JPH11313329A - Image data processing unit, its method and service medium - Google Patents

Abstract

PROBLEM TO BE SOLVED: To prevent the increase in the amount of codes on the occurrence of a random noise. SOLUTION: A motion vector detection circuit 10 calculates an absolute value difference sum (MEdistortion) and a means luminance separation residue (IntraMAD) of a luminance value and provides an output to a dummy data generating circuit 25. The dummy data generating circuit 25 uses the absolute value difference sum (MEdistortion) and the means luminance separation residue (InteraMAD) so as to discriminate whether or not image data outputted by the motion vector detection circuit 10 to a post-stage are a random noise. Furthermore, the dummy data generating circuit 25 controls each circuit so that the random noise is not coded with fidelity but minimizes the amount of codes to be produced.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to an image data processing apparatus and method, and a providing medium.
The present invention relates to an image data processing apparatus and method for reducing an amount of code generated for random noise, and a providing medium.

[0002]

2. Description of the Related Art MPEG is a technology for compressing moving image data.
Two schemes are known. In the MPEG2 system, data compression is performed using the temporal correlation between image frames and the spatial correlation within a frame, and the greater the temporal or spatial correlation, the more efficient the image data becomes. Data can be compressed well. However, when random noise called so-called "sand storm", which has no temporal correlation or spatial correlation, is compressed using MPEG, it is a noise image, but it has no meaning as image information. , The code amount cannot be compressed (decreased).

[0003] Therefore, a moving image with random noise mixed in the middle of a normal (no noise) image is used.
In the case of MPEG compression, the code amount of a section where a random noise image is compressed is significantly increased as compared with the code amount of a normal image.

[0004] By the way, the MPEG encoder that performs MPEG compression functions so that the amount of code generated at the time of data compression falls within a predetermined value (attack bit rate). It may be. Therefore, in order to prevent an increase in the code amount at the time of such random noise, the conventional image compression device shown in FIG.

[0005] The noise filter 32 includes one of a plurality of video materials (digital VTR input, analog input, telephone line input, satellite line input) that may have random noise mixed in the transmission path. The input is switched by the input switch 31. At the time of this switching, random noise may be generated and mixed into the video material.

[0006] The noise filter 32 is a low-pass filter or the like, and limits the band of image data to be passed to remove high-frequency components that are random noise.

[0007]

However, since the noise filter 32 limits the band of normal (non-noise) image data, the quality of the image data passing through the noise filter 32 is degraded. There was a problem that caused them to do so.

The present invention has been made in view of such a situation, and it is an object of the present invention to suppress an increase in a code amount when random noise occurs without deteriorating the quality of normal image data.

[0009]

According to the first aspect of the present invention, there is provided an image data processing apparatus, comprising: detecting means for detecting a motion vector of image data; and compensating for the motion of the image data based on the motion vector detected by the detecting means. Compensating means, switching means for switching a motion prediction mode of image data to one of a frame prediction mode and a field prediction mode, calculation means for calculating image data corresponding to the motion prediction mode switched by the switching means, Determining means for determining whether the data is random noise; a value of a motion vector to be supplied to the compensating means in accordance with the determination result of the determining means; and a control means for controlling a prediction mode to be switched by the switching means. It is characterized by having.

[0010] According to a fourth aspect of the present invention, there is provided an image data processing method comprising:
A detecting step of detecting a motion vector of the image data;
The compensation step of compensating the motion of the image data based on the motion vector detected in the detection step, the switching step of switching the motion prediction mode of the image data to one of the frame prediction mode and the field prediction mode, and the switching step. A calculating step of calculating image data corresponding to the motion prediction mode; a determining step of determining whether the image data is random noise; and a calculating step of the motion vector to be supplied to the compensating step in accordance with the determination result of the determining step. And a control step of controlling a prediction mode to be switched to the switching step.

According to a fifth aspect of the present invention, there is provided the providing medium, comprising: a detecting step of detecting a motion vector of the image data; a compensation step of compensating the motion of the image data based on the motion vector detected in the detecting step; A switching step of switching the prediction mode between the frame prediction mode and the field prediction mode, a calculation step of calculating image data corresponding to the motion prediction mode switched by the switching step, and whether or not the image data is random noise The image data processing apparatus performs a process including a determining step of determining whether a motion vector is to be supplied to a compensation step in accordance with a determination result of the determining step, and a control step of controlling a prediction mode to be switched to a switching step. A program to be executed is provided.

