that j is a natural number starting from 1, a coding state d (j) of the specific virtual buffer is obtained as follows before the coding of the predetermined area j is started.

For example, the initial value d0 of the virtual buffer of the intra-frame coded image is obtained as 10 × r / 31.

D (j) = d0 + B (j−1) × T (j−1) / MBCNT (Equation 1) Next, the reference quantization coefficient Q (j) of the predetermined area j is obtained as follows. Can be

However, the reaction coefficient r is defined as follows.

Q (j) = d (j) × 31 / r (expression 2) r = 2 × bit rate per second / number of frames per second (expression 3) The spatial activity mquant (j) of the predetermined area j is obtained as follows from a total of eight luminance values of a total of eight sub-blocks constituting the owning frame and field. However, the sub-block here is composed of 8 × 8 pixels.

Act (j) = 1 + min (var-sblk) | 8-sub-block (Equation 4) var-sub-block = (1/64) × SUM (Pk-Pmean) | k = 1,64 ... (Equation 5) Pmean = (1/64) × SUM (Pk) | k = 1,64 (Equation 6) Nact (j) = (2 × act (j) + avgact) / (act (j) +2 x avgact) (Equation 7) mquant (j) = Q (j) x Nact (j) (Equation 8) The value Nact (j) obtained from the above Equation 7 is from “0.5” to “2 The spatial activity mquant (j) has a value from “1” to “3” according to the MPEG2 standard.
It is determined to be a value of 1 ".

[0016]

However, these methods are applied to all input data, and are not always effective for all images.

That is, for a video including many edges such as a subtitle screen, encoding processing is performed by changing the quantization coefficient for each predetermined area in one frame according to the activity as usual. There has been a problem that the image after the encoding process is jagged or has a blurred edge.

The reason is that a video including many edges, such as a subtitle screen, has a characteristic that the luminance value locally changes drastically within one image. Therefore, if a quantization parameter using the modulation of the activity is used. This is because the distribution of the quantization width is widened, and modulation is performed so that the change in the quantization width appears finely.

Humans have the property of seeing a character as one continuous and uniform object having a more significant meaning than the background, and the same character differs when there is a change in the quantization parameter around the same character. Since encoding is performed using the quantization width, a dirty portion and a clean portion are generated, so the dirty portion is conspicuous,
The above contradictions will occur.

The present invention has been made in view of the above points, and an object thereof is to provide a jagged or partially blurred edge in a video after encoding processing even for a video including many edges such as a subtitle screen. It is to prevent that.

[0021]

In order to achieve the above object, according to the present invention, a video including many edges, such as a subtitle screen, is input using a quantization parameter which does not modulate an activity. A video signal is compression-encoded to prevent a jagged or partially blurred edge image.

In the present invention, when the video signal has at least a predetermined video pattern, the value Nact (j) of the above equation (7) may be considered to be fixed to "1.0", or the equation (8) becomes the following equation (9). It may be.

Mquant (j) = Q (j) × Nact (j) If not a predetermined video pattern = Q (j) if a predetermined video pattern (Equation 9) That is, the video according to the first aspect of the present invention. The encoding device is a video encoding device that compresses and encodes an input video signal, and includes a predetermined video pattern detection unit that detects whether the input video signal is a predetermined video pattern, and a plurality of the input video signals. A block dividing unit that divides the block video signal into pixels, an activity calculating unit that inputs a block video signal of the block dividing unit, calculates an activity between these block video signals, and a predetermined video pattern detecting unit. An output is input, and when at least the input video signal is the predetermined video pattern, the quantization width is determined independently of the activity, while the predetermined video pattern is determined. If not, the quantization width determining means for determining the quantization width according to the activity calculated by the activity calculating means, and the block video signal with the quantization width determined by the quantization width determining means. Quantizing means for quantizing, wherein the output of the quantizing means is subjected to variable-length coding to output a compressed coded sequence.

According to a second aspect of the present invention, there is provided a video encoding apparatus comprising:
A video encoding device that compresses and encodes an input video signal, wherein a time code synchronized with the input video signal is input, and it is detected whether the time code indicates a predetermined video section specified from the outside. Predetermined video section detection means,
A block dividing unit that divides the input video signal into block video signals composed of a plurality of pixels; an activity calculating unit that receives a block video signal of the block dividing unit and calculates an activity between these block video signals; The output of the predetermined video section detection means is input, and if the input video signal is the predetermined video section, the quantization width is determined independently of the activity. Means for determining a quantization width in accordance with the activity calculated by the means, and quantization means for quantizing the block video signal with the quantization width determined by the quantization width determination means. The signal obtained by the quantization means is subjected to variable-length encoding, and a compressed encoded sequence is output.

According to a third aspect of the present invention, there is provided a video encoding apparatus comprising:
An image encoding device for compression encoding an input image signal, wherein an external interrupt signal is input, an external interrupt detection means for detecting the presence or absence of the input of the external interrupt signal, and the input image signal comprising a plurality of pixels. A block dividing unit for dividing into block video signals, an input of a block video signal of the block dividing unit, an activity calculating unit for calculating an activity between these block video signals, and an output of the external interrupt detecting unit. In the section where at least the input video signal has the input of the external interrupt signal, the quantization width is determined independently of the activity, while in the section where the external interrupt signal is not input, the activity calculated by the activity calculating means is determined. Quantization width determining means for determining a quantization width according to And a quantizing means for quantizing the block video signal with a quantization width determined by the width determining means, and performing variable-length encoding on the signal generated by the quantizing means to output a compression-encoded sequence. It is characterized by.

According to a fourth aspect of the present invention, there is provided a video encoding apparatus comprising:
What is claimed is: 1. A video encoding apparatus for compressing and encoding an input video signal, comprising: a block dividing unit that divides the input video signal into block video signals composed of a plurality of pixels; and a block video signal of the block dividing unit. Activity calculating means for calculating an activity between block video signals, etc .; edge detecting means for detecting the presence or absence of an edge in a predetermined area of the input video signal; and irrelevant to the activity according to the output of the edge detecting means. A quantization width determining means for determining a quantization width according to the activity calculated by the activity calculating means, or a quantization width determined according to the activity calculated by the activity calculating means, and the block having the quantization width determined by the quantization width determining means. A quantizing means for quantizing a video signal, wherein an output of the quantizing means is subjected to variable-length encoding and compression encoding. And outputs a.

According to a fifth aspect of the present invention, there is provided a video encoding apparatus comprising:
What is claimed is: 1. A video encoding apparatus for compressing and encoding an input video signal, comprising: a block dividing unit that divides the input video signal into block video signals composed of a plurality of pixels; and a block video signal of the block dividing unit. Activity calculating means for calculating an activity between block video signals, and an edge image for determining whether or not the input video signal is an image including an edge based on a distribution of activities calculated by the activity calculating means. Detecting means, and inputting the output of the edge image detecting means,
At least in the section in which the input video signal is determined to be an edge image, the quantization width is determined independently of the activity, while in the section in which it is determined that the input video signal is not an edge image, the activity calculated by the activity calculating means is determined. A quantization width determining unit that determines a quantization width in accordance with the quantization width, and a quantization unit that quantizes the block video signal with the quantization width determined by the quantization width determination unit. It is characterized in that the output is variable-length coded and a compressed coded sequence is output.

According to a sixth aspect of the present invention, in the video encoding apparatus according to the first, second, third, fourth, or fifth aspect, the quantization width determining means may include an activity. When the quantization width is determined independently, the quantization width is determined to be a quantization width specified from the outside.

According to a seventh aspect of the present invention, in the video encoding device according to the first, second, third, fourth, or fifth aspect, the activity calculating means includes: The method is characterized in that a pixel-level correlation amount in a predetermined area in one frame or one field is measured.

According to an eighth aspect of the present invention, in the video encoding apparatus according to any one of the first, second, third, fourth and fifth aspects, the activity calculating means includes: The method is characterized in that a pixel-level correlation amount in a predetermined area between frames or fields is measured.

According to a ninth aspect of the present invention, in the video encoding apparatus according to the first aspect, the predetermined video pattern detecting means determines that a type of a color component of a predetermined area of the input video signal is smaller than a predetermined integer value K. It is characterized by detecting that the number is small.

According to a tenth aspect of the present invention, in the video encoding apparatus according to the first aspect, the predetermined video pattern detecting means includes a luminance component in a predetermined area of the input video signal, the luminance component being larger than a predetermined threshold X. It is characterized in that it is detected that the number of pixels having is equal to or greater than a predetermined integer value M.

According to an eleventh aspect of the present invention, in the video encoding apparatus according to the first aspect, the predetermined video pattern detecting means determines that the number of pixels whose luminance component in the predetermined area exceeds the predetermined number G1 is a predetermined integer N1. That is, the number of pixels smaller than the predetermined number G2 is detected to be equal to or larger than a predetermined integer value N2.

According to a twelfth aspect of the present invention, in the video encoding apparatus according to the first aspect, the predetermined video pattern detecting means detects that the input video signal is a video including subtitles. I do.

According to a thirteenth aspect of the present invention, in the video encoding apparatus according to the first aspect, the predetermined video pattern detecting means detects that the input video signal is a title screen or a menu screen. And

According to a fourteenth aspect of the present invention, in the video encoding apparatus according to the first aspect, the predetermined video pattern detecting means detects that the input video signal is a video including a computer graphics screen. It is characterized by.

According to a fifteenth aspect of the present invention, in the video encoding apparatus according to the first aspect, the predetermined video pattern detecting means detects that the input video signal is a video including animation. I do.

According to a sixteenth aspect of the present invention, in the video encoding apparatus according to the fourth aspect, the edge detecting means comprises:
It is characterized in that the presence or absence of an edge is detected based on the distribution of coefficients after the discrete cosine transform.

According to a seventeenth aspect of the present invention, in the video encoding apparatus according to the fourth aspect, the quantization width determining means determines that the number of areas determined to have an edge by the edge detecting means is within a predetermined area. If the value is equal to or more than the predetermined threshold F, the quantization width is determined independently of the activity, and if the value is less than the predetermined threshold F, the quantization width is determined according to the activity calculated by the activity calculating means. It is characterized by.

According to an eighteenth aspect of the present invention, in the video encoding apparatus according to the fifth aspect, the edge image detecting means has a distribution of the activity calculated by the activity calculating means other than a normal distribution in a predetermined area. In some cases, the image is detected as an edge image.

According to a nineteenth aspect of the present invention, in the video encoding apparatus according to the fifth aspect, the edge image detecting means determines a maximum value and a minimum value of the activity calculated by the activity calculating means in a predetermined area. When the difference is equal to or more than the predetermined threshold value X, the image is detected as an edge image.

