JP2000041244A - Image compression coder - Google Patents

Abstract

PROBLEM TO BE SOLVED: To uniformize image quality and to prevent disturbance in an image by correcting pull-down information. SOLUTION: An inverse telecine circuit 1 applies inverse telecine processing to a video signal based on an output of a 3:2 pull-down detection circuit 2 at first. A video compression circuit 4 compresses the video signal after inverse telecine processing. The 3:2 pull-down information generating circuit 5 detects pull-down information by a flag detected from a video stream and a 3:2 pull- down information correction circuit 6 corrects pull-down information by taking continuity of a telecine pattern into account. The pull-down information is stored in a memory 7 and the inverse telecine processing and coding processing are conducted again based on the pull-down information. Thus, disturbance of a decoded image of a video stream is prevented.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to an image compression encoding apparatus for compressing a video signal after inverse telecine processing.

[0002]

2. Description of the Related Art Conventionally, when an image created for a movie is broadcast by television broadcasting or the like, a frame (frame) (hereinafter, referred to as a film frame) recorded on a film of the movie and a frame of an NTSC signal (hereinafter referred to as a film frame) are used. Hereafter, NTS
C-frame). In a telecine conversion device, an image photographed on a film is electronically scanned and converted into an image signal.

In a movie, 24 frames are imaged every second. In contrast, NTSC signals are 29.97 frames per second (about 30
(Frame), one frame is composed of two fields. Therefore, in telecine conversion, 24 frames per second must be converted to an image of 30 frames per second (60 fields / second). Therefore, 3: 2 pull-down conversion in which one film frame is allocated to three fields or two fields is performed. Such a conversion order of the 3: 2 pull-down conversion is called a telecine pattern.

FIG. 15 is an explanatory diagram showing such a telecine conversion from a film frame to an NTSC frame.

[0005] The bold line shown in the upper part of FIG. 15 shows a film frame, and the middle line in the lower part on the upper and lower sides shows the NTSC field. NTS from film frame
At the time of conversion to a C frame, as shown in FIG. 15, an NTSC signal of 5 frames (10 fields) from 4 frames is created.

[0006] By the way, it is conceivable to compress and encode a telecine-converted video signal.

FIG. 16 is an explanatory diagram showing such an inverse telecine conversion from an NTSC frame to a film frame.

[0008] NTSC field image data (hereinafter referred to as
When converting the field data into movie frames, the telecine converter creates four frames from ten fields of field data. The middle line in the upper and lower two rows in FIG. 16 indicates the field data, and the thick line in the lower row indicates the film frame. It should be noted that the upper side of the middle line indicating the field data in FIG. 16 indicates the top field, and the lower side indicates the bottom field.

When two fields of field data are created based on one film frame in the telecine conversion, an original one film frame is created from the two fields of field data. If three field data are created based on one film frame in the telecine conversion, one film frame is created using two field data among the three field data.

FIG. 16 shows such an inverse telecine conversion. In this case, whether two-field data or three-field data is generated from the original one film frame is determined by detecting the correlation of the field data every other field.

That is, if the correlation of the field data every other field is small, it is determined that these field data are created from different film frames, and if the correlation is large, these field data are It is determined that they are created from the same film frame.

As shown in FIG. 16, when the correlation is small, one film frame is created from the two field data. When the correlation is large, the first two field data out of the three field data are generated. To make one film frame. That is, as shown in FIG. 16, when the correlation between the top fields is small, the first top field and the next bottom field are used for compression encoding, and when the correlation between the bottom fields is small, the first field is used. Are used for compression encoding. When the correlation between the top fields is large, the field data of two fields among the three fields from the first top field to the next top field is used for the compression encoding. Uses field data of two fields out of three fields from the first bottom field to the next bottom field for compression encoding. Thus, in the example of FIG.
The 3: 2 pull-down is detected by the correlation of the images.

FIG. 17 is a block diagram showing a conventional image compression encoding apparatus for performing inverse telecine conversion by detecting 3: 2 pulldown.

