JP2001359101A - Flashback processing method and digital video signal processing unit and digital broadcast receiver employing this method - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide a flashback processing method by which a memory quantity and an arithmetic quantity in the flashback processing of video data in compliance with the MPEG system can considerably be reduced. SOLUTION: In the case of selecting the flashback processing (S11, S12), canonical processing is applied to each motion vector for each of all macro blocks (S13), after obtaining normalizing vectors and a mean value of their scalar amounts (S14), a covariance matrix A is generated (S15). Furthermore, an eigen value and an eigen vector are generated from the covariance matrix A (S16), after deciding a mapping system to obtain a moving amount through the comparison of two eigen threshold values (S17), video data are mapped on a coordinate moved by the calculated moving amount from the coordinate position on just preceding video data having already been drawn, decoded video images are sequentially overwritten (S18, S19) and panorama images are drawn.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a flashback processing method for creating a panoramic video from a digital video signal, and a digital video signal processing apparatus and a digital broadcast receiving apparatus using the flashback processing method.

[0002]

2. Description of the Related Art As is well known, for analog broadcasting, processing such as motion detection, three-dimensional YC separation, noise reduction (NR) processing, and ip conversion are known as video processing techniques for creating high-quality video. This is done inside the receiver set. However, these processes are based on Pixel
The processing is performed in units, and the processing requires a huge memory capacity and processing amount. For this reason, in terms of price, such as using a line memory and a frame buffer and further performing each processing using a hard-wired logic, the processing is not always efficient.

On the other hand, as a new video processing technique, there has been proposed a method of realizing a panoramic drawing using a moving image on an analog TV using a flashback function. However, this technique itself measures, for each video frame (or field), how much and in what direction each image has moved between the previous frame (or field) and the current frame (or field). There is only a method that uses motion detection used in conventional analog broadcast receivers.

In this method, each Pixel in a certain window between frames or fields is used.
It is necessary to calculate a difference with respect to the luminance component of (pixel) and add the difference by the number of pixels in the window. When considering the accuracy of this calculation, it is important to increase the size of the target window, but by expanding the window, the amount of calculation becomes twice the window area. For this reason, there is a problem that a large window cannot always be used.

[0005] In recent years, the operation of digital broadcasting services has begun. However, the processing of high image quality itself used in digital broadcasting uses a conventional processing method, and new high image quality is used. The technology of transformation has not been born. Therefore, when starting operation of digital broadcasting, it has been considered to incorporate a flashback processing function into a receiver set. However, as described above, the conventional flashback processing method using motion detection has the above-mentioned problems, and a significant reduction in the memory capacity and processing amount in flashback processing has been required for the operation of digital broadcasting services. Expected.

[0006]

As described above, according to the conventional flashback processing method using motion detection, the P / P in the window between frames or between fields can be reduced.
An operation of adding the difference value for each pixel by the number of pixels in the window is required, and this processing requires high-speed arithmetic processing. Therefore, the above-described arithmetic processing circuit is required in dedicated hardware or an LSI. However, there is a problem in terms of cost. On the other hand, it has been proposed to provide a digital broadcast receiving apparatus with a flashback processing function. However, if the conventional processing method is applied as it is, the above-described problem will occur as it is.

SUMMARY OF THE INVENTION The present invention has been made in view of the above problems, and provides a flashback processing method capable of greatly reducing the amount of memory and the amount of calculation, and a digital video using the flashback processing method. It is an object to provide a signal processing device and a digital broadcast receiving device.

[0008]

In order to achieve the above object, a flashback processing method according to the present invention detects a motion vector for each frame or field from a video signal in macroblock units, and based on the detected motion vector. Compression encoding the video signal, inputting a digital video signal generated by adding motion vector data to the compression encoded video data, extracting the motion vector data from the digital video signal, When performing panoramic drawing by performing flashback processing on a reproduced video signal by decoding video data that has been compression-encoded based on the motion vector data and reproducing the video signal, the Calculating the moving direction and moving amount for each frame or field of the reproduced video signal Performs motion detection, and to perform a panoramic drawing by joining overlapped while shifting the pixel positions of the reproduced video signal for each frame or field based on the motion detection result.