According to a twentieth aspect of the present invention, in the video encoding apparatus according to the fifth aspect, the edge image detecting means determines that a maximum value of the activity calculated by the activity calculating means is equal to a predetermined threshold value X1 in a predetermined area.
If the minimum value is equal to or smaller than the predetermined threshold value X2, the image is detected as an edge image.

According to a twenty-first aspect of the present invention, in the video encoding apparatus according to the fifth aspect, the edge image detecting means determines that the number of activities having a predetermined threshold value G1 or more in a predetermined area is a predetermined integer value X1 or more. When the number of activities less than the predetermined threshold value G2 is equal to or more than a predetermined integer value X2, it is detected as an edge image.

According to a twenty-second aspect of the present invention, in the video encoding apparatus according to the fifth aspect, the edge image detecting means includes, in a predetermined area, one activity and M (M is 1 to 1) adjacent to the activity. If the difference between each of the activities and the number of activities exceeds the predetermined threshold T, the number of activities exceeds the predetermined number N, and the activity is detected as an edge image.

With the above arrangement, in the video encoding apparatus according to the first and the second aspects of the present invention, when at least the input video signal is a predetermined video pattern, the quantum coding determined independently of the activity is performed. While the variable length encoding process is performed using the quantization width, if the video pattern is not the predetermined video pattern, the variable length encoding process is performed using the quantization width determined according to the calculated activity. Therefore, when at least the input video signal is a predetermined video pattern such as a subtitle screen, the video after compression encoding does not become jagged or partially edge-blown video as in the related art. Video is obtained. If the input video signal is not a predetermined video pattern, a good video in which the activity is reflected can be obtained as in the conventional video after the compression encoding.

In the video encoding apparatus according to the present invention, when the input time code specifies a predetermined video pattern, the time code is determined irrespective of the activity. Since the variable-length encoding process is performed using the quantization width obtained in the above-described manner, in the same manner as described above, if at least the input video signal is a predetermined video pattern such as a subtitle screen, the video after the compression encoding is performed as in the related art. The image does not become jagged or partially edged, and a good image is obtained.

Further, in the video encoding apparatus according to the third and the third aspects of the present invention, in the video signal during the period in which the external interrupt signal is input, the quantization width determined independently of the activity is used. Since the variable-length encoding process is performed using the external interrupt signal in a predetermined video scene such as a subtitle screen, at least in the predetermined video scene, a video such as a subtitle screen after compression encoding is conventionally used. As described above, the image does not become jagged or partially edge-blown, and a good image can be obtained.

In addition, in the video encoding apparatus according to the fourth aspect and the dependent claims, a video signal having many edges is variable length using a quantization width determined independently of an activity. Since the encoding process is performed, in a video scene such as a subtitle screen including many edges, a video such as a subtitle screen after compression encoding is defined as a jagged or a partially edged video as in the related art. And good images can be obtained.

In addition, in the video encoding apparatus according to the present invention described in claim 5 and the dependent claims, the complexity of the image is determined based on the distribution of the activity, and the image including an edge such as an image with a complicated pattern is determined. In, the variable-length encoding process is performed using the quantization width determined independently of the activity, so that an image such as a subtitle screen including the edge is
A good image can be obtained without a jagged or partially blurred edge as in the related art.

In particular, in the video encoding apparatus according to the present invention, the activity is detected by measuring the correlation level of the pixel level in a predetermined area within one frame or one field of the input video signal. Can be reflected in the complexity of the image.

In the video encoding apparatus according to the present invention, the activity is measured and detected by measuring the pixel-level correlation in a predetermined area between frames or fields of the input video signal. Features can be reflected in the complexity of the image.

[0052]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS An embodiment of a video encoding apparatus according to the present invention will be described below with reference to the drawings.

(First Embodiment) FIG. 1 shows a configuration of a video encoding apparatus according to a first embodiment of the present invention.

In the figure, 101 is an input video signal, 1
02 is a block divider (block dividing means), 103 is a discrete cosine transformer, 104 is a quantizer (quantizing means), 105 is a variable length encoder, 106 is a used Q parameter, 107 is a compressed coded sequence, 108 is a code amount measuring device,
109 is an assigned coded value, 110 is a code amount controller, 1
11 is a Q parameter, 112 is an activity calculator (activity calculating means), 113 is a Q parameter modulator, 114 is a modulation Q parameter, 115 is a switch (quantization width determining means), 116 is a predetermined video pattern detector (predetermined Image pattern detecting means) 117 is a switching signal.

Hereinafter, the video encoding apparatus of FIG. 1 will be described in detail. The input video signal 101 is a predetermined video pattern detector 1
16 and the block divider 102. The predetermined video pattern detector 116 receives the input video signal 101, determines whether or not the frame is a predetermined video pattern in units of one frame or one field.
The input side of the switch 115 is connected to the Q parameter modulator 113
, And switches the input side of the switch 115 to the Q parameter 111 output of the code amount controller 110 in the case of a predetermined pattern.

The output of the block divider 102 is input to an activity calculator 112, and the activity calculator 112 obtains the activity of each predetermined area based on pixels within a frame or pixels between frames.

On the other hand, the output of the block divider 102 is input to the discrete cosine transformer 103, converted into frequency data, and input to the quantizer 104.

Here, the quantizer 104 quantizes the frequency data in the predetermined area with the determined quantization width, that is, the used Q parameter 106.
As described above, the modulated Q parameter 114 is used when the image is not a predetermined image pattern, and the unmodulated Q parameter 111 is used when the image is a predetermined image pattern.

The variable length encoder 105 includes a quantizer 104
Is subjected to variable-length coding to generate a compressed coded sequence 107.

At this time, the code amount measuring device 108 measures the generated code amount of the compression coded sequence 107, and the measured value and the assigned coded value 109 are input to the code amount controller 110.
The code amount controller 110 detects a difference between the two input values and calculates a Q parameter 111 based on, for example, Equation 2 in order to control the generated code amount in the next predetermined area.

The predetermined video pattern includes a screen including subtitles, a color bar, a still image, a flashing image, a test pattern, a title screen, an image created by computer graphics, an image including animation, and the like.
The image may be an image having a complicated pattern with little movement, or an image including a pattern including an edge locally.

A video having a high activity generally has a high coding difficulty, but the generated coding noise is hard to recognize. Conversely, coding noise tends to be more noticeable in images with low activity.

Therefore, by controlling the quantization parameter in accordance with the size of the activity, it is possible to suppress the deterioration of the image quality, and to improve the image quality of the entire video scene.

That is, the code amount controller 110 determines the Q parameter 111 so that the variable length encoder 105 can encode the video signal at an encoding rate according to the scene.

(First Example of Predetermined Video Pattern Detector) Next, as an example of the predetermined video pattern detector 116, a configuration for performing caption detection using an activity in one screen will be described with reference to FIG.

In FIG. 6, reference numeral 601 denotes an input video signal;
02 is a luminance component / color component decomposer, 603 is a luminance component, 6
04 is a color component, 605 is an activity detector, 606
Is a predetermined area activity, 607 is a predetermined area detector,
608 is a predetermined area control signal, 609 is a predetermined area number, 6
10 is a reference area detector, 611 is a reference area list, 61
2 is a reference region control signal, 613 is an activity reference region averager, 614 is a reference region activity average value,
615 is an activity difference value calculator, 616 is an activity difference value, 617 is an activity difference value threshold comparator, 618 is an activity threshold, 619 is a luminance component determination result, 620 is a color component type checker, 621 is the number of color component types, 622 is a color component threshold value judging device, 623 is a color component type number threshold value, 624 is a color component judgment result, 625 is a judgment result integrator, 626 is a predetermined pattern judgment result, 627 is a switch controller, and 628 is a switch signal. .

FIG. 7 shows the predetermined image pattern detector 1 shown in FIG.
FIG. 16 is a diagram for explaining the configuration of the 16th embodiment. The activity in one frame (for example, 720 × 480 pixels) is divided into predetermined regions (for example, 16 × 16 pixels).
Activities are obtained, and these are represented by numerical values. Reference numeral 705 denotes an area of one frame, 701 denotes an upper left corner area (for example, 48 × 48 pixels), 702 denotes an upper right corner area (48 × 48 pixels), and 703 denotes a lower left corner (for example, 48 × 48 pixels).
A region 704 is a lower right corner (for example, 48 pixels).
x48 pixels).

In the subtitle image, the activities at the four corners in one frame are almost the same, and only the central part of the screen changes. Moreover, since the color components are about two or three colors of black and white, it is considered that there is almost no variation in the color components.

Therefore, as shown in FIG. 7, the activity of a certain frame is calculated in the frame, the average value of the activities in the four corner areas 701, 702, 703, and 704 is calculated, and the four average values are calculated. If the difference between the minimum value and the maximum value is less than or equal to a certain threshold value α (α is a predetermined constant empirically determined), the frame is presumed to be an image in which the fluctuation of the activity is small around the screen.
The frame can be determined to be a predetermined pattern (caption).

For example, in FIG. 7, the average value of the activity in the area 701, the area 702, and the area 703 is “1.00”, and the average value of the activity in the area 704 is “1.
1875 ”, the difference between the minimum and maximum values of the average value of the activity in the four areas is“ 0.1875 ”, and if the threshold α is“ 3.00 ”, the amplitude of the activity in the four corner areas is Since the value is smaller than the activity threshold, it can be determined that the image is a caption image.

In FIG. 6, an input video signal 601 is input to a luminance component / color component
03, separated into color components 604.

The luminance component 603 is input to the in-screen activity detector 605. The activity detector 605 calculates an activity for each predetermined area in a frame using, for example, a pixel level correlation amount in a pixel in one frame, and outputs the activity as an activity 606 of the predetermined area.

The predetermined area detector 607 receives the input video signal 601 and outputs the in-screen activity detector 605.
The output timing of the activity 606 to be calculated is controlled. The reference area detector 610 detects four corner areas in one frame as reference areas. The in-screen activity detector 605 outputs the predetermined area activity 606 to the activity reference area averager 613, and the activity reference area averager 613 calculates a reference area activity average value 614. At this time, the reference region detector 608 inputs the predetermined region number 609 output from the predetermined region detector 607, and refers to a reference region list 611 indicating which pixel region is the reference region, and for each reference region. It outputs a reference area control signal 612 to control the calculation timing of the activity reference area averager 613.
For example, at the start point of each reference region, the calculation register for calculating the average value of the activity of the activity reference region averager 608 is reset, and the timing for obtaining the reference region activity average value 614 which is the average value of the activity for each reference region is determined. Control.