An input terminal 19 receives a video signal converted from a film frame to an NTSC frame by 3: 2 pull-down conversion. This video signal is supplied to the inverse telecine circuit 21 and also to the 3: 2 pull-down detection circuit 20.

The 3: 2 pull-down detection circuit 20 detects the field correlation of every other field of the input video signal and detects 3: 2 pull-down. This gives 3:
The two-pull-down detection circuit 20 determines whether the input video signal is an image having one film frame in three fields or an image having one film frame in two fields, and supplies the determination result to the inverse telecine circuit 21.

The inverse telecine circuit 21 performs an inverse telecine process based on the result of the judgment by the 3: 2 pull-down detection circuit 20, and outputs a video signal of 24 frames / sec to the video compression circuit 22. The video compression circuit 22 is an MPEG
A video stream is output by performing compression encoding based on the two standards.

As described above, in the apparatus shown in FIG. 17, the video signal after the inverse telecine processing is compression-encoded, and one image (1 image) is used as compared with the case where the inverse telecine conversion processing is not performed. Picture)). In addition, since the input interlaced image is converted into a progressive image, the compression encoding process can be performed in a frame unit instead of a field unit, and the image quality of a decoded image can be improved.

In the 3: 2 pull-down detection, the 3: 2 pull-down is detected by the correlation between the fields. However, when noise is mixed in an image, etc., 3: 2
Pulldown may not be detected correctly. For example, if an image that originally forms one film frame with three fields is erroneously determined to be one film frame with two fields, there is a problem that the compression efficiency is reduced.

Further, when a video signal is edited, it is conceivable that an inverse telecine process cannot be accurately performed on an image near an editing point. In this case, the image that has been correctly subjected to the inverse telecine processing is compressed as a progressive image, but if the inverse telecine processing is disabled, the image is compressed as it is in the original interlace image. A difference occurs in the image quality of the image.

Also, for an image with little motion, even if it is an image of one film frame in two fields, it may be erroneously determined to be an image of one film frame in three fields. The phenomenon that the movement of the camera becomes strange occurs.

[0021]

As described above, in the above-described conventional image compression coding apparatus, when 3: 2 pulldown cannot be detected correctly due to noise or the like, the compression efficiency is reduced. There was a problem that it was lowered. In addition, when the video signal is edited, there is a problem that a portion that is correctly subjected to inverse telecine processing and a portion that is not are generated, resulting in a difference in the image quality of the decoded image. Further, there is also a problem that the detection of 3: 2 pull-down may be erroneously performed on an image with a small amount of motion, and a phenomenon that the motion of the decoded image becomes strange may occur.

The present invention has been made in view of the above problems, and an object of the present invention is to provide an image compression encoding apparatus capable of improving compression efficiency by reliably detecting 3: 2 pulldown. Aim.

Another object of the present invention is to provide an image compression encoding apparatus capable of making the image quality uniform by reliably detecting 3: 2 pulldown.

Another object of the present invention is to provide an image compression encoding apparatus capable of preventing the motion of a decoded image from being abnormal by reliably detecting 3: 2 pulldown.

[0025]

According to the present invention, there is provided an image compression / encoding apparatus which receives an encoded stream obtained by encoding an inverse telecine-processed video signal and converts the encoded stream into a pull-down pattern employed in the inverse telecine processing. Pull-down information detecting means for detecting pull-down information based on the information, correcting means for correcting the pull-down information,
Inverse telecine means for performing inverse telecine processing on the video signal before inverse telecine processing based on the output of the correction means, and encoding means for encoding the output of the inverse telecine means again and outputting an encoded stream. Things.

In the present invention, the pull-down information detecting means receives the coded stream and detects pull-down information based on the pull-down pattern employed in the inverse telecine processing. The correction unit corrects the pull-down information. Based on the corrected pull-down information, the inverse telecine means performs inverse telecine processing again, and the video signal after the inverse telecine processing is encoded again by the encoding means.