According to the first aspect of the present invention, the detecting means detects a motion vector of the image data, and the compensating means compensates for the motion of the image data based on the motion vector detected by the detecting means. The switching means switches the motion prediction mode of the image data to one of the frame prediction mode and the field prediction mode, and the calculating means calculates the image data in accordance with the motion prediction mode switched by the switching means, Determines whether or not the image data is random noise, and the control means controls the value of the motion vector to be supplied to the compensation means in accordance with the determination result of the determination means, and the prediction mode to be switched by the switching means. .

[0013] In the image data processing method according to the fourth aspect and the providing medium according to the fifth aspect, the motion vector of the image data is detected in the detecting step, and the motion vector detected in the detecting step is detected in the compensating step. The motion prediction mode of the image data is switched between the frame prediction mode and the field prediction mode in the switching step in the switching step, and the motion prediction mode switched in the switching step is supported in the calculation step. In the determination step, it is determined whether the image data is random noise, and in the control step, the value of the motion vector to be supplied to the compensation step in accordance with the determination result of the determination step, The prediction mode to be switched to the switching step is controlled.

[0014]

DESCRIPTION OF THE PREFERRED EMBODIMENTS Embodiments of the present invention will be described below. In order to clarify the correspondence between each means of the invention described in the claims and the following embodiments, each means is described. When the features of the present invention are described by adding the corresponding embodiment (however, an example) in parentheses after the parentheses, the result is as follows.

That is, the image data processing apparatus according to the first aspect of the present invention includes a detecting means (for example, the motion vector detecting circuit 10 in FIG. 1) for detecting a motion vector of the image data and a motion vector detected by the detecting means. (For example, the motion compensation circuit 24 in FIG. 1) for compensating for the motion of the image data by using the switching means (for example, in FIG. 1) for switching the motion prediction mode of the image data between the frame prediction mode and the field prediction mode. Prediction mode switching circuit 1
2), a calculating means (for example, the calculating unit 13 in FIG. 1) for calculating the image data corresponding to the motion prediction mode switched by the switching means, and a determination for determining whether the image data is random noise or not. Means, and a control means (for example, a dummy data generation circuit 25 in FIG. 1) for controlling a value of a motion vector to be supplied to the compensation means in accordance with a result of the determination by the determination means and a prediction mode to be switched by the switching means. It is characterized by the following.

However, of course, this description does not mean that each means is limited to those described.

The configuration of an MPEG encoder to which the present invention is applied will be described with reference to FIG. The image data to be encoded is a motion vector detecting circuit 1 for each macroblock.
Input to 0. The motion vector detection circuit 10 processes the image data of each frame as an I picture, a P picture, or a B picture according to a predetermined sequence set in advance. It is determined in advance whether an image of each sequentially input frame is processed as a picture of I, P, or B.

The image data of the frame processed as an I picture is transferred from the motion vector detecting circuit 10 to the front original image section 11a of the frame memory 11 and stored. The image data of the frame processed as a B picture is stored in the reference original image. The image data of the frame transferred and stored in the image unit 11b and processed as a P picture is stored in the rear original image unit 11c.
Is transferred and stored.

At the next timing, B
When an image of a frame to be processed as a picture or a P picture is input, the image data of the first P picture that has been stored in the rear original image section 11c is transferred to the front original image section 11a, and the next B The image data of the picture is stored (overwritten) in the original image section 11b, and the image data of the next P picture is stored (overwritten) in the rear original image section 11c. Such operations are sequentially repeated.

The signal of each picture stored in the frame memory 11 is read therefrom, and the prediction mode switching circuit 12 performs frame prediction mode processing or field prediction mode processing.

Further, under the control of the prediction determination circuit 14, the calculation unit 13 performs calculation of intra-picture prediction, forward prediction, backward prediction, or bidirectional prediction. Which of these processes is to be performed is determined according to the prediction error signal (the difference between the reference image to be processed and the predicted image corresponding to the reference image). For this reason, the motion vector detection circuit 10 generates a sum of absolute values (or a sum of squares) of the prediction error signal used for this determination.

Here, the frame prediction mode and the field prediction mode in the prediction mode switching circuit 12 will be described.