The reference area activity average value 614 is
The average value of the activities in the four reference areas is stored in the activity difference value calculator 615, and the difference between the minimum value and the maximum value is calculated. That is, the amplitude of the activity in the four corner areas in one frame is calculated,
It is output as the activity difference value 616.

Here, the activity difference threshold comparator 611 compares whether or not the activity difference 616 is below the activity threshold 618. If the activity difference is below the activity threshold 618, the subtitle is determined based on the luminance component. For example, the luminance component determination result 619 is a 1-bit signal, and is output as the value “1” at the end of one frame. On the other hand, if it is not below, it is determined from the luminance component that it is not a caption, so the luminance component determination result 619 is set to the value “0”.
Output as

The activity threshold 618 is a predetermined constant and can be empirically given from outside before encoding.

The color component 603 is the color component type checker 6
20 and outputs, for example, the number of color component types 621 indicating how many color components exist in one frame. The color component threshold value determiner 622 determines that the number of color component types 621 falls below the color component type number threshold value 623. The subtitles are compared if they are below, and if they are below the captions, for example, the color component determination result 62
4 is a 1-bit signal, and is output as the value “1” at the end of one frame. On the other hand, if it is not below, it is determined from the color component that it is not a caption, and the color component determination result 624 is output as a value “0”. When checking the type of the color component, the color component type checker 620 does not recognize the color component that is very infrequent as a color component, and determines the type of the color component that is more frequent than the predetermined number J. Number of color component types 621
It is good.

The color component threshold value 623 is a predetermined constant, and may be empirically provided from outside before encoding.

The judgment result integrator 625 receives the luminance component judgment result 619 and the color component judgment result 624 as inputs. For example, when both the luminance component judgment result 619 and the color component judgment result 624 are “1”, a predetermined pattern , And outputs the predetermined pattern determination result 626 as “1”. Otherwise, the predetermined pattern determination result 6
26 is output as "0".

The switch controller 627 outputs a switching signal 628 according to the predetermined pattern determination result 626 as, for example, “1” when the image is a predetermined image pattern, and “0” when the image is not a predetermined image pattern.

When it is desired to discriminate at least a video scene which is a subtitle, the activity threshold 618 may be changed to a large value, or the color component threshold 623 may be set large to loosen the criterion. The judgment result integrator 62
5 is a luminance component determination result 619 and a color component determination result 624
If any of the above is “1”, a predetermined pattern (caption)
And the predetermined pattern determination result 626 may be output as “1”.

In order to simplify the determination, it is possible to determine whether the pattern is a predetermined pattern using only one of the luminance component and the color component. Any method may be used as long as it can be determined that the image pattern is a small but complicated image pattern or an image pattern including an edge.

The area to be selected as the reference area does not necessarily need to be the four corners of one frame. For example, the upper part or the lower part of the screen is selected. The determination may be made, and even when the four corners are set as the reference area, the four corners need not have the same number of pixel areas.

Incidentally, the method of discriminating the subtitle screen may be pattern matching with the subtitle, or if the portion where the subtitle appears is clear, the change of the video pattern activity in that portion may be greatly changed. May be discriminated, and any discrimination method may be used.

(Second Example of Predetermined Image Pattern Detector) Next, as a second example of the predetermined image pattern detector 116,
An example of the characteristics of subtitles and title screens utilizing the fact that the types of color components of an input video signal are small.

FIG. 8 shows the configuration of the predetermined image pattern detector of the second example.

In the figure, reference numeral 801 denotes an input video signal;
02 is a luminance component / color component separation circuit, 803 is a color component, 8
04 is a color component type checker, 805 is the number of color component types, 80
6 is a color component type number threshold, 807 is a color component threshold value judging device, 8
08 is a color component determination result, 809 is a switch controller, 8
10 is a switching signal.

An input video signal 801 is input to a luminance component / color component separation circuit 802, and in particular, extracts only a color component 803.

The color component 803 is the color component type checker 8
For example, a color component type number 805 indicating the number of color components present in one frame is output, and the color component threshold value determiner 807 determines that the color component type number 805 is lower than the color component type number threshold value 806. It is determined whether or not the color component is determined to be a subtitle.
8 is a 1-bit signal, and is output as the value "1" at the end of one frame. On the other hand, if it is not below, it is determined from the color component that it is not a subtitle.
08 is output as “0”. The color component type checker 8
In reference numeral 04, when examining the types of color components, a color component with a very low frequency is not recognized as a color component, and a color component type with a frequency higher than a predetermined number J may be set as the color component type 805. .

When the color component determination result 808 is “1” in accordance with the color component determination result 808, the switch controller 809 sets the Q parameter having no activity modulation, “0”.
At the time of the switching, the switching signal 810 is switched so as to select the Q parameter having the activity modulation.

(Third Example of Predetermined Image Pattern Detector) Next, as a third example of the predetermined image pattern detector 116,
One example of the characteristics of a subtitle and a title screen that utilizes a point in which a luminance component of an input video signal has an extremely bright portion is given.
The image refers to, for example, an image in which white characters scroll, an image in which a title appears, an image in which diamonds shine, an explosion scene, a starry sky, a shining water surface, etc., in the center of a screen with black or blue back. . Examples include images in which logos and marks of a production company created by computer graphics are embedded, and images in which a large number of people wearing white clothes move on a dark screen with animation.

FIG. 9 shows an example in which the distribution of the luminance component of the video signal is biased toward the bright side.

The horizontal axis 901 in FIG.
Denotes the number, 903 denotes a luminance component distribution curve, and 904 denotes a luminance component threshold X.

As shown in FIG. 9, when the distribution of the luminance component is examined in a predetermined area of the screen, images such as white letters and stars, humans dressed in white clothes, sandy beaches, mirrors, grass, glittering water surfaces, and the like are obtained. , The luminance component shows a large value in a specific area.

Therefore, in a predetermined area of the screen, the number indicating a luminance component larger than the luminance component threshold X in FIG. 9 is counted, and if the number is equal to or larger than a predetermined integer M, it may be determined that the image pattern is a predetermined image pattern. .

FIG. 10 is a configuration diagram of the predetermined image pattern detector of the third example.

In FIG. 10, 1001 is an input video signal, 1002 is a luminance component / color component separation circuit, 1003 is a luminance component, 1004 is a luminance component threshold, 1005 is a counter, 1006 is a luminance component threshold determiner, and 1007 is a constant M , 1008 is a distribution determiner, 1009 is a luminance component determination result, 1010 is a switch controller, and 1011 is a switch signal.

An input video signal 1001 is input to a luminance component / color component separation circuit 1002, and in particular, a luminance component 1003
Take out only.

The luminance component 1003 is input to a luminance component threshold value judging unit 1006, and it is determined whether the luminance component 1003 exceeds the luminance component threshold value 1004. If the luminance component 1003 exceeds the luminance component 1003, the count value of the counter 1005 is increased by one. . The distribution determiner 1008 determines that the value of the counter 1005 is a constant M (10
07) is determined for each predetermined area, and is output as a luminance component determination result 1009. The luminance component determination result 1009 is, for example, a 1-bit signal, and a value “1” indicates a predetermined video pattern,
In the case of "0", it may be determined that it is not a predetermined video pattern. In accordance with this, the switch controller 1010 selects a Q parameter having no activity modulation when it is a predetermined video pattern, and selects a Q parameter having activity modulation when it is not a predetermined video pattern, and performs quantization. The switching signal 1011 may be set so as to perform the switching. The predetermined area may be one frame or a plurality of pixels.

The count value of the counter 1005 is reset to zero for each predetermined area.

(Fourth Example of Predetermined Image Pattern Detector) Next, as a fourth example of the predetermined image pattern detector 116,
An example of a characteristic of a subtitle or a title screen that utilizes a point in which a luminance component of an input video signal has an extremely bright portion and a dark portion. The image refers to an image in which white characters scroll, an image in which a title appears, or an image in which the headlights of a car are illuminated countlessly in the dark in the center of a black or blue-backed screen. Many luminous stars in the dark universe created by computer graphics, images embedded with production company logos and marks, and many animations in a dark screen with animation An image is another example.

FIG. 11 shows an example in which the distribution of the luminance component of the video signal has a bias on the bright side and the dark side.

The horizontal axis 1101 in FIG.
102 denotes the number, 1103 denotes a luminance distribution curve, 11
04 is the luminance component threshold X1, 1105 is the luminance component threshold X2
It is.

As shown in FIG. 11, when the distribution of the luminance component is examined in a predetermined area of the screen, when an image in which a white character, a star, or a person wearing white clothes moves finely in a dark background, a specific It is clear that the luminance component shows a large value in the region and the other specific region has a portion where the background is dark and the luminance component locally shows a small value.

Therefore, in a predetermined area of the screen, the number of luminance components larger than the luminance component threshold value X1 in FIG. 11 is counted, and the number of the luminance components is equal to or larger than the predetermined integer N1 and smaller than the luminance component threshold value X2. Is greater than or equal to a predetermined integer N2, it may be determined that the pattern is a predetermined video pattern.

FIG. 12 is a configuration diagram of the predetermined image pattern detector of the fourth example.

In FIG. 12, reference numeral 1201 denotes an input video signal; 1202, a luminance component / color component separation circuit; 1203, a luminance component; 1204, a luminance component threshold X1; 1205, a second luminance component threshold X2; Counter, 1
207 is a second counter, 1208 is a first luminance component determiner, 1209 is a second luminance component determiner, 1210 is a constant N1, 1211 is a constant N2, 1212 is a distribution determiner,
1213 is a luminance component determination result, 1214 is a switch controller, and 1215 is a switch signal.

An input video signal 1201 is input to a luminance component / color component separation circuit 1202, and in particular, a luminance component 1203
Take out only.

The luminance component 1203 is input to a first luminance component determiner 1208 and a second luminance component determiner 1209. The first luminance component determiner 1208 outputs the luminance component 1
It is determined whether or not 203 exceeds the luminance component threshold value X1 (1204).
Is incremented by one. Similarly, the second luminance component determiner 1209 determines that the luminance component 1203 is the luminance component threshold X2
(1205) The value smaller than (1205) is determined, and if smaller, the count value of the second counter 1307 is increased by one.

The distribution judging unit 1212 has the first counter 12
06 is greater than or equal to the constant N1 (1210),
And the counter value of the second counter 1207 is a constant N2
(1211) When it is equal to or more than the predetermined pattern, it is determined that the signal is a predetermined pattern. For example, the luminance component determination result 1213 is a 1-bit signal. Set to “0”.