[0027]

Embodiments of the present invention will be described below in detail with reference to the drawings. FIG. 1 is a block diagram showing an embodiment of an image compression encoding apparatus according to the present invention.

A video signal is input to the input terminal 19 on a field basis. This video signal has been converted from a film frame to an NTSC frame by 3: 2 pull-down conversion. This video signal (field data)
Is supplied to the inverse telecine circuit 1 and
It is also supplied to the 3: 2 pull-down detection circuit 2.

The 3: 2 pull-down detection circuit 2 has a memory (not shown), accumulates input field data, and detects an image correlation based on field data of every other field. For example, the 3: 2 pull-down detection circuit 2 accumulates the difference between the pixel values of the pixels at the same position on the screen of the first and last field data among the three consecutive field data over the entire image, and And a predetermined threshold value are compared to detect a correlation between the images. When the correlation between the two images is large, the 3: 2 pull-down detection circuit 2 determines that the field data of these three consecutive fields are based on the same film frame, and determines the correlation between the two images. Is smaller, it is determined that the field data of these three consecutive fields are not based on the same film frame.

When the 3: 2 pull-down detecting circuit 2 determines that the field data of the three fields used for detecting the 3: 2 pull-down is based on the same film frame, the field data of these three fields is used. A detection result indicating a pattern (pull-down pattern) for restoring one frame using data is output.
Also, the 3: 2 pull-down detection circuit 2 provides information indicating whether the first field among the three fields used for 3: 2 pull-down detection is a top field or a bottom field (hereinafter referred to as top field information). ) Is also output.

Conversely, if the accumulated value exceeds a predetermined threshold, the 3: 2 pull-down detection circuit 2 determines that the field data before and after the two fields are based on different film frames, and It is determined that only the field data of the first two fields of the three consecutive fields are based on the same film frame. In this case, the 3: 2 pull-down detection circuit 2 outputs a detection result indicating a pattern (pull-down pattern) for restoring one film frame using the field data of these two fields together with the top field information.

The detection result (hereinafter, referred to as repeat information) from the 3: 2 pull-down detection circuit 2 and the top field information are a combination pattern (pull-down pattern) of field data for restoring one film frame.
Is shown. The 3: 2 pull-down detection circuit 2 outputs a detection result of the pull-down pattern based on the repeat information and the top field information. These pieces of information are supplied to the inverse telecine circuit 1 via the switch 3.

The inverse telecine circuit 1 creates a film frame from the input field-based video signal based on the repeat information and the top field information. That is, when repeat information indicating that field data of two fields is based on one film frame is input, the inverse telecine circuit 1 creates one film frame from the field data of these two fields, When repeat information indicating that the field data is based on one film frame is input, one film frame is created from the field data of the first two fields among these three fields.

The video signal of 24 frames / second from the inverse telecine circuit 1 is supplied to the video compression circuit 4. The video compression circuit 4 compresses an input video signal and outputs a video stream by a predetermined compression coding such as the MPEG2 standard. Also,
The video compression circuit 4 also outputs an encoding log in which the result of the compression encoding process is recorded.

For example, in the MPEG2 standard, a flag called repeat_first_field for transmitting repeat information and a flag called top_field_first for transmitting top field information are provided. The repeat information and the top field information based on the detection of the 2 pull-down are transmitted on the video stream by these flags. Also, the same information can be transmitted by the encoding log.

The 3: 2 pull-down information generation circuit 5 indicates what inverse telecine processing has been performed on the video signal input via the terminal 19 based on information such as the video stream and the encoding log. : 2 pull-down information is generated. The 3: 2 pull-down information is supplied to the 3: 2 pull-down information correction circuit 6.

The 3: 2 pulldown information correction circuit 6 corrects the 3: 2 pulldown information based on the input 3: 2 pulldown information and the like. In this case, the 3: 2 pull-down information correction circuit 6 can correct the 3: 2 pull-down information in consideration of the continuity of the telecine pattern. Further, the 3: 2 pull-down information correction circuit 6 can also provide information on a correction range, information on a reference image, information on a division boundary, and the like from outside by an operator. Based on the information,
The 3: 2 pulldown information can be corrected. The 3: 2 pull-down information correction circuit 6 has various correction modes. By appropriately applying these various correction modes, the original material, that is, the original film frame before the telecine conversion is faithfully reproduced. The 3: 2 pull-down information is corrected so as to perform an inverse telecine process that can be performed.