The above method utilizes the fact that the compression-encoded video data is generated based on the motion vector, and calculates the moving direction and the moving amount for each frame or field of the reproduced video signal. Since the detection is performed, it is not necessary to perform a complicated calculation such as the conventional motion detection for calculating the movement amount.

[0010]

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

(First Embodiment) FIG. 1 is a block diagram showing a configuration of a digital broadcast receiving apparatus equipped with a flashback processing function as a first embodiment according to the present invention. In the digital broadcasting in this embodiment, it is assumed that the MPEG-2 system is adopted for encoding video data and audio data.

In FIG. 1, a digital broadcast wave transmitted from a broadcasting station is input to a tuner 2 through an antenna 1. The tuner 2 selects a channel specified by the user, and the selected signal is input to the transport decoder 3. This transport decoder unit 3
After demodulating the channel selection signal, the channel selection signal is separated into three data of video, audio, and other data. Each of the separated data is stored in a memory 4 connected to the decoder unit 3.
Is stored in

Of the data stored in the memory 4, the data of the motion vector added to the video data
Memory 5 through bus line BUS under the control of PU 6
Is transferred and stored. The video data and the audio data among the data stored in the memory 4 are input to the video decoder unit 7 and the audio decoder unit 8, respectively.

The video decoder unit 7 performs an MPEG-2 video decoding process to extract actual video data. This video data is transferred to the back-end processor unit 9 and connected to the back-end processor unit 9. Is stored in the frame memory 10. The back-end processor unit 9 follows an instruction from the host CPU 6
The video data stored in the frame memory 10 is read out and processed as appropriate to generate an actual video signal. The video signal generated here is input to the display 11 and displayed on the screen as a video. The audio decoder section 8 converts the audio signal into an actual analog audio signal by an MPEG-2 audio decoding process.
The sound is reproduced from No. 3.

The host CPU 6 has a bus line BUS
In response to a user instruction input from a remote controller (hereinafter referred to as a remote controller) 15 through an infrared receiving unit 14 connected to the CPU, required processing blocks (for example, a video decoder unit 7, an audio decoder unit 8, a back-end processor unit 9)
) To manage and control the processing operation, and here, it is assumed that a control function for flashback processing is incorporated.

Prior to describing the operation of the digital broadcast receiving apparatus having the above configuration, a conventional flashback processing method will be briefly described with reference to FIG.

In FIG. 2, first, the target Pixel
, A certain window (window) is formed, and the size of this window moves the target window up, down, left, and right from the same display position with respect to the immediately preceding frame or field. Then, as shown in Expression (1), each Pixel in the target window is set between frames or between fields.
, The difference is added by Pixel in the window, the position where the added value becomes smallest is determined, and it is determined that the position has been moved from the previous frame or field to this position. Do.

[0018]

(Equation 1)

This method utilizes a conventional motion detection method, calculates a difference between luminance components of each Pixel in a window between frames or between fields, and further calculates the difference in the window. Pix
An operation of adding el is required. When considering the accuracy of this calculation, it is important to increase the size of the target window. However, by expanding the window, the amount of calculation becomes twice as large as the window area. Can not do.

On the other hand, video data of digital broadcasting has MP
EG-2 is used, and there is a motion vector generated during the MPEG-2 encoding process. The present invention focuses on this point, and since the motion vector can be regarded as the same as the above-described motion detection result, by using it for flashback processing, the amount of operation of the flashback processing device or digital broadcast receiving device can be reduced. It is also possible to reduce the amount of memory associated therewith.

In the configuration shown in FIG. 1, the host CP
By U6, arithmetic processing is performed according to the flowcharts shown in FIGS.

In FIG. 3, first, when the user (viewer) selects the flashback processing with the remote controller 15,
The host CPU 6 determines this (S11, S12),
Each motion vector P for each macro block stored in the memory 5 is read out, and a canonicalization process is performed in accordance with Equation (2) so that the average value in the vector becomes 0, and the result is stored in the memory 5 again. It is stored (S13).