When the frame prediction mode is set, the prediction mode switching circuit 12 outputs the luminance signal supplied from the motion vector detection circuit 10 as it is to the subsequent operation unit 13. That is, in this case, the luminance signal of the odd field line and the luminance signal of the even field line are mixed. In this frame prediction mode, one motion vector corresponds to one frame in macroblock units.

On the other hand, the prediction mode switching circuit 1
2, in the field prediction mode, a signal input from the motion vector detection circuit 10 is separated into an odd field line luminance signal and an even field line luminance signal and output to the arithmetic unit 13. In this case, for each odd field in macroblock units, 1
Pieces of motion vectors correspond to each other, and another one of the motion fields corresponds to even-numbered fields.

The motion vector detecting circuit 10 outputs to the prediction mode switching circuit 12 the absolute value sum of the prediction error in the frame prediction mode and the absolute value sum of the prediction error in the field prediction mode. The prediction mode switching circuit 12 compares the absolute value sum of the prediction errors in the frame prediction mode and the field prediction mode, performs processing corresponding to the prediction mode having a small value, and outputs the data to the arithmetic unit 13.
Output to

However, such processing is actually performed by the motion vector detection circuit 10. That is, the motion vector detection circuit 10 outputs a signal having a configuration corresponding to the determined mode to the prediction mode switching circuit 12, and the prediction mode switching circuit 12 outputs the signal as it is to the subsequent operation unit 13.

In addition, the motion vector detecting circuit 10 determines whether the intra prediction, the forward prediction, the backward prediction, or the bidirectional prediction is to be performed in the prediction determining circuit 14 as described below. Generate the sum of absolute values of prediction errors.

That is, as the sum of the absolute values of the prediction errors of the intra prediction, the absolute value | ΣAij | of the sum ΣAij of the macroblock signals Aij of the reference picture and the macroblock signal A
The difference between the sum 絶 対 | Aij | of the absolute values | Aij | The sum of absolute values | Aij−Bij | of the difference Aij−Bij | between the signal Aij of the macroblock of the reference image and the signal Bij of the macroblock of the predicted image is used as the absolute value sum of the prediction error of the forward prediction. Bij |. In addition, the absolute value sum of the prediction error between the backward prediction and the bidirectional prediction is obtained in the same manner as in the forward prediction (by changing the predicted image to a different predicted image from that in the forward prediction).

The sum of these absolute values is supplied to the prediction judgment circuit 14. The prediction determination circuit 14 selects the smallest absolute value sum of the prediction errors of the forward prediction, the backward prediction, and the bidirectional prediction as the sum of the absolute values of the prediction errors of the inter prediction (inter determination). Furthermore, the absolute value sum of the prediction error of the inter prediction and the absolute value sum of the prediction error of the intra prediction are compared, and the smaller one is selected, and the mode corresponding to the selected absolute value sum is set as the prediction mode. select. That is, if the sum of the absolute values of the prediction errors of the intra prediction is smaller, the intra prediction mode is set. If the sum of the absolute values of the prediction errors in the inter prediction is smaller, the mode in which the corresponding sum of the absolute values is the smallest among the forward prediction, backward prediction, and bidirectional prediction modes is set.

As described above, the motion vector detecting circuit 10
Supplies the macroblock signal of the reference image to the arithmetic unit 13 via the prediction mode switching circuit 12 in a configuration corresponding to the mode selected by the prediction mode switching circuit 12 among the frame or field prediction modes. Prediction prediction circuit 14 of the four prediction modes
The motion vector (Xvec, Yvec) between the predicted image and the reference image corresponding to the selected prediction mode is detected, and is output to the variable length coding circuit 18 and the motion compensation circuit 14. As described above, as the motion vector, the motion vector having the smallest absolute value sum of the corresponding prediction errors is selected.

Further, as shown in FIG. 2, the motion vector detecting circuit 10 calculates the luminance value C of the macroblock of the predicted image.
The sum of the absolute values | Cij−Rij | of the differences between ij and the luminance values Rij of the macroblocks of the reference image Σ | Cij−Rij | (the absolute value difference sum (MEdistortion) of the luminance values) and the luminance of the macroblocks of the prediction image Value Cij and the average C of the luminance values of the macroblocks
Sum of absolute value of difference of ave | Cij−Cave | Σ | Cij−Cav
e | (average luminance separation residual (IntraMAD)), and outputs it to the dummy data generation circuit 25 together with the motion vector (Xvec, Yvec) described above.