The 3: 2 pulldown information corrected by the 3: 2 pulldown information correction circuit 6 is supplied to a 3: 2 pulldown memory 7. Memory 7 is corrected 3: 2
The pull-down information is stored and supplied to the inverse telecine circuit 1 via the switch 3.

The inverse telecine circuit 1 receives the corrected 3: 2 pull-down information via the switch 3 and performs inverse telecine processing on the video signal from the input terminal 19 again using the 3: 2 pull-down information. It has become. The video signal subjected to the inverse telecine processing again by the inverse telecine circuit 1 is supplied to the video compression circuit 4, and the video compression circuit 4 compresses and encodes the input video signal again and outputs a video stream. .

Next, the operation of the embodiment configured as described above will be described with reference to FIGS. FIG. 2 is a flowchart for explaining the operation of the embodiment. FIG. 3 is an explanatory diagram for explaining the operation of the embodiment. FIG. 3A shows a 3: 2 pull-down detection circuit 2
And FIG. 3D shows the inverse telecine processing based on the corrected 3: 2 pull-down information. In addition, FIG.
(C) shows a process of correcting 3: 2 pull-down information, and shows a reference image and correction information, respectively. Also,
FIGS. 3A to 3D show upper and lower black circles.
It shows the field data of the top field or the bottom field, respectively. The horizontal direction in FIG. 3 corresponds to the time in field units. In FIG. 3, a solid line frame surrounding a black circle indicates a combination pattern (pull-down pattern) of field data determined or created as the same film frame.

As is apparent from the solid frame in FIG. 3, the pattern (pull-down pattern) of the combination of field data for restoring one film frame is 4
Kind can be considered. That is, a pattern (hereinafter, referred to as a pattern P1) for restoring a film frame by two field data of a top field and a bottom field following the top field, and three fields of a top field and a bottom and a top field following the top field A pattern for restoring a film frame by data (hereinafter referred to as pattern P2),
A pattern (hereinafter referred to as a pattern P3) for restoring a film frame by two field data of a bottom field and a top field following the bottom field, and three field data of a bottom field and a top and bottom field following the bottom field. There are four patterns for restoring the film frame (hereinafter referred to as pattern P4).

Now, it is assumed that the video signal shown in FIG. This video signal is obtained by subjecting the film frame shown in FIG. 3D to telecine conversion. This video signal is supplied to an inverse telecine circuit 1 and a 3: 2 pull-down detection circuit 2, and a 3: 2 pull-down detection and encoding process are performed in step S1 in FIG.

The 3: 2 pull-down detection circuit 2 detects the correlation between the first and last fields of the continuous field data of three fields to detect a pull-down pattern. Based on the detection result of the 3: 2 pull-down detection circuit 2, the inverse telecine circuit 1 performs an inverse telecine process on the input video signal.

When the 3: 2 pull-down detection circuit 2 detects a pull-down pattern for creating one film frame from the first two fields of the field data of the three fields used for detection, the 3: 2 pull-down detection circuit 2 detects these three fields. The last field is the first field for detecting a pull-down pattern using the next three consecutive fields.

Now, it is assumed that the detection result of the pull-down pattern is shown by the solid frame in FIG. 3A due to the influence of noise and the like. The inverse telecine circuit 1 performs an inverse telecine process based on the detection result of the pull-down pattern. That is, for the period A in FIG. 3A, a film frame using only two fields of data is continuously created.

The output of the inverse telecine circuit 1 is supplied to a video compression circuit 4 and subjected to a predetermined compression encoding such as MPEG2. In the video stream from the video compression circuit 4, a repeat first field flag and a top field first flag are inserted. The video compression circuit 4 also outputs an encoding log indicating a log of the inverse telecine processing and the compression processing.