[0023]

(Equation 2)

The host CPU 6 reads again the canonicalized input vector P 'from the memory 5 for the canonicalized input vector P', and normalizes the input vector P '. An average value of the input vector P ″ and its scalar amount S is obtained by using Expressions (3) and (3 ′), and the result is stored in the memory 5 (S14).

[0025]

(Equation 3)

When the normalization processing is completed for each of the motion vectors of all the macroblocks within the range of the motion extraction, the host CPU 6 calculates the covariance matrix A using the equation (4).
Is created, and the result is stored in the memory 5 (S1).
5).

[0027]

(Equation 4)

When the creation of the covariance matrix A is completed using the motion vectors of all the macroblocks which are the target range of the motion extraction, the host CPU 6 uses the following equation (5).
An eigenvalue and an eigenvector are created from the covariance matrix A, and the result is stored in the memory 5 (S16).

[0029]

(Equation 5)

Here, the host CPU 6 reads the eigenvalue obtained by equation (5), and then
A comparison process is performed for the two threshold values α1 and α2, and a mapping method for obtaining the movement amount is determined based on the comparison result (S17). FIG. 4 shows an example of the processing flow.

In FIG. 4, the host CPU 6 first
With respect to the eigenvalue λ1 corresponding to the first axis serving as the main axis obtained by the principal component analysis, a magnitude relationship with α1 inherent in the system is examined (S21 to S23).

If λ1 is larger than the threshold α1 of the system and larger than the threshold β of another system, the variance of the motion vector is concentrated in the vector direction indicated by the eigenvector Φ1. Then, it is determined that the movement can be represented by the moving amount V in the direction indicated by the eigenvector (S24). On the other hand, when λ1 is larger than α1, but the difference is smaller than β value, the eigenvector Φ1 of only the first axis as the main axis
It is not possible to represent the moving direction only by using only the moving amount V using the eigenvector Φ2 of the second axis.
Is obtained (S25). The movement vector V when λ1 is larger than α1 is given by the following equation (6).

[0033]

(Equation 6)

On the other hand, if λ1 is smaller than α1, the host CPU 6 performs comparison processing with the threshold α2 inherent in the system (S26). In this comparison method, first, λ1 is divided by λ2 as in Expression (7), and μ is obtained as a result.

[0035]

(Equation 7)

Further, in the host CPU 6, μ and α2
Is examined (S27), and if μ is larger than α2, a basic movement vector V ′ for each macroblock is obtained by equation (8) (S28). Subsequently, for the frame data of the immediately preceding video stored in the frame memory 10 connected to the back-end processor unit 9,
The movement is performed by the movement amount calculated by the equation (8), and the motion of each pixel in a certain area with the video data in which the current decoding result is stored is calculated by using the motion detection of the equation (9). A detailed movement amount (movement vector W) is determined (S29).

[0037]

(Equation 8)

[0038]

(Equation 9)

On the other hand, when λ1 is smaller than α1 and μ is smaller than α2, the host CPU 6 determines that each macroblock has a large variance and is not gathered in a certain direction. It is determined that the corresponding motion vector itself is handled as the movement amount (S30).

After the calculation of each moving amount is completed using the above α1 and α2, the host CPU 6, as shown in FIG. 5, replaces the decoded video data with the video data immediately before being drawn. Is mapped on the coordinates moved by the calculated movement amount from the coordinate position of, and the decoded video is overwritten on the area on the new mapped coordinates (S18 in FIG. 3). Thereafter, the processes of steps S13 to S18 are repeated for a designated time (S19). Thus, panoramic drawing can be performed.

As described above, in the present embodiment, the eigenvalue expansion is performed based on the motion vector for each macroblock, utilizing the fact that the video data is in the MPEG-2 format.
Since the flashback process is performed by calculating the moving distance and the moving vector from the eigenvalue and the eigenvector, it is not necessary to perform a complicated operation such as the conventional motion detection for calculating the moving amount. Further, by calculating the movement amount based on the eigenvectors of a plurality of axes, it is possible to perform more accurate mapping processing in the flashback processing as compared with a method using an average value or the like.