In general, the absolute value difference sum MEdistorti of luminance values
“on” indicates the amount of deformation of the macroblock of the reference image. The smaller the value is, the smaller the amount of deformation is. On the contrary, the value of the object image that moves while deforming like a soap bubble is large. The average luminance separation residual IntraMAD is
It is known that there is a correlation with the degree of difficulty of encoding a macroblock, and that the larger this value is, the more difficult it is to encode. This is because there are higher-order frequencies in the macroblock, which is a property typified by an image such as a fine tie pattern, for example. It is also rare that the absolute value difference sum MEdistortion of the luminance values of image data other than random noise and the average luminance separation residual IntraMAD show large values at the same time.

When the motion vector detecting circuit 10 reads out the I-picture image data from the front original image section 11a, the prediction determining circuit 14 sets the intra-frame or field (image) intra-prediction mode (performs motion compensation) as the prediction mode. Mode), and set the switch 13d of the operation unit 13
To the contact a side. As a result, the image data of the I picture is input to the DCT mode switching circuit 15.

The DCT mode switching circuit 15 sets D to a state in which odd field lines and even field lines are mixed (frame DCT mode) or separated (field DCT mode).
Output to the CT circuit 16.

That is, the DCT mode switching circuit 15
Compares the coding efficiency when DCT processing is performed by mixing data of odd fields and even fields, and the coding efficiency when DCT processing is performed in a separated state,
Select a mode with good coding efficiency.

For example, the input signal has a configuration in which odd field lines and even field lines are mixed, and the difference between the signal of the vertically adjacent odd field line and the signal of the even field line is calculated. Find the sum (or sum of squares) of the values.

Further, the input signal is configured such that the lines of the odd field and the even field are separated from each other, and the difference between the signal of the odd field adjacent vertically and the difference of the signal of the lines of the even field are calculated. Then, the sum (or sum of squares) of the respective absolute values is obtained.

Further, the DCT mode corresponding to the smaller value is set by comparing the two (the sum of absolute values). That is, if the former is smaller, the frame DCT mode is set, and if the latter is smaller, the field DCT mode is set.

Then, the data having the configuration corresponding to the selected DCT mode is output to the DCT circuit 16, and the DCT flag indicating the selected DCT mode is transmitted to the variable length coding circuit 1.
8 and the motion compensation circuit 24.

When the prediction mode switching circuit 12 selects a frame prediction mode (a mode in which odd lines and even lines are mixed), the DCT mode switching circuit 1
5, it is highly possible that the frame DCT mode (a mode in which odd lines and even lines are mixed) is selected,
When the prediction mode switching circuit 12 selects a field prediction mode (a mode in which data of an odd field and an even field are separated), the DCT mode switching circuit 15 converts a field DCT mode (data of an odd field and an even field) into a field DCT mode. (Separated mode) is likely to be selected.

However, the mode is not always selected in this way. In the prediction mode switching circuit 12, the mode is determined so that the absolute value sum of the prediction error becomes small, and in the DCT mode switching circuit 15, , The mode is determined so that the coding efficiency is good.

The image data of the I picture output from the DCT mode switching circuit 15 is input to the DCT circuit 16 where it is subjected to DCT processing and converted into DCT coefficients. The DCT coefficients are input to the quantization circuit 17, quantized at a quantization scale corresponding to the data storage amount (buffer storage amount) of the transmission buffer 19, and then input to the variable length coding circuit 18.

The variable length coding circuit 18 is
The image data (in this case, I-picture data) supplied from the quantization circuit 17 is converted into a variable-length code such as a Huffman code in accordance with the supplied quantization scale (scale) and transmitted. Output to the buffer 19.

The variable length coding circuit 18 also sets a quantization scale (scale) by the quantization circuit 17 and a prediction mode (intra-picture prediction, forward prediction, backward prediction, or bidirectional prediction) by the prediction determination circuit 14. Mode indicating whether the motion vector has been detected), a motion vector from the motion vector detection circuit 10,
A prediction flag (a flag indicating whether the frame prediction mode or the field prediction mode has been set) from the prediction mode switching circuit 12 and a DCT flag (either the frame DCT mode or the field DCT mode) output from the DCT mode switching circuit 15 are set. Are input, and these are also variable-length coded.