The 3: 2 pull-down information generation circuit 5 detects 3: 2 pull-down information from the output of the video compression circuit 4 in step S2. The 3: 2 pull-down information generation circuit 5 outputs the detected pull-down information to the 3: 2 pull-down information correction circuit 6.

The 3: 2 pull-down information correction circuit 6 detects a portion where the inverse telecine processing may not be performed normally in step S3. Normal telecine patterns are: patterns, P1, P2, P3, P4,
.. Can be considered to repeat. For example, 3:
The 2 pull-down information correction circuit 6 detects from the repeat first field flag that the period A in FIG. 3A is continuously the same pull-down pattern, and determines an abnormal portion of the inverse telecine processing.

When the 3: 2 pull-down information correction circuit 6 detects in step S4 that there is an abnormality in the pull-down detection, the process shifts to step S5 to correct the 3: 2 pull-down information. For example, the pull-down information correction circuit 6 corrects the 3: 2 pull-down information based on the continuity of the telecine pattern.

That is, the pull-down information correction circuit 6 uses the patterns immediately before and after the period A as reference images and creates a pull-down pattern for maintaining the continuity of the telecine pattern using field data existing between these patterns. Detect whether or not can be done. FIG. 3B shows a pattern serving as a reference image. Since the reference image B1 is a pattern P4 and the reference image B2 is a pattern P2, patterns P1, P2,.
, P1 can be created.

In one cycle of the telecine pattern, ten field data are required.
.Times.n (n is an integer) +2 field data may be present between the reference images B1 and B2. Reference image B1,
Since there are twelve field data between B2, the 3: 2 pull-down information correction circuit 6 uses the reference image B1
, B2, correction information for inserting a pull-down pattern for maintaining the continuity of the telecine pattern (FIG. 3)
(C)) is created, and the period A of the original 3: 2 pull-down information is replaced with this correction information. The 3: 2 pull-down information correction circuit 6 creates the corrected 3: 2 pull-down information (FIG. 3D) and outputs it to the 3: 2 pull-down memory 7.

As described above, since the telecine pattern does not change during the normal telecine processing,
The pull-down pattern for the period A can be easily derived. Although the correction information is obtained using the pull-down pattern immediately before and immediately after the portion where the 3: 2 pull-down is not normally detected as a reference image, a predetermined correction range is designated, and within this range, the continuation of the telecine pattern is performed. Obviously, the 3: 2 pull-down information may be corrected so as to maintain the characteristics.

Next, inverse telecine processing and video compression processing are performed again on the same video signal. In this case, the switch 3 is a 3: 2 pull-down memory 7
Is supplied to the inverse telecine circuit 1. Inverse telecine circuit 1
Performs inverse telecine processing on the input field data according to the pull-down pattern shown in FIG.
The video signal is converted into a video signal of 24 frames / sec and is converted
To supply. The video compression circuit 4 performs compression coding such as the MPEG2 standard on the input video signal and outputs a video stream.

Incidentally, whether or not the inverse telecine processing has been normally performed cannot be confirmed unless the encoded video is actually decoded and viewed. That is,
There is no guarantee that the result of the automatic correction is always correct.

Therefore, the 3: 2 pull-down information correction circuit 6
In consideration of this point, the camera has various correction modes in addition to the automatic correction mode as shown in FIG.

FIG. 4 is an explanatory diagram for explaining another correction mode of the 3: 2 pull-down information correction circuit 6. FIG. 4 corresponds to FIG. 3, in which FIG. 4A shows 3: 2 pull-down information before correction, FIG. 4B shows correction information, and FIG.
(C) shows the corrected 3: 2 pull-down information.

The correction mode shown in FIG. 4 results in invalidating the inverse telecine processing. Now 3:
It is assumed that the 2 pull-down information generation circuit 5 has acquired the pull-down information shown in FIG. As shown in FIG. 4A, the pull-down information maintains the continuity of the telecine pattern.