(Second Embodiment) Next, as a second embodiment according to the present invention, it is possible to adopt a low-priced digital broadcast receiving apparatus that does not have a host CPU with a very high performance. A possible flashback processing method will be described. Note that the digital broadcast receiving apparatus to which the flashback processing method of the present embodiment is applied basically has the first
Since the configuration may be the same as that shown in FIG. 1 in the embodiment, the description is omitted here by referring to FIG.

In the present embodiment, the flashback process is performed by using information of only a part of the screen for the motion vector, and using a simple average value for calculating the movement amount.
The reason why flashback processing is performed using a part of the screen is that, when panning (rotating) with camera work, a part corresponding to a part of the screen can be used without using the motion vector of the entire screen. This is because even when compared with the case where the movement amount is calculated by extracting the movement amount, the movement amount is substantially equivalent. The method is described below.

FIG. 6 is a flowchart showing a procedure for realizing flashback according to this embodiment. First, when the user (viewer) selects the flashback processing with the remote control 15, the host CPU 6 determines this (S3).
1, S31), out of the motion vectors of all the macroblocks stored in the memory 5, the macroblocks other than the macroblocks in the vicinity of the center on the display screen and not in contact with the upper, lower, left and right ends. Is read out (S33).

The reason for this is that, in a normal screen, the camera pans (rotates) in accordance with the movement of the center part, so that the movement amount of the center part is smaller than that of the periphery, and the motion vectors for all macroblocks are used. Rather than doing, except for the macro block around the center, and
This is because the use of the motion vectors of the macroblocks not in contact with the upper, lower, left, and right ends improves the accuracy in calculating the movement amount.

Further, when a motion vector of a macroblock which is in contact with the display screen vertically and horizontally is used, the relation of the motion with the immediately preceding frame (field) may not always be correctly expressed. This is because there is a possibility that the accuracy of the movement amount can be improved by using a macroblock other than the vicinity of the center and not in contact with the top, bottom, left and right.

In particular, when a simple average value of each motion vector as shown in the present embodiment is used, it is equivalent to using only the first axis when eigenvalues are expanded, so that the movement amount of the motion vector itself is used. This is because it is necessary to increase the accuracy of

Hereinafter, a method of calculating the movement amount will be described. First, the host CPU 26 calculates the average value of the motion vector according to the equation (10), and stores the result in the memory 25 again (S34).

[0049]

(Equation 10)

After the calculation of the movement amount by the equation (10) is completed, the calculation is performed on the decoded video data from the coordinate position on the video data immediately before being drawn, as in the case of FIG. It maps on the coordinates moved by the moved amount, and overwrites the decoded image on the area on the new mapped coordinates (S35). Thereafter, the above steps S33 to S35 are repeated for a designated time (S36). Thus, panoramic drawing can be performed.

As described above, in the present embodiment, the movement of the screen means that the camera is panning (rotating) with the camera work in order to follow the target person or object, or the camera is on the rail. We are paying attention to that we are moving. That is, since the camera pans (rotates) or moves in accordance with the movement of the central portion, the property that the movement amount of the entire screen can be represented by the motion vector corresponding to the macroblock of the peripheral portion is used. are doing.

Furthermore, rather than using the motion vectors for all macroblocks, the use of only the motion vectors of the macroblocks corresponding to a part of the screen, especially the part other than the vicinity of the center of the display screen, makes it possible to reduce the amount of motion. The difference between the motion vector around the center of the blank screen and the motion vector other than around the center of the display screen corresponding to the background is removed.

Thus, the accuracy of a simple moving amount calculation method can be improved. In particular, low-cost TVs that do not have a very high-performance host CPU
In e.g., the eigenvalue expansion process is executed by the host C
Although the load of the PU is applied, according to the method of the present embodiment, only a part of the screen is used for the motion vector, and the flashback process is performed using a simple average value for calculating the movement amount. It will be practical enough.