Accordingly, the switch controller 1214
If the switching signal 1215 is set so as to select a Q parameter having no activity modulation at the time of the predetermined video pattern, and to select a Q parameter having activity modulation at the time of not the predetermined video pattern, and perform quantization. good. The predetermined area may be one frame or a plurality of pixels.

The count values of the first counter 1206 and the second counter 1207 are reset to zero for each predetermined area.

Here, an example of the configuration for detecting four predetermined video patterns has been described. The predetermined video patterns are a screen including subtitles, a color bar, a still image, a flashing image, a test pattern, a title screen, and a computer. Images with complicated patterns even with little movement, such as images created by graphics, images with animation,
Or an image including a pattern that includes an edge locally,
Either is acceptable.

For example, in the title screen or the menu screen, the color components at the four corners are considered to be substantially the same color as in the subtitle screen. If the number is not more than the predetermined number, it may be determined that the screen is a title screen. Further, since an image created by computer graphics tends to have a specific number of color components due to the limitation of the number of simultaneous colors of the computer, etc., it is checked whether the number of color components is equal to or less than a predetermined number. If the number is equal to or less than the number, the image may be determined to be an image created by computer graphics. Also, since the animation is a solid painting of a picture, a predetermined area may be taken in the horizontal or vertical direction, the number of the color components may be checked, and if the number is less than the predetermined number, it may be determined that the animation is an animation. The method of determining the predetermined video pattern may be any.

(Second Embodiment) FIG. 2 shows the configuration of a video encoding apparatus according to a second embodiment of the present invention.

In the figure, reference numeral 201 denotes an input video signal;
02 is a block divider, 203 is a discrete cosine transformer,
204 is a quantizer, 205 is a variable length encoder, 206 is a used Q parameter, 207 is a compressed coded sequence, 208 is a code amount measuring device, 209 is an assigned coded value, 210 is a code amount controller, and 211 is a code amount controller. A Q parameter, 212 is an activity calculator, 213 is a Q parameter modulator, 214 is a modulation Q parameter, and 215 is a switch. Also, 2
16 is a time code, 217 is a time code list, 2
Reference numeral 18 denotes a switching signal, and 219, a predetermined video section detector (predetermined video section detection means).

In this embodiment, as compared with the first embodiment, a section in which the modulation of the activity is cut off is not specified from the signal characteristics of the input video, but the contents of the video source are previously known by a person. In this case, by simply specifying the approximate time code where an image with many subtitles and edges is present, quantization with a quantization width that does not modulate the activity and variable-length encoding can be performed.

In the video coding apparatus shown in FIG. 2, an input video signal 201 is input to a block divider 202. The time code 21 is synchronized with the input video signal 201.
6 is input. The time code 216 corresponds to V in the video signal.
It may be an ITC or an input of a different system from the input video signal, for example, LTC or ASTC. In addition, VIT
C is time code information written in the vertical blanking or specific horizontal blanking of the video signal, and it is assumed that one or more values are uniquely written in each frame. For example, LTC (Loading Ti
meCode) is read from a different track from the video signal of the tape track during high-speed playback, and the ASTC (AcouStic Tim
eCode) is read from a different track from the video signal of the tape track during low-speed playback.

The predetermined video section detector 219 inputs the time code 216 and the time code list 217. The time code list 217 is a list indicating a start position and an end position of a time code of a subtitle scene for which modulation of an activity is not desired. Predetermined video section detector 219
Is the switching signal 2 when the time code 216 is included in the section specified by the time code list 217.
The value of 18 is switched and the input of the switch 215 is changed to the unmodulated Q parameter 211 of the activity.
, If not, switch signal 21
8 and switches the input of the switch 215 to the modulation Q parameter 214 which is the output of the Q parameter modulator 213.
Switch to the side. That is, the predetermined video section detector 219
Determines whether or not the frame is a predetermined video pattern in units of one frame or one field based on the time code 216. If the video pattern is not the predetermined video pattern, the input side of the switch 215 is modulated by the output of the Q parameter modulator 213. Switching to the Q parameter 214 side, and in the case of the predetermined video pattern, switching the input side of the switch 215 to the Q parameter 211 side which is the output of the code amount controller 210.

The output of the block divider 202 is input to an activity calculator 212, and the activity calculator 212 obtains an activity for each predetermined area based on pixels in a frame or pixels between frames.

On the other hand, the output of the block divider 202 is input to the discrete cosine transformer 203, becomes frequency data, and is input to the quantizer 204.

The quantizer 204 quantizes the frequency data of the predetermined area using the determined quantization width, that is, the used Q parameter 206. If the used Q parameter 206 is not a predetermined video pattern as described above, the quantization is performed. The Q parameter 214 is used. If the image pattern is a predetermined video pattern, the Q parameter 211 that is not modulated is used.

The variable-length encoder 205 includes a quantizer 204
Is subjected to variable length encoding to generate a compressed encoded sequence 207.

At this time, the code amount measuring device 208 measures the generated code amount of the compressed coded sequence 207, receives the measured value and the assigned coded value 209 as input, detects the difference between the measured value and the assigned coded value 209, and detects the difference between the two. Q parameter 2 for controlling the amount of generated code
11 is calculated by the code amount controller 210 based on, for example, the equation (2).

The predetermined video pattern includes a screen including subtitles, a color bar, a still image, a flashing image, a test pattern, a title screen, an image created by computer graphics, an image including animation, and the like.
The image may be an image having a complicated pattern with little movement or an image including a pattern including a local edge.

A video having a high activity generally has a high coding difficulty, but the generated coding noise is hard to recognize. Conversely, coding noise tends to be noticeable in an image with low activity.

Therefore, by controlling the quantization parameter in accordance with the size of the activity, the deterioration of the image quality can be suppressed, and the image quality of the entire video scene can be improved.

That is, the code amount controller 210 determines the Q parameter 211 so that the variable length encoder 205 can encode the video signal at an encoding rate according to the scene.

Note that the time code list 21 is set to be wider than a section which is considered to include a predetermined video pattern such as a subtitle scene.
7 may be set to determine the quantization width without modulating the activity, or may be set to be narrower than a section including a predetermined video pattern such as a subtitle scene.

(Third Embodiment) FIG. 3 shows a configuration of a video encoding device according to a third embodiment of the present invention.

In the figure, reference numeral 301 denotes an input video signal, 3
02 is a block divider, 303 is a discrete cosine transformer,
304 is a quantizer, 305 is a variable length coder, 306 is a used Q parameter, 307 is a compression coded sequence, 308 is a code amount measuring device, 309 is an assigned coded value, 310 is a code amount controller, and 311 is a code amount controller. Q parameter, 312 is an activity calculator, 313 is a Q parameter modulator, 314 is a modulation Q parameter, 315 is a switch, 316 is an external interrupt signal, 317 is an external interrupt detector (external interrupt detection means), and 318 is a switch signal. It is.

In the video encoding apparatus shown in FIG. 3, an input video signal 301 is input to a block divider 302. In the present embodiment, an external interrupt signal 316 that is not necessarily synchronized with the input video signal 301 is input. As the external interrupt signal 316, information written in the video signal may be read.
The data may be read from a system different from 01. Further, for example, when a human presses a button or the like, the external interrupt signal 316 is output.
May be turned on.

The external interrupt detector 317 inputs the external interrupt signal 316, and when the external interrupt signal 316 is on, that is, in a subtitle scene where it is not desired to modulate the activity, the switching signal 318 is switched, and The input side of 315 is switched to the Q parameter 311 side to cut off the modulation of the activity. When the external interrupt signal is off, the input side of the switch 315 is switched to the side of the modulation Q parameter 314 which is the output of the Q parameter modulator 313.

That is, the external interrupt detector 317 determines whether or not there is an external interrupt in a predetermined section unit such as one frame, one field, or a plurality of pixels or a plurality of frames based on the external interrupt signal 316. When the external interrupt signal 316 is off, the switch 31
5 is switched to the modulation Q parameter 314, which is the output of the Q parameter modulator 313, and when the external interrupt signal 316 is on, that is, for example, in a subtitle scene, the input of the switch 315 is connected to the code amount controller. Switching is made to the Q parameter 311 side, which is the output of 310.

The output of the block divider 302 is input to an activity calculator 312, and the activity calculator 312 obtains the activity of each predetermined area based on the pixels in the frame or the pixels between the frames.

On the other hand, the output of the block divider 302 is input to the discrete cosine transformer 303, becomes frequency data, and is input to the quantizer 304.

Here, the quantizer 304 quantizes the frequency data of the predetermined area using the determined quantization width, that is, the used Q parameter 306. When the used Q parameter 306 is not a predetermined video pattern as described above, Uses a modulated Q parameter 314, and uses a non-modulated Q parameter 311 for a predetermined video pattern.

The variable length encoder 305 includes a quantizer 304
The variable-length coding is performed on the quantized code output as the compressed code sequence 307.

At this time, the code amount measuring unit 308 measures the generated code amount of the compression coded sequence 307, and the code amount controller 310 inputs the measured value and the assigned coded value 309, and detects the difference. Then, the Q parameter 311 is calculated based on, for example, Equation 2 in order to control the generated code amount in the next predetermined area.

Note that the predetermined video pattern includes a screen including subtitles, a color bar, a still image, a flashing image, a test pattern, a title screen, an image created by computer graphics, an image including animation, and the like.
The image may be an image having a complicated pattern with little movement or an image including a pattern including a local edge.

A video having a high activity generally has a high coding difficulty, but the generated coding noise is hard to recognize. Conversely, coding noise tends to be noticeable in an image with low activity.

Therefore, by controlling the quantization parameter in accordance with the size of the activity, the deterioration of the image quality can be suppressed, and the image quality of the entire video scene can be improved.

That is, the code amount controller 310 determines the Q parameter 311 so that the variable length encoder 305 can encode the video signal at an encoding rate corresponding to the scene.

The unit for detecting whether the external interrupt signal 316 is on or off may be any unit, such as one frame, one field, a plurality of pixels, a plurality of frames, or any other predetermined section.

Note that the activity modulation may be turned off when the delay time of a plurality of frames elapses after the external interrupt signal 316 is turned on.

(Fourth Embodiment) FIG. 4 shows the configuration of a video encoding apparatus according to a fourth embodiment of the present invention.

In the figure, reference numeral 401 denotes an input video signal;
02 is a block divider, 403 is a discrete cosine transformer,
404 is a quantizer, 405 is a variable length coder, 406 is a used Q parameter, 407 is a compressed coded sequence, 408 is a code amount measuring device, 409 is an assigned coded value, 410 is a code amount controller, and 411 is a code amount controller. Q parameter, 412 is an activity calculator, 413 is a Q parameter modulator, 414 is a modulation Q parameter, 415 is a switch, 416 is an edge detector (edge detecting means), and 417 is a switch signal.