The 3: 2 pull-down information correction circuit 6 corrects the input pull-down information. For example, the 3: 2 pull-down information correction circuit 6 changes the repeat first field flag to information (= “0”) in which two fields constitute one picture image in the correction range of FIG. FIG. 4B shows correction information based on this flag.

The 3: 2 pull-down information correction circuit 6 is configured as shown in FIG.
The corrected 3: 2 pull-down information shown in FIG.
And store it. The inverse telecine circuit 1 performs the inverse telecine processing again based on the corrected 3: 2 pull-down information. In this way, the inverse telecine processing shown in FIG. 4C is performed to create a film frame.

That is, using the corrected 3: 2 pull-down information is equivalent to not performing the inverse telecine processing on the correction range. For example, if the decoded image cannot be normally reproduced from the video stream obtained by the video compression circuit 4 as a result of performing the automatic correction mode of FIG. 3, the correction mode of FIG. 4 is adopted.

When frame processing is performed on an image in the correction range, the motion of the decoded image may be abnormal if the inverse telecine processing is performed and the compression encoding is performed. Also in this case, by adopting the correction mode of FIG. 4, it is possible to prevent a problem from occurring in the video stream.

FIG. 5 is an explanatory diagram for explaining another correction mode of the 3: 2 pull-down information correction circuit 6. FIG.
5A shows an abnormal portion (shaded portion) of pull-down detection based on 3: 2 pull-down information before correction, FIG. 5B shows a correction range, and FIG. 5C shows 3: 2 pull-down information after correction. Is shown.

The correction mode shown in FIG. 5 integrates a plurality of areas where the pull-down detection is determined to be abnormal, and performs correction based on the continuity of the telecine pattern for the integrated area.

For example, in a still image or an image with little motion, even if the images are different, the field correlation is large. Therefore, when 3: 2 pulldown is detected, it may be erroneously determined to be the field data based on the same image. is there.
In such a case, although the pull-down detection is normally abnormal, the telecine pattern may be continuous for a part of the period, and it may be possible to determine that the pull-down has been normally detected.

FIG. 5A shows such a state. The thick line portion in FIG. 5A indicates an area where the continuity of the telecine pattern is first confirmed in the 3: 2 pull-down information correction circuit 6 and it is determined that the pull-down detection has been normally performed. The shaded portion in FIG. 5A indicates that the telecine pattern is discontinuous.
Areas N1 and N where pull-down detection is determined to be abnormal
2 is shown.

The 3: 2 pull-down information correction circuit 6 first determines whether or not the same correction as in FIG. 3 can be performed on the areas N1 and N2. Here, these areas N1,
It is assumed that the correction shown in FIG. 3 cannot be performed for any of N2. In this case, the 3: 2 pull-down information correction circuit 6 sets the entire period from the head position of the area N1 to the end position of the area N2 as a correction range, and can perform correction for maintaining the continuity of the telecine pattern in this correction range. It is determined whether or not.

When it is determined that the correction is possible in the correction range shown in FIG. 5B, the 3: 2 pull-down information correction circuit 6 operates to maintain the continuity of the telecine pattern in this correction range. Correction information is created, and corrected 3: 2 pull-down information is created. By using the corrected 3: 2 pull-down information for the second inverse telecine processing, a normal video stream can be obtained.

FIG. 5 shows an example in which there are two portions where the detection of 3: 2 pull-down is determined to be abnormal.
It is clear that any number of abnormal portions of 3: 2 pulldown detection are effective.

FIG. 6 is an explanatory diagram for explaining another correction mode of the 3: 2 pull-down information correction circuit 6. 6A and 6B correspond to FIG. 3, FIG. 6A shows 3: 2 pull-down information before correction, FIG. 6B shows a reference image, and FIG.
FIG. 6C shows correction information, and FIG. 6D shows 3: 2 after correction.
Shows pull-down information.

In the correction mode shown in FIG. 6, a predetermined reference image is designated by an operation of an operator, and a pull-down pattern based on a telecine pattern is set assuming that the reference image is correct.