(Third Embodiment) Next, as a third embodiment of the present invention, a macroblock other than the center peripheral portion used in the second embodiment is used. A flashback processing method using the used eigenvalue expansion will be described. Note that the digital broadcast receiving apparatus to which the flashback processing method of the present embodiment is applied may have the same configuration as that shown in FIG. 1 in the first embodiment. Description is omitted.

FIG. 7 is a flowchart showing a procedure for realizing flashback according to this embodiment. First, when the user (viewer) selects the flashback processing with the remote controller 15, the host CPU 6 determines this (S4).
1, S42), out of the motion vectors of all the macroblocks stored in the memory 5, the macroblocks other than the macroblocks in the vicinity of the center on the display screen and not in contact with the upper, lower, left and right ends. Is read out (S43). The reason is as described in the second embodiment.

Next, the host CPU 6 calculates the equation (1) for each of the read motion vectors P for each macro block.
A canonicalization process is performed according to 1) so that the average value in the vector becomes 0, and the result is stored in the memory 45 again (S44).

[0057]

[Equation 11]

The host CPU 6 reads out the canonicalized input vector P 'again from the memory 5 for the canonicalized input vector P', and normalizes the input vector P '. The average value of the input vector P ″ and its scalar amount S is obtained by using Expressions (12) and (12 ′), and the result is stored in the memory 5 (S45).

[0059]

(Equation 12)

When the normalization processing is completed for each of the motion vectors of all the macroblocks within the range of the motion extraction, the host CPU 6 creates a covariance matrix A using Expression (13), and as a result, Is stored in the memory 5 (S4
6).

[0061]

(Equation 13)

When the creation of the covariance matrix A is completed using the motion vectors of all the macroblocks which are the target areas for motion extraction, the host CPU 6 calculates the covariance matrix A from the covariance matrix A using the equation (14). Eigenvalues and eigenvectors are created, and the results are stored in the memory 5 (S47).

[0063]

[Equation 14]

Here, after reading the eigenvalue obtained by the equation (14), the host CPU 6 compares the two threshold values α1 and α2 inherent in the system, and moves based on the comparison result. A mapping method for determining the quantity is determined (S48). How to determine this mapping method
The processing flow shown in FIG. 4 can be applied as it is.

After the calculation of each moving amount is completed using the above α1 and α2, the host CPU 6 performs the same operation as in FIG.
The decoded video data is mapped onto the coordinates that have been moved by the calculated movement amount from the coordinate position on the video data just before being drawn, and the area corresponding to the new mapped coordinates is mapped. Is overwritten with the decoded video (S49). Thereafter, the processing of steps S43 to S49 is repeated for a designated time (S5
0). Thus, panoramic drawing can be performed.

The flashback processing method of the present embodiment is a method that utilizes the advantages of both the first embodiment and the second embodiment. Compared to the first embodiment, the accuracy of the motion vector itself is smaller. It is excellent in the following points.

First, the fact that the screen moves means that the camera is panning (rotating) by camera work or the camera is moving on rails in order to follow the target person or object. This means that this embodiment focuses on this point.

That is, in the method according to the present embodiment, the camera pans (rotates) or moves in accordance with the movement of the center portion, and therefore, the motion vector corresponding to the macroblock in the peripheral portion represents the movement amount of the entire screen. Take advantage of the property that it is possible. Furthermore, rather than using the motion vectors for all the macroblocks, only the motion vectors of the macroblocks corresponding to a part of the screen, particularly the part other than the vicinity of the center of the display screen, are used. As a result, it is possible to remove the difference between the motion vector around the center of the screen that does not move much and the motion vector other than around the center of the display screen corresponding to the background. Also,
It is possible to use a motion vector based on the basic camera work itself as a main component.

On the other hand, in the principal component analysis, different from the second embodiment, eigenvalue expansion is performed, a moving distance and a moving vector are calculated from the plurality of eigenvalues and the plurality of eigenvectors, and flashback processing is performed. This eliminates the need for complicated calculations such as motion detection in calculating the amount of movement. In addition, since the movement amount is determined based on the eigenvectors of a plurality of axes, more accurate mapping can be performed.