The transmission buffer 19 temporarily stores the input data, and converts the data corresponding to the storage amount into the quantization circuit 17.
Output to When the remaining data amount increases to the allowable upper limit, the transmission buffer 19 increases the quantization scale of the quantization circuit 17 by the quantization control signal,
The data amount of the quantized data is reduced. Conversely, when the remaining data amount decreases to the allowable lower limit value, the transmission buffer 19 sends the quantization signal to the quantization circuit 17.
By reducing the quantization scale of, the data amount of the quantized data is increased. In this way, overflow or underflow of the transmission buffer 19 is prevented.

The data stored in the transmission buffer 19 is read out at a predetermined timing and output to a transmission path.

On the other hand, the I picture data output from the quantization circuit 17 is input to the inverse quantization circuit 20 and inversely quantized according to the quantization scale supplied from the quantization circuit 17. The output of the inverse quantization circuit 20 is input to an IDCT (Inverse Discrete Cosine Transform) circuit 21, subjected to inverse discrete cosine transform processing, supplied to a forward prediction image unit 23 a of a frame memory 23 via an arithmetic unit 22, and stored. Is done.

The motion vector detecting circuit 10 receives the image data of each frame sequentially input, for example, as I, B, P, B, P, B... After the image data of the input frame is processed as an I picture, the image data of the next input frame is processed as a P picture before the image of the next input frame is processed as a B picture. This is because a B picture involves backward prediction and cannot be decoded unless a P picture as a backward predicted image is prepared first.

Therefore, after the processing of the I picture, the motion vector detecting circuit 10 starts the processing of the picture data of the P picture stored in the rear original picture section 11c. Then, as in the case described above, the sum of the absolute values of the inter-frame differences (prediction errors) in macroblock units is supplied from the motion vector detection circuit 10 to the prediction mode switching circuit 12 and the prediction determination circuit 14. Prediction mode switching circuit 12
And the prediction determination circuit 14 sets a frame / field prediction mode or a prediction mode of intra prediction, forward prediction, backward prediction, or bidirectional prediction in accordance with the absolute value sum of the prediction errors of the macroblocks of the P picture. I do.

When the intra-picture prediction mode is set, the operation unit 13 switches the switch 13d to the contact a as described above. Therefore, this data is transmitted to the transmission path via the DCT mode switching circuit 15, the DCT circuit 16, the quantization circuit 17, the variable length coding circuit 18, and the transmission buffer 19, like the I picture data. Also,
This data is supplied to and stored in the backward prediction image unit 23b of the frame memory 23 via the inverse quantization circuit 20, the IDCT circuit 21, and the arithmetic unit 22.

When the forward prediction mode is set,
The switch 13d is switched to the contact b, and the image (in this case, the I-picture image) data stored in the forward prediction image section 23a of the frame memory 23 is read out. Motion compensation is performed corresponding to the motion vector output from the circuit 10. That is, the motion compensating circuit 24
4, when the setting of the forward prediction mode is instructed, the read address of the forward prediction image unit 23a corresponds to the motion vector from the position corresponding to the position of the macroblock currently output by the motion vector detection circuit 10. The data is read out by shifting by an amount and predicted image data is generated.

The predicted image data output from the motion compensation circuit 24 is supplied to the calculator 13a. Arithmetic unit 13a
Subtracts the prediction image data corresponding to the macroblock supplied from the motion compensation circuit 24 from the macroblock data of the reference image supplied from the prediction mode switching circuit 12, and outputs the difference (prediction error). I do. This difference data is transmitted to the transmission path via the DCT mode switching circuit 15, the DCT circuit 16, the quantization circuit 17, the variable length coding circuit 18, and the transmission buffer 19. Also,
The difference data is locally decoded by the inverse quantization circuit 20 and the IDCT circuit 21 and input to the arithmetic unit 22.

The arithmetic unit 22 is also supplied with the same data as the predicted image data supplied to the arithmetic unit 13a. The calculator 22 adds the prediction image data output from the motion compensation circuit 24 to the difference data output from the IDCT circuit 21. As a result, image data of the original (decoded) P picture is obtained. The image data of the P picture is supplied to and stored in the backward prediction image section 23b of the frame memory 23.