In the video encoding apparatus shown in FIG. 4, an input video signal 401 is supplied to an edge detector 416 and a block divider 4.
02 is input. The edge detector 416 receives the input video signal 401 as input, determines whether there is an edge in a predetermined area,
If it is determined that there is no edge, the switch 4
15 is switched to the modulation Q parameter 414 side which is the output of the Q parameter modulator 413, and when it is determined that there is an edge, the input side of the switch 415 is switched to the Q parameter which is the output of the code amount controller 410. Switch to 411 side.

The output of the block divider 402 is input to an activity calculator 412, and the activity calculator 412 obtains the activity of each predetermined area based on the pixels in the frame or the pixels between the frames.

On the other hand, the output of the block divider 402 is input to the discrete cosine transformer 403, becomes frequency data, and is input to the quantizer 404.

Here, the quantizer 404 quantizes the frequency data in the predetermined area using the determined quantization width, that is, the used Q parameter 406, but determines that the used Q parameter 406 has no edge as described above. Modulation Q
Using the parameter 414, if it is determined that the image has an edge, the Q parameter 411 that is not modulated is used.

The variable length coder 405 is
Is subjected to variable length encoding to generate a compressed encoded sequence 407.

At this time, the code amount measuring device 408 measures the generated code amount of the compressed coded sequence 407, and the code amount controller 410 inputs the measured value and the assigned coded value 409, and detects the difference. Then, the Q parameter 411 is calculated based on, for example, Equation 2 in order to control the generated code amount in the next predetermined area.

Images having edges include screens including subtitles, color bars, still images, flashing images, test patterns, title screens, images created by computer graphics, images including animations, and the like.
The image may be an image having a complicated pattern with little movement or an image including a pattern including a local edge.

A video having high activity generally has a high coding difficulty, but generated coding noise is hard to recognize. Conversely, coding noise tends to be noticeable in an image with low activity.

Therefore, by controlling the quantization parameter in accordance with the size of the activity, the deterioration of the image quality can be suppressed, and the image quality of the entire video scene can be improved.

That is, the code amount controller 410 determines the Q parameter 411 so that the variable length encoder 405 can encode the video signal at an encoding rate according to the scene.

The unit for determining the presence or absence of an edge may be any of a plurality of pixels, one frame, one field, or a plurality of frames.

The switching timing for determining whether to use the quantization width without activity modulation or to use the quantization width with activity modulation based on the determination of the presence or absence of an edge is a unit for determining the presence or absence of an edge. Each unit may be used, or a unit in which a plurality of regions for determining the presence or absence of an edge is collected, a frame unit, a field unit, a plurality of frame units, or the like.

For example, assuming that the switching timing is a frame unit, the number of areas for determining the presence / absence of an edge is, for example, 1350, and a frame determined to be an area having a threshold F or more (eg, 1000) or more is a frame. The quantization width may use a quantization width without modulation of the activity, and in a frame other than that, the quantization width may use a quantization width with modulation of the activity.

(First Example of Edge Detector) Next, a first example of the edge detector 416 will be described.

FIG. 13 shows a change in the luminance component in the horizontal direction of an image having an edge.

FIG. 21A shows a screen 1 which has been blacked out.
A white character 1302 "A" is drawn on the entire screen of the image 301, and FIG. 2B shows the luminance component by cutting FIG. 1A by a dotted line 1303. FIG.

As apparent from FIG. 13B, the luminance component 1304 has two sharply changing peaks exceeding the threshold value α1305. When the pattern in the screen has an edge, one or more steep changes occur in any of the horizontal direction, the vertical direction, and the oblique direction of the luminance component or the color component. If such a phenomenon can be measured, the presence or absence of an edge can be determined.

In the first example of the edge detector 416, when a luminance component is extracted in a horizontal direction at a certain portion of one screen,
A case where an image is determined to be an edge image when a change equal to or more than a certain threshold value X is equal to or more than a predetermined number N will be described.

FIG. 14 is a configuration diagram of a first example of the edge detector 416.

In the figure, reference numeral 1401 denotes an input video signal;
1402 is a luminance component / color component decomposer, 1403 is a luminance component, 1404 is a luminance component memory, 1405 is a luminance component difference device, 1406 is a luminance component threshold value, 1407 is a luminance component difference value, 1408 is a luminance component difference judgment device, 1409 Is a luminance component determination result, 1410 is a color component, 1411 is a color component memory, 1412 is a color component differentiator, 1413 is a color component threshold, 1414 is a color component difference value, 1415 is a color component difference determiner, and 1416 is color component determination. As a result, 1417 is an edge luminance component counter, 1418 is a luminance component edge determiner, 1
419, a luminance component edge reference value; 1420, an edge color component counter; 1421, a color component edge determiner;
Is a color component edge reference value, 1423 is a judgment result integrator, 1
424 is a switch controller and 1425 is a switch signal.

An input video signal 1401 is input to a luminance component / color component decomposer 1402 and is decomposed into a luminance component 1403 and a color component 1410.

The brightness component 1403 is stored in the brightness component memory 14.
04 and the luminance component differentiator 1405, which subtracts the luminance component of the predetermined region stored in the luminance component memory 1404 from the luminance component 1403 and subtracts the result from the luminance component 1403. As 1407,
Output to the luminance component difference determiner 1407.

When the luminance component difference value 1407 is larger than the luminance component threshold value 1406, the luminance component difference judgment unit 1408 increments the count value of the edge luminance component counter 1417 by one. In addition, the luminance component edge determiner 1418 provides an edge luminance component counter 14 for each predetermined area.
Looking at the count value of No. 17, the luminance component edge reference value 141
9 is determined, and if it is larger, it is determined that the image has an edge.
For example, 9 is set to a value “1”.

Similarly, a luminance component / color component decomposer 1402
The color component 1410 separated by
And the color component differentiator 1412
Reference numeral 412 denotes a color component memory 1411 from a color component 1410.
Is subtracted from the color component one predetermined region before, and the result is output to the color component difference determination unit 1415 as a color component difference value 1414.

When the color component difference value 1414 is larger than the color component threshold value 1413, the color component difference judgment unit 1415 increments the count value of the edge color component counter 1420 by one. The color component edge determiner 142
1 is determined for each predetermined area by checking the count value of the edge color component counter 1420 and determining whether or not the count value is larger than the color component edge reference value 1422. If the count value is larger, it is determined that the image has an edge. The color component determination result 1416 is set to, for example, a value “1”.

The judgment result integrator 1423 receives the luminance component judgment result 1409 and the color component judgment result 1416 as inputs and, for example, when both the luminance component judgment result 1409 and the color component judgment result 1416 are “1”, It is determined that the image is an edge image, and the determination result is output to the switch controller 1424 as a value “1”, for example.

The switch controller 1424 selects a Q parameter having no activity modulation if it is a predetermined video pattern according to the output of the determination result integrator 1423, and has activity modulation if it is not the predetermined video pattern. The switching signal 1425 is set so that the Q parameter is selected and quantized. The predetermined area may be one frame or a plurality of pixels.

The count values of the edge luminance component counter 1417 and the edge color component counter 1420 are reset for each predetermined area.

In the edge detector 416 of the first example, the luminance component determination result 1411 and the color component determination result 1411
12 is a decision result integrator 1
When 423 is performed, the image is determined to be an edge image. However, when one of the luminance component determination result 1411 and the color component determination result 1412 is a determination result having an edge, the image is determined to be an edge image. It may be determined.

In the edge detector 416 of the first example, both the luminance component and the color component are used. However, the determination may be performed using only the luminance component, or the determination may be performed using only the color component. May be.

(Second Example of Edge Detector) Next, a second example of the edge detector 416 will be described.

Taking the change in the luminance of an image having an edge as an example, if the difference between the luminance components of the adjacent pixels is larger than a certain threshold, the pixel can be determined to be an edge, and if the determination area is small, the pixel is small. The change can be grasped finely. In other words, as shown in FIG. 15, the edge type is (a) an image area in which the edge runs in the horizontal direction (horizontal type), and (b)
Image area with vertical edges running (vertical type),
It is considered that either (c) a type in which the edge runs diagonally to the upper left (upward diagonal type), or (d) a type in which the edge runs diagonally to the upper right (upward diagonal). Of course, a color component may be used as a reference.

For example, when comparing the luminance components, the area 1
In 501, 1503, 1505, and 1507, the luminance level is high and there are many bright pixels.
4, 1506, and 1508 assume that there are many dark pixels.

Incidentally, those which do not correspond to any of FIGS. 15 (a), (b), (c) and (d) may be of a type having no edge.

Here, an example in which only the horizontal type and the vertical type are discriminated for the sake of simplicity will be described. However, there are four types of patterns to be determined: a horizontal type, a vertical type, a diagonally rising left, and a diagonally rising right. Good, more classification types may be added, and for example, only diagonal types may be determined.

FIG. 16 is a configuration diagram of a second example of the edge detector 416.

In the figure, reference numeral 1601 denotes an input video signal;
1602 is a luminance component / color component decomposer, 1603 is a luminance component, 1604 is a predetermined area blocking unit, 1605 is an upper surface average value calculator, 1606 is a lower surface average value calculator, 160
7 is a left side average value calculator, 1608 is a right side average value calculator,
1609, 1610, 1611 and 1612 are average values of luminance components, 1613 and 1614 are differentiators, 1615 and 16
16 is a difference value, 1617 is a horizontal type determination threshold, 1618 is a horizontal type determiner, 1619 is a vertical type determination threshold, 1620 is a vertical type determiner, 1621 is a determination result integrator, 1622 is a switch controller, and 1623 is a switch. Signal, 1624
Is a counter and 1625 is a threshold X.

The input video signal 1601 is input to a luminance component / color component decomposer 1602, and is decomposed into a luminance component 1603 and a color component.

The luminance component 1603 is divided by a predetermined area blocking unit 1604 into a predetermined area, for example, a 16 × 16 pixel area, and an upper surface average value calculator 1605, a lower surface average value calculator 1606, a left surface average value calculator 1607, and a right surface average It is input to the value calculator 1608.

[0187] The upper surface average value calculator 1605 is shown in FIG.
An average value 1609 of the luminance component of the area 1501 (upper 8 × 16 pixels) in FIG. Lower surface average calculator 160
6 calculates an average value 1610 of the luminance components of the area 1502 (8 × 16 pixels at the bottom) in FIG. The left-side average calculator 1607 obtains the average value 1611 of the luminance component of the area 1503 (16 × 8 pixels on the left) in FIG. 15B. The right side average value calculator 1608 calculates the average value 1 of the luminance component of the area 1504 (16 × 8 pixels on the right side) in FIG.
612.