Now, it is assumed that the 3: 2 pull-down information generated by the 3: 2 pull-down information generating circuit 5 is as shown in FIG. The 3: 2 pull-down information correction circuit 6 receives the 3: 2 pull-down information and
The period A in (a) is a correction range. Here, the operator refers to the 3: 2 pull-down information in FIG.
An arbitrary image within the correction range is designated as a reference image by a time code, the number of frames from the head, or the like. FIG.
As shown in (a), the pull-down pattern is continuous for the images immediately before and after the reference image (FIG. 6 (b)).

The 3: 2 pull-down information correction circuit 6 sets a pull-down pattern according to the telecine pattern before and after the reference image. The 3: 2 pull-down information correction circuit 6 determines that the set pull-down pattern is correct when the telecine pattern is continuous at the boundary of the correction range.

FIG. 6C shows a pull-down pattern set based on the reference image. The first pull-down pattern in the correction range is the pattern P4, and the pattern of the immediately preceding image is the pattern P3. Therefore, the telecine pattern is continuous at this boundary. Also,
Since the last pull-down pattern in the correction range is the pattern P4, and the pattern of the next image is the pattern P1, the telecine pattern is continuous even at this boundary portion. Therefore, the 3: 2 pull-down information correction circuit 6
Replaces the generated pull-down pattern as correction information with a pattern in the correction range of the input 3: 2 pull-down information. Thus, normal 3: 2 pull-down information shown in FIG. 6D can be obtained.

In FIG. 6, the correction is performed over the entire correction range. However, the correction range is divided into several correction ranges.
By specifying a reference picture for each of the divided correction ranges, the correction can be similarly performed. If it is known in advance that the video to be compression-encoded is edited only in frame units, that is, if it is known that the image always starts from the top field, the specified division correction range Even if the boundary of the data is compressed and encoded in the bottom field, the division is automatically performed in the top field, so that the boundary division according to the actual editing can be performed. Correction is possible.

FIG. 7 is an explanatory diagram for explaining another correction mode of the 3: 2 pull-down information correction circuit 6. FIG. 7 corresponds to FIG. 3. FIG. 7A shows 3: 2 pull-down information before correction, FIG. 7B shows a reference image, and FIG.
7C and 7D show correction information, and FIG. 7E shows 3: 2 pull-down information after correction.

The correction mode shown in FIG. 7 corresponds to a video signal or the like that is being edited, and specifies an editing point or the like to specify a discontinuous point of the telecine pattern and divide the correction range. It was made.

Assume that the 3: 2 pull-down information generated by the 3: 2 pull-down information generating circuit 5 is as shown in FIG. The 3: 2 pull-down information correction circuit 6 receives the 3: 2 pull-down information and
The period A in (a) is a correction range. Here, the operator instructs the boundary of the division, for example, to divide the correction range at the editing point. 3: 2 pulldown information generation circuit 5
Means that the image immediately before and after the correction range is the reference image B1,
Set as B2.

The 3: 2 pull-down information correction circuit 6 sets a pull-down pattern according to the telecine pattern before and after with reference to the reference images B1 and B2. When the pull-down pattern does not cross the division boundary at the division boundary portion of the correction range, the 3: 2 pull-down information correction circuit 6
It is determined that the set pull-down pattern is correct.

FIG. 7C shows a pull-down pattern set on the basis of the reference image B1.
(D) shows a pull-down pattern set based on the reference image B2. Based on the reference image B1,
A pattern P2 is set near the division boundary. However, the pattern P2 includes, as an element, field data F1 that exceeds the division boundary. Therefore, the 3: 2 pull-down information correction circuit 6 determines that the pull-down pattern cannot be corrected based on the reference image B1.

On the other hand, with reference to the reference image B2, a pattern P2 is set near the division boundary. The field data constituting this pattern P2 is located at the boundary end within the correction range. Accordingly, the 3: 2 pull-down information correction circuit 6 determines that the correction of the pull-down pattern based on the reference image B2 is possible.