(Fourth Embodiment) Next, a flashback processing method according to a fourth embodiment of the present invention will be described. Note that the digital broadcast receiving apparatus to which the flashback processing method of the present embodiment is applied may have the same configuration as that shown in FIG. 1 in the first embodiment. Description is omitted.

The flashback processing method of the present embodiment differs from the first to third embodiments in that the moving direction of the entire screen is roughly calculated as the first step in the processing up to the actual video data mapping. The second step is to perform a hierarchical process of calculating a movement amount of a macroblock near a corner located on the opposite side to the movement direction and using this value as mapping data.

FIG. 8 is a flowchart showing a procedure for realizing flashback according to the present embodiment. First, when the user (viewer) selects the flashback processing with the remote controller 15, the host CPU 6 determines this (S5).
1, S52), among all the macroblocks stored in the memory 5, using the motion vectors of the macroblocks other than the macroblock near the center of the display screen, calculate the average value of the motion vectors according to the equation (15). The calculation is performed, and the result is stored in the memory 5 again (S53).

[0073]

(Equation 15)

Here, the host CPU 6 reads the result of the movement vector V obtained by the equation (15) from the memory 5, and from which vector the movement direction has been moved in any of the nine areas shown in FIG. Is checked, and the result is stored in the memory 5 (S54).

Next, as shown in FIG. 10, a region (edge peripheral portion) corresponding to a corner opposite to the moving direction is searched (S55), and the region is set as a moving amount calculation region.
A motion vector corresponding to this area is extracted (S5).
6) The movement amount is calculated for the extracted motion vector by the following method.

First, a plurality of motion vectors (input vectors) in an area corresponding to the movement amount calculation area are expressed by the following equation (1).
The vector is normalized according to 6), and the result is stored in the memory 5 (S57).

[0077]

(Equation 16)

The host CPU 6 reads again the canonicalized input vector P ′ from the memory 5 for the canonicalized input vector P ′, and normalizes the input vector P ′. The average value of the input vector P ″ and its scalar amount S is obtained using Expressions (17) and (17 ′), and the result is stored in the memory 5 (S58).

[0079]

[Equation 17]

When the normalization processing is completed for each of the motion vectors of all the macroblocks within the range of the motion extraction, the host CPU 6 creates a covariance matrix A using Expression (18), and as a result Is stored in the memory 5 (S5
9).

[0081]

(Equation 18)

When the creation of the covariance matrix is completed using the motion vectors of all the macroblocks in the range of motion extraction, the host CPU 6 calculates the eigenvalues and the eigenvectors from the covariance matrix A using equation (19). Create
The result is stored in the memory 5 (S60).

[0083]

[Equation 19]

Here, after reading the eigenvalue obtained by equation (19), the host CPU 6 compares the two threshold values α1 and α2 inherent in the system, and determines the movement amount from the comparison result. The required mapping method is determined (S61). The method of determining this mapping method is shown in FIG.
Can be applied as it is.

After the calculation of each moving amount is completed by using the above α1 and α2, the host CPU 6 performs the same operation as in FIG.
The decoded video data is mapped onto the coordinates that have been moved by the calculated movement amount from the coordinate position on the video data just before being drawn, and the area corresponding to the new mapped coordinates is mapped. Is overwritten with the decoded video (S62). Thereafter, the processes of steps S53 to S62 are repeated for a designated time (S6
3). Thus, panoramic drawing can be performed.

As described above, the flashback processing method according to the present embodiment employs a method of calculating the movement amount using a two-stage via-larky structure. This method realizes the same principle as that of first performing corner alignment when a plurality of partially overlapped still images are bonded. In other words, first, the direction of movement is calculated from the average value of the motion vector, and the amount of movement is calculated at the center of the corner located in the opposite direction to the direction by more detailed eigenvalue expansion. Is to be mapped. Such a hierarchical structure enables more detailed mapping.