The differentiator 1613 calculates the average value 16 of the luminance component.
09 and 1610 are input, and the difference value 1615 is output to the horizontal type decision unit 1618. The differentiator 1614 receives the average values 1611 and 1612 of the luminance components and outputs the difference value 1616 to the vertical type determiner 1620.

The horizontal type decision unit 1618 calculates the difference value 1615
Is greater than or equal to the horizontal type determination threshold 1617, there is a difference in brightness, so the count value of the counter 1624 is incremented by one.

The vertical type decision unit 1619 calculates the difference value 1616
Is greater than or equal to the vertical type determination threshold value 1619, there is a difference in brightness, so the count value of the counter 1624 is incremented by one.

The decision result integrator 1621 determines the number of areas determined to be horizontal or vertical, that is, the counter 1624, for each predetermined area in which a plurality of blocked areas (for example, N, N is a predetermined integer) are collected. When the count value is a constant X (16
25) Check if there are more than X, and if more than X,
The determination result that the image has many edges is output to the switch controller 1622.

The switching controller 1622 selects a Q parameter without activity modulation when the image is a predetermined image pattern, and modulates the activity when the image is not a predetermined image pattern, according to the output of the judgment result integrator 1621. The switching signal 1623 is set so that the Q parameter is selected and quantized. The predetermined area may be one frame or a plurality of pixels.

The counter value of the counter 1624 is reset to zero for each predetermined area.

(Third Example of Edge Detector) Next, a third example of the edge detector 416 will be described.

In the third example of the edge detector 416, an example will be described in which the presence or absence of an edge is determined from the distribution state of the coefficients after the input video signal is subjected to the discrete cosine transform.

Here, when the sub-blocks of the image are orthogonally transformed, the power is concentrated on the low-frequency components in the flat image area.

On the other hand, as shown in FIG. 17, in the image area where the edges run in (a) horizontal, (b) vertical, and (c) oblique directions, in the conversion area (d),
“Significant conversion coefficients” having relatively large absolute values occur in the areas of 01, 1702, and 1703. A document showing an extremely efficient orthogonal transform coding method by taking advantage of this phenomenon of coefficients after performing discrete cosine transform and introducing a technique to cleverly encode the position and size of significant transform coefficients. As an example of WHChen and W.K.Pratt: "Scene
Adaptive Coder ", IEEE Trans.Commun., COM-32,3, pp.22
5-232 (Mar.1984).

FIG. 18 shows an edge detector 416 of the third example.
FIG.

In the figure, reference numeral 1801 denotes an input video signal;
1802 is a luminance component / color component decomposer, 1803 is a luminance component, 1804 is a predetermined area blocking unit, 1805 is a discrete cosine transformer, 1806 is a threshold cut processing unit, 18
07 is a threshold value X, 1808 is a horizontal edge determiner, 180
9 is a vertical edge determiner, 1810 is a diagonal edge determiner, 1811 is a horizontal edge threshold, 1812 is a vertical edge threshold, 1813 is a diagonal edge threshold, 1814 is a decision result integrator, and 1815 is a switch controller. , 1816
Is a switching signal, and 1817 is a constant F.

An input video signal 1801 is input to a luminance component / color component decomposer 1802 and is decomposed into a luminance component 1803 and a color component.

The luminance component 1803 is divided by a predetermined area blocking unit 1804 into a predetermined area, for example, an area of 16 × 16 pixels, and is output to a discrete cosine transformer 1805. The discrete cosine transformer 1805 is a two-dimensional DCT
(Discrete cosine transform) is performed to obtain a DCT coefficient. This output is input to a threshold cut processor 1806, and the value of the coefficient equal to or smaller than the threshold X (1807) is forcibly set to zero.

The threshold cut processor 1806 sets the DCT coefficient equal to or smaller than the threshold X to “0”, and outputs this DCT coefficient to the horizontal edge determiner 1808, the vertical edge determiner 1809, and the oblique edge determiner 1810. You.

The horizontal type edge determiner 1808 is, for example,
The number of coefficients larger than the value “0” of the area 1701 in FIG. 17 is counted, and if the number exceeds the horizontal edge threshold 1811, it can be determined that the image area has horizontal edge running. The result is set to, for example, “1” and output to the determination result integrator 1814.

Similarly, the vertical edge determiner 1809 is
For example, the number of coefficients larger than the value “0” of the area 1702 in FIG. 17 is counted, and if the number exceeds the vertical edge threshold 1812, it can be determined that the image area has vertical edge running. The determination result is set to, for example, “1” and output to the determination result integrator 1814.

Similarly, the vertical edge determiner 1810
For example, the number of coefficients larger than the value “0” of the area 1703 in FIG. 17 is counted, and if the number exceeds the oblique edge threshold 1813, it can be determined that the image area has an oblique edge run. The determination result is set to, for example, “1” and output to the determination result integrator 1814.

The judgment result integrator 1814 determines, for each predetermined area in which a plurality of blocked areas (for example, N, N is a predetermined integer) are collected, the number of areas determined to be horizontal, vertical, or oblique. It is checked whether or not the number is a constant F (1817) or more. If the number is F or more, it is determined that the image has many edges, and the determination result is output to the switch controller 1815.

The switch controller 1815 selects a Q parameter having no activity modulation if the video pattern is the predetermined video pattern, and modulates the activity if the video image is not the predetermined video pattern, according to the output of the determination result integrator 1814. The switching signal 1816 is set so that the Q parameter is selected and quantized. The predetermined area may be one frame or a plurality of pixels.

In the above case, one area may correspond to any of the horizontal type, the vertical type, and the diagonal type. The mold need not necessarily be limited to the three types of horizontal type, vertical type, and oblique type, and may be reduced or increased.

[0209] The method of edge detection by the edge detector 416 is not limited to the above example, and may be any method.

(Fifth Embodiment) FIG. 5 shows the configuration of a video encoding device according to a fifth embodiment of the present invention.

In the figure, reference numeral 501 denotes an input video signal;
02 is a block divider, 503 is a discrete cosine transformer,
504 is a quantizer, 505 is a variable length coder, 506 is a used Q parameter, 507 is a compressed coded sequence, 508 is a code amount measuring device, 509 is an assigned coding value, 510 is a code amount controller, and 511 is a code amount controller. Q parameter, 512 is an activity calculator, 513 is a Q parameter modulator, 514 is a modulation Q parameter, 515 is a switch, 516 is an edge image detector (edge image detection means), 517 is a switching signal, and 518 is activity. .

In the video coding apparatus shown in FIG. 5, an input video signal 501 is input to an edge image detector 516 and a block divider 502. The edge image detector 516 outputs the activity 5 output from the activity calculator 512.
18 is input, the presence / absence of an edge in a predetermined area is determined from the distribution of the activity 518. When it is determined that the edge has no edge, the input side of the switch 515 is connected to the modulation Q parameter 514 which is the output of the Q parameter modulator 513. Side, and when it is determined to have an edge, the input side of the switch 515 is switched to the Q parameter 511 side which is the output of the code amount controller 510.

The output of the block divider 502 is input to the activity calculator 512, and the activity calculator 512 obtains the activity of each predetermined area based on the pixels in the frame or the pixels between the frames.

On the other hand, the output of the block divider 502 is input to the discrete cosine transformer 503, converted into frequency data, and input to the quantizer 504.

Here, the quantizer 504 quantizes the frequency data in the predetermined area using the determined quantization width, that is, the used Q parameter 506, but the used Q parameter 506 must have no edge as described above. If it is determined, the modulation Q parameter 514 is used, and if it is determined that it has an edge, the Q parameter 511 without modulation is used.

The variable length encoder 505 includes a quantizer 504
Is subjected to variable length encoding to generate a compressed encoded sequence 507.

At this time, the code amount measuring device 508 measures the generated code amount of the compressed coded sequence 507, and inputs the measured value and the assigned coded value 509 to the code amount controller 510 to detect the difference. Then, the Q parameter 511 is calculated based on, for example, Equation 2 in order to control the generated code amount in the next predetermined area.

Note that images having edges include screens including subtitles, color bars, still images, flashing images, test patterns, title screens, images created by computer graphics, images including animation, and the like.
Any of an image having a complicated pattern with little movement or an image including a pattern including a local edge may be used.

A video having a high activity generally has a high coding difficulty, but the generated coding noise is hard to recognize. Conversely, coding noise tends to be noticeable in an image with low activity.

Therefore, by controlling the quantization parameter in accordance with the size of the activity, it is possible to suppress the deterioration of the image quality, and to improve the image quality of the entire video scene.

That is, the code amount controller 510 determines the Q parameter 511 so that the variable length encoder 505 can encode a video signal at an encoding rate according to a scene.

The unit for checking the distribution of the activity may be any of a plurality of pixels, one frame, one field, a plurality of frames.

The unit for detecting the edge image may be any of a plurality of pixels, one frame, one field, or a plurality of frames.

After the detection of the edge image or not, the switching timing for determining the quantization width depending on whether the image is an edge image may be performed for each unit for judging the presence or absence of an edge or for judging the presence or absence of a plurality of edges. Even if a unit has multiple areas
Any of a frame unit, a field unit, and a plurality of frame units may be used.

(First Example of Edge Image Detector) Next, a first example of the edge image detector 516 will be described.

Here, particularly when considering an image having many edges locally on the screen, an extremely high activity portion and an overall low activity portion appear, and the distribution of the activity is as shown in FIG. It is presumed to be distribution.
Therefore, the activity distribution 1901 in FIG.
Is compared with the threshold value U1902, the ones that are equal to or larger than the first threshold value U1902 are counted, the number is set to N1, and the second threshold value L19
If the number is 03 or less and the number is N2, then N1
When ≧ X1 and N2 ≧ X2, it may be determined that the image is an edge image.

FIG. 20 shows the configuration of the first example of the edge image detector.

In FIG. 20, 2001 is an input video signal, 2002 is a luminance component / color component decomposer, 2003 is a luminance component, 2004 is a predetermined area blocking unit, 2005
Is an activity detector, 2006 is a predetermined area extractor,
2007 is a first threshold value determiner, 2008 is a second threshold value determiner, 2009 is a threshold value G1, 2010 is a threshold value G2, 20
11 is the first counter, 2012 is the second counter, 2
013 is a judgment result integrator, 2014 is a constant X1, 201
5 is a constant X2, 2016 is a switch controller, 2017
Is a switching signal.

An input video signal 2001 is input to a luminance component / color component decomposer 2002 and is decomposed into a luminance component 2003 and a color component.