After the data of the I picture and the data of the P picture are stored in the forward prediction image section 23a and the backward prediction image section 23b, the motion vector detection circuit 10 next executes the processing of the B picture. The prediction mode switching circuit 12 and the prediction determination circuit 14 correspond to the magnitude of the absolute value sum of the inter-frame difference in macroblock units,
The frame / field mode is set, and the prediction mode is set to any of the intra prediction mode, the forward prediction mode, the backward prediction mode, and the bidirectional prediction mode.

As described above, in the intra prediction mode or the forward prediction mode, the switch 13d is set to the contact a or b.
Can be switched to At this time, the same processing as in the case of the P picture is performed, and the data is transmitted.

On the other hand, when the backward prediction mode or the bidirectional prediction mode is set, the switch 13d is switched to the contact point c or d.

In the backward prediction mode in which the switch 13d is switched to the contact point c, the picture (in this case, the picture of the P picture) stored in the backward prediction picture section 23b is read out and the motion compensation circuit 24 is read. As a result, motion compensation is performed corresponding to the motion vector output from the motion vector detection circuit 10. That is, when the setting of the backward prediction mode is instructed by the prediction determination circuit 14,
The motion vector detection circuit 10 reads the data by shifting the read address of the backward prediction image unit 23b from the position corresponding to the position of the currently output macroblock by the amount corresponding to the motion vector, and generates the prediction image data. I do.

The predicted image data output from the motion compensation circuit 24 is supplied to the calculator 13b. Arithmetic unit 13b
Subtracts the predicted image data supplied from the motion compensation circuit 24 from the macroblock data of the reference image supplied from the prediction mode switching circuit 12, and outputs the difference. This difference data is stored in the DCT mode switching circuit 1
5, transmitted to the transmission path via the DCT circuit 16, the quantization circuit 17, the variable length coding circuit 18, and the transmission buffer 19.

In the bidirectional prediction mode in which the switch 13d is switched to the contact point d, the image data (in this case, the I-picture image) stored in the forward prediction image section 23a and the data stored in the backward prediction image section 23b. Data (in this case, a P-picture image) is read out, and the motion compensating circuit 24 outputs the motion vector detecting circuit 1
Motion compensation is performed corresponding to the motion vector output by 0.

That is, when the prediction determination circuit 14 instructs the setting of the bidirectional prediction mode, the motion compensation circuit 24 determines the read addresses of the forward predicted image section 23a and the backward predicted image section 23b, and the motion vector detection circuit 10 The data is read out from the position corresponding to the position of the macroblock being output by shifting the data by an amount corresponding to the motion vector (in this case, two motion vectors for the forward predicted image and the backward predicted image). Generate data.

The predicted image data output from the motion compensation circuit 24 is supplied to the calculator 13c. Arithmetic unit 13c
Subtracts the average value of the prediction image data supplied from the motion compensation circuit 24 from the macroblock data of the reference image supplied from the motion vector detection circuit 10 and outputs the difference. This difference data is transmitted to the transmission path via the DCT mode switching circuit 15, the DCT circuit 16, the quantization circuit 17, the variable length coding circuit 18, and the transmission buffer 19.

The picture of the B picture is not stored in the frame memory 23 because it is not regarded as a predicted picture of another picture.

In the frame memory 23, the forward predicted image section 23a and the backward predicted image section 23b are switched between banks as necessary, and stored in one or the other for a predetermined reference image. Can be switched and output as a forward predicted image or a backward predicted image.

In the above description, the processing for the luminance signal has been mainly described, but the chrominance signal is similarly processed and transmitted. As a motion vector for processing a chrominance block, a motion vector obtained by halving the motion vector of the corresponding luminance signal in the vertical and horizontal directions is used.

The dummy data generation circuit 25 calculates the absolute value difference sum MEdistorti input from the motion vector detection circuit 10.
Using on and the average luminance separation residual IntraMAD, it is determined whether or not the image data output to the subsequent stage by the motion vector detection circuit 10 is random noise based on a predetermined determination criterion.

For example, if random noise image data as shown in FIG. 3A is faithfully encoded, the image has no meaning as information. There is no practical problem even if it is different from FIG. Therefore, the dummy data generation circuit 25 controls each circuit so as not to encode the random noise faithfully but to reduce the amount of codes generated as much as possible.