The present invention is not limited to the above embodiment. For example, in the case where a plurality of motion vectors are used as one input vector and eigenvalue expansion is performed by increasing the number of dimensions, or a time series of motion vectors over several frames is treated as one input vector, and It is also applicable to eigenvalue expansion.

The present invention is not limited to the digital video signal of the MPEG-2 system, but can be similarly applied to video data using a motion vector corresponding thereto. Furthermore, it is needless to say that the present invention can be applied not only to the digital broadcast receiving apparatus but also to a digital video signal processing apparatus such as a DVD reproducing apparatus.

The flashback processing using the moving amount can be applied to a camera shake correction of a digital camera or a digital TV camera. As this application method,
There is a method of absorbing the movement of the screen due to camera shake by moving the video data of the next frame in the direction 180 degrees (reverse direction) to the motion vector using the motion vector. As described above, by performing motion detection using a motion vector, camera shake correction can be realized without performing detailed motion detection used in conventional motion adaptation or the like, and as a result, the amount of memory and the amount of calculation can be significantly reduced. It can be reduced.

[0090]

As described above, according to the present invention, in flashback processing used in panorama drawing in digital broadcasting and the like, video data is mapped using a motion vector created during video encoding processing. Accordingly, a panoramic drawing can be realized without performing the motion detection used in the conventional motion adaptation or the like in detail.

Therefore, it is possible to provide a flash-back processing method capable of greatly reducing the amount of memory and the amount of calculation, and to provide a digital video signal processing device and a digital broadcast receiving device using the flash-back processing method. .

[Brief description of the drawings]

FIG. 1 is a block diagram showing a configuration of a digital broadcast receiving apparatus to which a flashback processing method according to a first embodiment of the present invention is applied.

FIG. 2 is a schematic diagram for explaining a conventional flashback processing method.

FIG. 3 is a flowchart illustrating a flashback processing method according to the first embodiment.

FIG. 4 is a flowchart illustrating a method of determining a movement amount according to the first embodiment;

FIG. 5 is a view showing an example of a flashback result according to the first embodiment;

FIG. 6 is a flowchart illustrating a flashback processing method according to a second embodiment of the present invention.

FIG. 7 is a flowchart illustrating a flashback method according to a third embodiment of the present invention.

FIG. 8 is a flowchart illustrating a flashback method according to a fourth embodiment of the present invention.

FIG. 9 is a diagram showing nine quadrants in a moving direction according to a fourth embodiment.

FIG. 10 is a diagram showing a relationship between a moving direction of the entire screen and an area in which detailed verification of the moving direction is performed in the fourth embodiment.

[Explanation of symbols]

DESCRIPTION OF SYMBOLS 1 ... Antenna, 2 ... Tuner part, 3 ... Transport decoder part, 4 ... Memory, 5 ... Memory, 6 ... Host CP
U, 7: video decoder unit, 8: audio decoder unit, 9: back-end processor unit, 10: frame memory, 11
... display, 12 ... audio amplifier, 13 ... speaker, 14 ... infrared receiver, 15 ... remote control.

Continued on front page F-term (reference)

Claims (10)