The luminance component 2003 is divided into a predetermined area, for example, an area of 16 × 16 pixels, by a predetermined area blocking unit 2004, and an activity detector 2005 calculates an activity for each predetermined area, and the predetermined area extractor 2
006.

The predetermined region extractor 2006 limits the region for which the activity distribution is to be examined, and outputs a plurality of activities to the first threshold value determiner 2007 and the second threshold value determiner 2008.

The first threshold value determiner 2007 determines that the activity output from the predetermined area extractor 2006 is the threshold value G1.
(2009), the first counter 201
The count value of 1 is incremented by one.

The second threshold value determiner 2008 determines that the activity output from the predetermined area extractor 2006 is equal to the threshold value G2.
If (2010) or less, the second counter 201
The count value of 2 is incremented by one.

Note that the count values of the first counter 2011 and the second counter 2012 are reset to zero for each predetermined area in which the distribution of the activity is checked.

The judgment result integrator 2013 determines that the count value of the first counter 2011 is equal to or larger than the constant X1 (2014) for each predetermined region in which a plurality of blocked regions (for example, N, N is a predetermined integer) are collected. If the count value of the second counter 2012 is equal to or larger than the constant X2 (2015), it is determined that the image has many edges, and the determination result is output to the switch controller 1815.

The switch controller 2016 selects a Q parameter having no activity modulation when it is a predetermined video pattern according to the output of the judgment result integrator 2013, and modulates the activity when it is not the predetermined video pattern. The switching signal 2017 is set so that the Q parameter is selected and quantized. The predetermined area may be one frame or a plurality of pixels.

It is also possible to check whether the distribution of the activity is other than the normal distribution as shown in FIG. 21 in a predetermined area. The reason for this is that an image having a normal distribution of activity can be determined to be an image having flat characteristics, and is therefore expected to have few edges.

In the case of an edge image having a complicated picture, it is determined that the change in the activity is large. Therefore, it may be determined whether the maximum value and the minimum value of the activity have a large difference, and the edge image may be determined. For example, if the difference between the maximum value and the minimum value of the activity in a certain predetermined area is equal to or larger than the threshold value X, the activity can be determined to have a wide distribution, and thus can be determined to be an edge image. Further, for example, the edge image may be determined by checking whether or not the predetermined area has a large difference between the maximum value of the activity equal to or larger than the threshold X1 and the minimum value equal to or smaller than the threshold X2.

(Second Example of Edge Image Detector) Next, a second example of the edge image detector 516 will be described.

When focusing on the activity in a certain area, if a certain activity takes a value extremely different from that of the surrounding activities, when the quantization width is determined in accordance with the activity, the quantization width is set in the area close to the activity. Therefore, an image having such a unique portion may be determined to be an edge image.

FIG. 22 shows an example in which the activity is different from the surrounding activities. The activity in the area 2201 has a difference of “3” or more from the surrounding activities, and indicates a different value from the activity in the surrounding area. If there is such a difference in an adjacent area, the encoding is performed with a higher code amount than the surrounding area at the time of encoding, so that when there is noise, the noise becomes conspicuous. If there are continuous edges, the difference will be more noticeable.

FIG. 23 shows the second example of the edge image detector 516.
FIG. 3 is a configuration diagram of an example of FIG.

In FIG. 23, 2301 is an input video signal, 2302 is a luminance component / color component decomposer, 2303 is a luminance component, 2304 is a predetermined area blocking unit, 2305
Is an activity detector, 2306 is an activity memory, 2307 is an adjacent activity difference determiner, 23
08 is a threshold T, 2309 is a counter, 2310 is a judgment result integrator, 2311 is a switch controller, 2312 is a switch signal, and 2313 is a constant N.

An input video signal 2301 is input to a luminance component / color component decomposer 2302, and is decomposed into a luminance component 2303 and a color component.

The luminance component 2303 is divided by the predetermined area blocking unit 2304 into a predetermined area, for example, an area of 16 × 16 pixels. The activity detector 2305 calculates an activity for each predetermined area and outputs it to the activity memory 2306.

Although the activity memory 2306 stores a plurality of activities, the adjacent activity 2307 stores the maximum M (M
Is a natural number equal to or less than 8) activities, and if any of the difference values exceeds the threshold value T (2308), it is determined that the image area has an edge, and the counter 2309 is reset. Increment by one.

Incidentally, the count value of the counter 2309 is reset to zero for each predetermined area.

The judgment result integrator 2310 determines whether the count value of the counter 2309 is equal to or greater than a constant N (2313) for each predetermined area in which a plurality of blocked areas (for example, N, N is a predetermined integer) are collected. Is determined to be an image having many edges, and the determination result is determined by the switch controller 2
Output to 311.

The switch controller 2311 selects a Q parameter having no activity modulation when it is a predetermined video pattern according to the output of the decision result integrator 2310, and modulates the activity when it is not the predetermined video pattern. The switching signal 2312 is set so that the Q parameter is selected and quantized. The predetermined area may be one frame or a plurality of pixels.

When examining the distribution of activities,
It is not always necessary to cut in the horizontal direction, and it may be in the horizontal, vertical, or oblique direction, or the same operation may be performed by dividing one screen into a predetermined rectangular area.

In the first embodiment, the method of detecting a predetermined video pattern does not need to capture characteristics from the luminance and color components of the video signal. For example, a signal indicating a predetermined video pattern written in the above may be read, and any method of detecting the predetermined video pattern may be used.

In the first to fifth embodiments, the quantization width used by the quantization width determiner at the time of encoding may be specified from outside, and it is determined that the image is a predetermined video pattern or edge image. In the section to be performed, the quantization width held in advance may be specified.

In the first to fifth embodiments, the section in which a predetermined video pattern or an edge image is estimated to be included may be broadened by loosening the criterion, or may be set to a tighter criterion. It may be narrow.

In the first, fourth, and fifth embodiments, the section determined to be a predetermined video pattern or an edge image may be further widened or narrowed.

In the first to fifth embodiments, the unit for performing the discrete cosine transform and the unit for obtaining the activity may be the same or different.

In the first to fifth embodiments, the amount of correlation between pixel levels in a screen (one frame or one field) may be obtained as the activity of a video signal. The correlation amount may be obtained, or both of them may be obtained.

Further, in the first to fifth embodiments, the activity calculators 112, 212, 312, 4
An example of the calculation method of 12, 512 will be described below. Here, for example, the variance at the pixel level of the pixel is obtained as the activity. The variance of the pixel level is calculated as in the following Expression 10, where N pixel levels are X (1), X (2),..., X (N). Note that M in Equation 10 is the following Equation 11
Shows the average value of the N pixel levels calculated by. Also,
P is the number of pixels in a predetermined area for calculating the activity.

[0258] The calculation of the activity is not limited to the variance at the pixel level, but may be the sum of the differences between adjacent pixel levels, or the difference or the sum from the average value. In addition, an average value or a maximum value of the section of the pixel level may be used, or a median (median value), a minimum value, a statistically most frequent one, and the like, in which a change in the pixel level of the section can be measured. , Any may be sufficient.

In addition, the section for obtaining the average value of the activity may be the same size as the predetermined area for obtaining the activity, or an area larger than the predetermined area, for example, 16 ×
The 16 pixel regions may be in units of 30 pixels, in units of one frame, or in units of one field.

In the first to fifth embodiments, the activity may be calculated in units of one frame or one field divided into a plurality of pixels, or one frame or one field, or Any of a plurality of screen units may be used.

[0261]

As described above, according to the video encoding apparatus of the present invention described in claim 1 and the dependent claims, at least when the input video signal is a predetermined video pattern, it is irrelevant to the activity. Since the variable length encoding process is performed using the quantization width determined in the above, at least when the input video signal is a predetermined video pattern such as a subtitle screen,
It is possible to obtain a good image that does not become a jagged or partially blurred image as in the related art.

Further, according to the video encoding apparatus of the invention described in claim 2 and the dependent claims, when the input time code specifies a predetermined video pattern,
Since the variable length encoding process is performed using the quantization width determined independently of the activity, as described above, if at least the input video signal is a predetermined video pattern such as a subtitle screen, the conventional jagged pattern is used. Or a good image which does not become an image in which the edges are frayed partially.

Further, according to the video encoding apparatus of the invention described in claim 3 and the dependent claims, a person or the like watching a video can generate an external interrupt signal in a predetermined video scene such as a subtitle screen. For example, during this occurrence period, the variable-length encoding process is performed using the quantization width determined independently of the activity. Therefore, in a predetermined video scene such as a subtitle screen, the image is jagged or partially cut as in the related art. A good image that does not become an image with a sharp edge can be obtained.

In addition, according to the video encoding apparatus of the invention described in claim 4 and the dependent claims, in a video signal including many edges, the quantization width determined independently of the activity is used. Since the variable-length encoding process is performed, in a video scene such as a subtitle screen including many edges, a good image that does not become a jagged or partially blurred edge as in the related art can be obtained.

In addition, according to the video encoding apparatus of the invention described in claim 5 and the dependent claims, the complexity of the image is determined based on the distribution of the activity, and the edge of a subtitle image or the like having a complicated picture is determined. For images containing, the variable-length encoding process is performed using the quantization width determined independently of the activity, so good subtitles that do not become jagged or partially edge-blown images as in the past An image such as an image is obtained.

In particular, according to the video encoding apparatus of the present invention, the activity is detected by measuring the pixel level correlation amount of a predetermined area in one frame or one field of the input video signal. And the features in the frame can be reflected in the complexity of the image.

In the video encoding apparatus according to the present invention, the activity is detected by measuring the pixel level correlation in a predetermined area between frames or fields of the input video signal. Features can be reflected in the complexity of the image.

[Brief description of the drawings]

FIG. 1 is a block diagram illustrating a configuration of a video encoding device according to a first embodiment of the present invention.

FIG. 2 is a block diagram illustrating a configuration of a video encoding device according to a second embodiment of the present invention.

FIG. 3 is a block diagram illustrating a configuration of a video encoding device according to a third embodiment of the present invention.

FIG. 4 is a block diagram illustrating a configuration of a video encoding device according to a fourth embodiment of the present invention.

FIG. 5 is a block diagram illustrating a configuration of a video encoding device according to a fifth embodiment of the present invention.

FIG. 6 is a block diagram illustrating a configuration of a first example of a predetermined pattern detector according to the first embodiment of the present invention.

FIG. 7 is a diagram showing an activity distribution for explaining an operation of a predetermined video pattern detector.

FIG. 8 is a block diagram showing a configuration of a second example of the predetermined image pattern detector.