That is, the dummy data generation circuit 25
If the predicted image is determined to be random noise, a dummy motion vector (0,0) having a motion vector component of 0
Is supplied to the motion compensation circuit 24, and the prediction mode switching circuit 12 and the DCT mode switching circuit 15 are prohibited from switching to the field prediction, and are switched to the frame prediction that generates a smaller amount of code than the field prediction. Let it. Further, the prediction determination circuit 14 prohibits intra-picture prediction (intra prediction), and among inter predictions in which block distortion is not conspicuous subjectively, a smaller amount of code is generated as compared to backward prediction and bidirectional prediction. Set the forward prediction mode.

It should be noted that the amount of generated code can be minimized by using random noise as a skipped macroblock. However, in this case, the noise stops on the playback screen, which is unnatural, and is not regarded as a skipped macroblock. I decided that.

On the other hand, if it is determined that the noise is not random noise, the dummy data generation circuit 25 does not execute the above processing.

FIG. 4 is a plot of the result of calculating the absolute value difference sum MEdistortion of the macroblock which is random noise and the average luminance separation residual IntraMAD. Note that the horizontal axis is the absolute value difference sum MEdistortion, and the vertical axis is the average luminance separation residual IntraMAD. It is clear from the figure that the result of calculating the macroblock of the random noise in the region where the absolute value difference sum MEdistortion is 12000 or more and the average luminance separation residual IntraMAD is larger than the absolute value difference sum MEdistortion. Is concentrated.

Therefore, based on the experimental results shown in FIG. 4, as a criterion for noise determination in the dummy data generation circuit 25, as shown in FIG.
A region where on is a threshold of 12000 or more and the average luminance separation residual IntraMAD is larger than the absolute difference sum MEdistortion is defined as a random noise region, and the absolute difference sum MEdistortion and the average luminance separation residual IntraMAD are calculated. If the result exists in this area, it is determined that the noise is random noise.

Next, the noise determination processing of the dummy data generation circuit 25 will be described with reference to the flowchart of FIG. In step S1, the dummy data generation circuit 25 determines whether the value of the absolute value difference sum MEdistortion input from the motion vector detection circuit 10 is equal to or greater than a threshold (12000). If it is determined that the value of the absolute value difference sum MEdistortion is equal to or larger than the threshold, the process proceeds to step S2.

In step S2, the dummy data generation circuit 25 determines whether or not the average luminance value separation residual IntraMAD is equal to or greater than the absolute value difference sum MEdistortion. When it is determined that the average luminance value separation residual IntraMAD is equal to or larger than the absolute value difference sum MEdistortion (belonging to the random noise area), the process proceeds to step S3.

In step S3, the dummy data generation circuit 25 supplies the motion vector (0,0) to the motion compensation circuit 24. The dummy data generation circuit 25 includes the prediction mode switching circuit 12 and the DCT mode switching circuit 1
For 5, switching to field prediction is prohibited, and switching to frame prediction is performed. Further, the prediction determination circuit 1
For 4, the intra prediction (intra prediction) is prohibited, and the forward prediction mode of the inter prediction is set.

If it is determined in step S1 that the value of the absolute value difference sum MEdistortion is equal to or smaller than the threshold value, or in step S2, the average luminance separation residual
If it is determined that IntraMAD is equal to or smaller than the absolute value difference sum MEdistortion, the dummy data generation circuit 25 ends the processing. Therefore, normal MPEG encoding is performed on a normal image that is not random noise.

As described above, according to the present embodiment, even if an image containing random noise between normal images is encoded, the code amount of the section of the random noise is a normal image. Since there is no increase in the code amount of the section, for example, even if random noise occurs in the channel chB of the four multiplexed channels chA to chD, as shown in FIG. The code amount increases in the section, and the other multiplexed channels chA, c
The hC band is not violated, and the code amount does not increase even in the random noise section as shown in FIG.

FIG. 8 shows a code amount A generated when the present invention is applied to an image in which random noise is generated periodically without applying the present invention, and a code amount B generated when the present invention is applied to the same image. From the increase rate of code amount B to code amount A (B
-A) / A calculated and plotted. In the figure, the horizontal axis indicates the reproduction time, and the vertical axis indicates the rate of increase. According to the figure, the rate of increase of the code amount of each frame (black square) after the occurrence of noise (black circle) is-on average.
This is about 5%, which means that the code amount B is 5% greater than the code amount A.
% Less. Note that even if the decrease is about several percent, since the code amount A when random noise is encoded without applying the present invention is large, the absolute amount of the code amount B is greatly reduced. For example, 5% of the code amount A is about 3 megabytes, which is equivalent to the code amount for one channel of normal satellite broadcasting.