[Claims] An input signal detects a motion vector for each frame or field from a video signal in units of macroblocks, and compression-codes the video signal based on the motion vector. A digital video signal generated by adding motion vector data, extracting the motion vector data from the input digital video signal, and decoding the compression-encoded video data based on the motion vector data. A flashback processing method of performing a flashback process on a reproduced video signal to perform panoramic drawing, comprising: a moving direction for each frame or field of the reproduced video signal from data of a motion vector extracted at the time of decoding; Motion detection means for performing motion detection for calculating the amount of movement; A flashback processing method, comprising: a panorama drawing means for performing a panorama drawing by superimposing and connecting the pixel positions of the reproduced video signal for each frame or field based on the detection result of the means. 2. The flashback according to claim 1, wherein said motion detecting means includes a moving amount calculating step of calculating a moving amount using a motion vector itself for each macroblock of said reproduced video signal. Processing method. 3. The motion detecting means generates a covariance matrix from motion vectors of a plurality of macroblocks in the reproduced video signal, performs eigenvalue expansion from the covariance matrix, and obtains each component obtained by the eigenvalue expansion. 2. The flashback processing method according to claim 1, further comprising a moving amount calculating step of calculating a moving amount from an eigenvalue and an eigenvalue vector corresponding to each of the axes. 4. A moving amount calculating step according to claim 1, wherein said moving detecting means includes a moving amount calculating step of calculating an average value of the motion vectors of a plurality of macroblocks in said reproduced video signal to obtain a moving amount. Flashback processing method. 5. The motion detection means calculates an average value from motion vectors of a plurality of macroblocks in the reproduced video signal, detects a rough motion, and then, based on the rough motion, detects the motion of the plurality of macroblocks. Create a covariance matrix from the motion vector, perform eigenvalue expansion from the covariance matrix, and calculate the amount of movement from the eigenvalue and eigenvalue vector corresponding to each axis of each component obtained by the eigenvalue expansion. 2. The flashback processing method according to claim 1, further comprising an amount calculation procedure. 6. The motion detecting means generates a covariance matrix from motion vectors of a plurality of macroblocks in the reproduced video signal, performs eigenvalue expansion from the covariance matrix, and obtains each component obtained by the eigenvalue expansion. After performing the first motion detection using the eigenvalues and eigenvalue vectors corresponding to the respective axes, the image data of the immediately preceding frame or field is further compared with the coordinates of each target pixel in the macroblock. The panorama drawing means includes a movement amount calculation procedure of a hierarchical structure for calculating a movement amount of each pixel in the macro block by performing a second movement detection between the video data of the current frame or the field. From the detection result of the first motion detection and the amount of movement of each pixel by the second motion detection,
2. The flashback processing method according to claim 1, further comprising a processing procedure for adjusting a display address for each pixel. 7. The motion detecting means, as the second motion detection, a luminance difference between frames or fields,
7. The flashback processing method according to claim 6, further comprising a moving amount calculating step of calculating a moving value for each pixel using a difference between both luminance and color difference. 8. The method according to claim 2, wherein the motion detecting means includes at least two or more processing procedures, and includes at least one or more eigenvalues and at least one or more predetermined threshold values. 2. The flashback processing method according to claim 1, further comprising a selection step of automatically selecting a processing procedure based on a magnitude relation between the above and a magnitude relation between an average value and a preset threshold value. 9. An input signal comprising: detecting a motion vector for each frame or field from a video signal in units of macroblocks; compressing and encoding the video signal based on the motion vector; A digital video signal generated by adding motion vector data, extracting the motion vector data from the digital video signal, and decoding the compression-encoded video data based on the motion vector data. Decoding means for reproducing the original video signal; and motion detection for calculating a moving direction and a moving amount for each frame or field of the reproduced video signal output from the decoder means from the data of the motion vector extracted by the decoding means. Then, based on the motion detection result, the pixel position of the reproduced video signal is Digital video signal processing apparatus characterized by comprising a flashback processing means for performing a panoramic drawing by joining overlapped while shifting every field. 10. A motion vector for each frame or field is detected from a video signal in macroblock units,
The video signal is compression-encoded based on the motion vector, and is used in a digital broadcasting system that broadcasts a digital video signal generated by adding motion vector data to the compression-encoded video data. A tuner section for receiving and selecting and outputting a digital video signal to be broadcast, a demodulation section for synchronously detecting a tuning output of the tuner section and demodulating the digital video signal; and a digital video demodulated by the demodulation section. A decoding unit that extracts the motion vector data from the signal, decodes the compression-encoded video data based on the motion vector data, and reproduces the original video signal; and a motion vector extracted by the decoding unit. Moving direction and moving for each frame or field of the reproduced video signal output from the decoder unit from the data Flashback processing means for performing motion detection for calculating the amount, and performing panorama drawing by superimposing and connecting pixel positions of the reproduced video signal while shifting the pixel position for each frame or field based on the motion detection result. A digital broadcast receiving apparatus characterized by the above-mentioned.