FIG. 9 is a distribution diagram of luminance of a predetermined video pattern.

FIG. 10 is a block diagram showing a configuration of a third example of the predetermined image pattern detector.

FIG. 11 is a distribution diagram of luminance of a predetermined video pattern.

FIG. 12 is a block diagram showing a configuration of a fourth example of the predetermined image pattern detector.

FIG. 13 is a diagram illustrating a change in luminance of an image having an edge.

FIG. 14 is a diagram illustrating a configuration of a first example of an edge detector.

FIG. 15 is a diagram showing classification of edge types.

FIG. 16 is a diagram showing a configuration of a second example of the edge detector.

FIG. 17 is a diagram illustrating a relationship between coefficients after discrete cosine transform and types of edges.

FIG. 18 is a diagram illustrating a configuration of a third example of the edge detector.

FIG. 19 is a diagram showing a distribution of activities of an edge image.

FIG. 20 is a diagram illustrating a configuration of a first example of an edge image detector.

FIG. 21 is a diagram illustrating a distribution of activities of an image that is not an edge image.

FIG. 22 is a diagram illustrating an activity having a large difference from an adjacent activity.

FIG. 23 is a diagram illustrating a configuration of a second example of the edge image detector.

[Explanation of symbols]

Reference Signs List 101 input video signal 102 block divider (block dividing means) 103 discrete cosine transformer 104 quantizer (quantizing means) 105 variable length encoder 106 used Q parameter 108 code amount measuring instrument 110 code amount controller 111 Q parameter 112 activity calculator (activity calculating means) 113 Q parameter modulator 114 modulation Q parameter 115 switch (quantization width determining means) 116 predetermined video pattern detector (predetermined video pattern detecting means) 216 time code 219 predetermined video section detector (Predetermined video section detection means) 316 external interrupt signal 317 external interrupt detector (external interrupt detection means) 416 edge detector (edge detection means) 516 edge image detector (edge image detection means) 602 luminance component / color component decomposition 605 Screen activity 606 Predetermined area activity 607 Predetermined area detector 613 Activity reference area averager 615 Activity difference value calculator 620 Color component type checker 625 Judgment result integrator 626 Predetermined pattern judgment result 627 Switching controller 1005 Luminance component Threshold judgment unit 1008, 1212 Distribution judgment unit 1301 Screen 1405 Luminance component difference unit 1408 Luminance component difference judgment unit 1411 Color component memory 1412 Color component difference unit 1415 Color component difference judgment unit 1417 Edge luminance component counter 1418 Luminance component edge judgment unit 1420 Edge Color component counter 1421 Color component edge determiner 1604 Predetermined area blocking device 1605 Upper surface average calculator 1606 Lower surface average calculator 1607 Left surface average calculator 1608 Right side average value calculator 1613 Difference unit 1618 Horizontal type decision unit 1805 Discrete cosine transform unit 1808 Horizontal type edge decision unit 1809 Vertical type edge decision unit 1810 Oblique type edge decision unit 2004 Predetermined area blocking unit 2005 Activity detector 2006 Predetermined area extraction unit 2306 Activity memory 2307 Adjacent activity difference determiner

Claims (22)

[Claims] 1. A video encoding apparatus for compressing and encoding an input video signal, comprising: a predetermined video pattern detecting means for detecting whether the input video signal is a predetermined video pattern; A block dividing unit that divides the block video signal into pixels; an activity calculating unit that inputs a block video signal of the block dividing unit and calculates an activity between these block video signals; An output is input, and when at least the input video signal is the predetermined video pattern, the quantization width is determined independently of the activity, whereas when the input video signal is not the predetermined video pattern, the quantization width is calculated by the activity calculation unit. Quantization width determining means for determining a quantization width according to an activity; Quantizing means for quantizing the block video signal with a quantization width determined by the setting means, wherein the output of the quantizing means is subjected to variable-length encoding to output a compressed encoded sequence. Video encoding device. 2. A video encoding apparatus for compressing and encoding an input video signal, wherein a time code synchronized with the input video signal is input, and wherein the time code indicates a predetermined video section specified from outside. A predetermined video section detecting means for detecting whether or not the block video signal is input; a block video signal for dividing the input video signal into block video signals comprising a plurality of pixels; An activity calculating unit for calculating an activity between the input unit and an output of the predetermined video section detection unit, and when the input video signal is the predetermined video section, a quantization width is determined independently of the activity. If it is not the predetermined video section, the quantization width is set according to the activity calculated by the activity calculating means. And a quantizing unit for quantizing the block video signal with the quantization width determined by the quantizing width determining unit, wherein the signal generated by the quantizing unit is variable-length. A video encoding device that encodes and outputs a compressed encoded sequence. 3. A video encoding device for compressing and encoding an input video signal, comprising: an external interrupt signal input; an external interrupt detection means for detecting whether or not the external interrupt signal is input; A block dividing unit for dividing into block video signals composed of a plurality of pixels; an activity calculating unit for inputting a block video signal of the block dividing unit and calculating an activity between these block video signals; and the external interrupt detecting unit And the quantization width is determined independently of the activity in at least the section where the input video signal has the input of the external interrupt signal, and the activity calculating means in the section where the external interrupt signal is not input. Quantization width determination that determines the quantization width according to the activity calculated by And a quantization means for quantizing the block video signal with the quantization width determined by the quantization width determination means, wherein the signal obtained by the quantization means is subjected to variable-length coding, A video encoding device for outputting a coded sequence. 4. A video encoding apparatus for compressing and encoding an input video signal, comprising: a block dividing unit for dividing the input video signal into a block video signal composed of a plurality of pixels; and a block video signal of the block dividing unit. And an activity calculating means for calculating the activity between these block video signals, an edge detecting means for detecting the presence or absence of an edge in a predetermined area of the input video signal, and an output of the edge detecting means, A quantization width determination unit that determines a quantization width independently of the activity, or determines a quantization width according to the activity calculated by the activity calculation unit; and a quantum determined by the quantization width determination unit. Quantizing means for quantizing the block video signal with a quantization width, wherein the output of the quantizing means is variable-length coded. And outputting a compressed coded sequence. 5. A video encoding device for compressing and encoding an input video signal, comprising: a block dividing unit for dividing the input video signal into a block video signal comprising a plurality of pixels; and a block video signal of the block dividing unit. And an activity calculating means for calculating an activity between these block video signals; and, based on a distribution of activities calculated by the activity calculating means, whether or not the input video signal is an image including an edge. Edge image detection means for determining, and input the output of the edge image detection means, and determine the quantization width independently of the activity in at least the section where the input video signal is determined to be an edge image, In the section determined not to be an edge image, the activity calculated by the activity calculating means is determined. A quantization width determining unit that determines a quantization width according to the activity; and a quantization unit that quantizes the block video signal with the quantization width determined by the quantization width determining unit. A video coding apparatus, wherein the output of the means is variable-length coded and a compressed coded sequence is output. 6. The quantization width determining means, when determining a quantization width irrespective of an activity, determines the quantization width to a quantization width specified from the outside. The video encoding apparatus according to claim 1, 2, 3, 4, or 5. 7. The apparatus according to claim 1, wherein said activity calculating means measures a pixel-level correlation amount of a predetermined area in one frame or one field of said input video signal. 3. The video encoding device according to claim 4 or claim 5. 8. The apparatus according to claim 1, wherein said activity calculating means measures a pixel level correlation amount in a predetermined area between frames or fields of said input video signal. The video encoding device according to claim 4. 9. The method according to claim 1, wherein the predetermined image pattern detecting means determines that a type of a color component in a predetermined area of the input video signal is a predetermined integer value K.
2. The method as claimed in claim 1, wherein the number is smaller than the number.
A video encoding device according to claim 1. 10. The predetermined video pattern detection means detects that the number of pixels whose luminance component in a predetermined area of the input video signal has a luminance component larger than a predetermined threshold X is equal to or larger than a predetermined integer value M. The video encoding device according to claim 1, wherein: 11. The predetermined image pattern detecting means determines that the number of pixels whose luminance component exceeds a predetermined number G1 in a predetermined area is equal to or greater than a predetermined integer value N1 and the number of pixels smaller than the predetermined number G2 is a predetermined integer value. 2. The video encoding apparatus according to claim 1, wherein the video encoding apparatus detects that the number is equal to or greater than N2. 12. The video encoding apparatus according to claim 1, wherein the predetermined video pattern detection unit detects that the input video signal is a video including subtitles. 13. The video encoding apparatus according to claim 1, wherein the predetermined video pattern detection unit detects that the input video signal is a title screen or a menu screen. 14. The video encoding apparatus according to claim 1, wherein the predetermined video pattern detection unit detects that the input video signal is a video including a computer graphics screen. 15. The video encoding apparatus according to claim 1, wherein the predetermined video pattern detection unit detects that the input video signal is a video including an animation. 16. The video encoding apparatus according to claim 4, wherein said edge detecting means detects the presence or absence of an edge based on a distribution of coefficients after discrete cosine transform. 17. The quantization width determining means, if the number of areas determined to have an edge by the edge detecting means is equal to or greater than a predetermined threshold F within a predetermined area, regardless of the activity. 5. The video encoding device according to claim 4, wherein the quantization width is determined according to the activity calculated by the activity calculation means when the value is smaller than the predetermined threshold F. 18. The image processing apparatus according to claim 5, wherein the edge image detecting means detects an edge image when a distribution of the activity calculated by the activity calculating means is other than a normal distribution in a predetermined area. A video encoding device according to claim 1. 19. The edge image detecting means determines a difference between a maximum value and a minimum value of the activity calculated by the activity calculating means in a predetermined area by a predetermined threshold X.
6. The video encoding apparatus according to claim 5, wherein if the above is the case, the image is detected as an edge image. 20. The edge image detecting means, wherein a maximum value of the activity calculated by the activity calculating means is equal to or more than a predetermined threshold X1 in a predetermined area,
6. The video encoding apparatus according to claim 5, wherein when the minimum value is equal to or less than the predetermined threshold value X2, the image is detected as an edge image. 21. The edge image detecting means, wherein, in a predetermined area, the number of activities having a predetermined threshold value G1 or more is a predetermined integer value X1 or more, and the number of activities less than the predetermined threshold value G2 is a predetermined integer value X2 or more. 6. The video encoding apparatus according to claim 5, wherein in such a case, the image is detected as an edge image. 22. The edge image detecting means wherein a difference between one activity and M activities (M is an integer from 1 to 8) adjacent to the activity exceeds a predetermined threshold value T in a predetermined area. 6. The video encoding apparatus according to claim 5, wherein when the number of images exceeds a predetermined number N, the image is detected as an edge image.