The present invention can be applied to multi-channel satellite broadcasting because real-time processing is possible.

The computer program for performing each of the above processes can be provided to the user via a medium such as a magnetic disk or a CD-ROM, or a network medium such as the Internet or a digital satellite. it can.

[0080]

As described above, the image data processing apparatus according to claim 1, the image data processing method according to claim 4,
According to the providing medium of the present invention, whether or not the image data is random noise is detected, and the motion vector and the prediction mode are controlled in accordance with the detection result. It is possible to suppress an increase in the code amount when random noise occurs without deteriorating data quality.

[Brief description of the drawings]

FIG. 1 is a block diagram illustrating a configuration of an MPEG encoder to which the present invention has been applied.

FIG. 2 is a diagram illustrating correspondence between a predicted image and a macroblock of a reference image.

FIG. 3 is a diagram illustrating encoding of random noise.

FIG. 4 is a diagram showing characteristics of random noise.

FIG. 5 is a diagram showing a random noise region.

FIG. 6 is a flowchart illustrating a noise determination process of the dummy data generation circuit 25 of FIG. 1;

FIG. 7 is a diagram illustrating a reduction in the code amount generated for random noise.

FIG. 8 is a diagram illustrating a rate of a decrease in a code amount generated with respect to random noise.

FIG. 9 is a diagram illustrating an example of a configuration of a conventional image compression device.

[Explanation of symbols]

11 frame memory, 12 prediction mode switching circuit, 13 operation unit, 14 prediction judgment circuit, 15 DC
T mode switching circuit, 16 DCT circuit, 17 quantization circuit, 18 variable length coding circuit, 19 transmission buffer, 20 inverse quantization circuit, 21 IDCT circuit, 22
Arithmetic unit, 23 frame memory, 24 motion compensation circuit, 25 dummy data generation circuit, 31 input switch, 32 noise filter, 33 MPEG encoder

Claims (5)

[Claims] 1. An image data processing apparatus for compressing image data, comprising: detecting means for detecting a motion vector of the image data; and compensation for compensating for the motion of the image data based on the motion vector detected by the detecting means. Means for switching the motion prediction mode of the image data to one of a frame prediction mode and a field prediction mode; and calculation means for calculating the image data corresponding to the motion prediction mode switched by the switching means. Determining means for determining whether the image data is random noise; a value of the motion vector to be supplied to the compensating means in response to a result of the determination by the determining means; and a prediction mode for switching by the switching means An image data processing apparatus comprising: control means for controlling the image data. 2. The image processing apparatus according to claim 1, wherein the determining unit determines whether the image data is random noise based on a sum of absolute differences between luminance values of the image data and an average luminance separation residual. 2. The image data processing device according to 1. 3. The image data processing apparatus according to claim 1, wherein the control unit further controls prediction determination of the image data in accordance with a determination result of the determination unit. 4. An image data processing method for compressing image data, comprising: a detecting step of detecting a motion vector of the image data; and a compensation for compensating a motion of the image data based on the motion vector detected in the detecting step. A switching step of switching a motion prediction mode of the image data to one of a frame prediction mode and a field prediction mode; and a calculation step of calculating the image data corresponding to the motion prediction mode switched by the switching step. A determining step of determining whether the image data is random noise; a value of the motion vector to be supplied to the compensation step in accordance with a determination result of the determining step; and a prediction mode for switching to the switching step And a control step of controlling Image data processing method. 5. An image data processing apparatus for compressing image data, comprising: a detecting step of detecting a motion vector of the image data; and a compensation for compensating for the motion of the image data based on the motion vector detected in the detecting step. A switching step of switching a motion prediction mode of the image data to one of a frame prediction mode and a field prediction mode; and a calculation step of calculating the image data corresponding to the motion prediction mode switched by the switching step. A determining step of determining whether the image data is random noise; a value of the motion vector to be supplied to the compensation step in accordance with a determination result of the determining step; and a prediction mode for switching to the switching step For executing a process including a control step of controlling A providing medium characterized by providing gram.