Therefore, the 3: 2 pull-down information correction circuit 6
Uses only the pull-down pattern shown in FIG. 7D as correction information, and replaces the divided correction range of the input 3: 2 pull-down information with this pattern. The input 3: 2 pull-down information is used as it is for the division correction range based on the reference image B1. Thus, FIG.
3: 2 pull-down information shown in FIG.

As described above, when the video editing point exists in the area where the 3: 2 pull-down detection is abnormal, the telecine pattern is discontinuous at the editing point. Therefore, by dividing the correction range at the editing point, the continuity of the telecine pattern can be maintained within the divided correction range.

Considering the continuity of the telecine pattern 3:
By generating the two pull-down information, it can be corrected to correct 3: 2 pull-down information.

If it is known in advance that the video to be compression-encoded is edited only in frame units, even if the boundary of the designated divided correction area is compression-encoded in the bottom field. By automatically performing division at the top field, boundary division according to actual editing can be performed, and effective 3: 2 pull-down correction can be performed.

FIG. 8 is an explanatory diagram for explaining another correction mode of the 3: 2 pull-down information correction circuit 6. 8 corresponds to FIG. 3, FIG. 8A shows 3: 2 pull-down information before correction, FIG. 8B shows a plan of correction information,
FIG. 8C shows the correction information.

The correction mode shown in FIG. 8 corresponds to a video signal or the like being edited. Generally, a telecine pattern is not considered when editing a video. Therefore,
When editing is performed, it is normally 3:
In many cases, two pull-down conversion cannot be performed. FIG.
Considers this point.

In the correction modes shown in FIGS. 6 and 7, (division)
If the telecine pattern is not continuous at the boundary position of the correction range, it is determined that correction is impossible. In contrast, FIG.
Is an example in which a part of the correction information is used even when the telecine pattern is not continuous at the boundary of the divided correction range.

Now, it is assumed that the 3: 2 pull-down information generated by the 3: 2 pull-down information generating circuit 5 is as shown in FIG. An operator specifies a division boundary of the correction range. 3: 2 pulldown information generation circuit 5
Sets, for example, the image immediately before the correction range as the reference image B1.

The 3: 2 pull-down information correction circuit 6 sets a pull-down pattern according to the telecine pattern with reference to the reference image B1. When the 3: 2 pull-down information correction circuit 6 determines that the set pull-down pattern is correct when the pull-down pattern does not cross the division boundary at the division boundary portion of the correction range, it is determined that the pull-down pattern is correct as shown in FIGS. Is the same as

However, when the reference image B1 is used as a reference, as shown in FIG. 8B, the pull-down pattern becomes a pattern P2 near the division boundary, and the pattern P2 has the field data F1 beyond the division boundary as an element. .
Therefore, the pull-down pattern set based on the reference image B1 is discontinuous at the division boundary.

However, even in this case, the 3: 2 pull-down information correction circuit 6 makes the correction information valid by changing only the pull-down pattern near the division boundary as shown in FIG. 8C. In FIG. 8C, by making the pull-down pattern of the pattern P1 continuous near the division boundary, correction within the division correction range is enabled.

Thus, the 3: 2 pull-down information correction circuit 6 corrects the division correction range of the input 3: 2 pull-down information using the correction information shown in FIG. FIG.
By using the correction mode described above, it is possible to normally correct most of the correction range.

In FIG. 8, 3: 2 pull-down information is not corrected in a part of the correction range near the division boundary, and one film frame is set by two field data. The video compression circuit 4
For a portion where 3: 2 pull-down detection is normally performed, compression encoding is performed as a progressive frame. However, an image that forms one film frame with two field data without correction is not necessarily an interlaced image.

Therefore, if it can be determined that the image is not an interlaced image, the image is determined to be a progressive frame, and if the image is an interlaced image, the compression code can be set without setting the progressive frame. Perform the conversion. For a picture set as a progressive frame, only frame processing is adopted and compressed during compression encoding processing, so that image quality can be improved compared to a video stream field-processed as an interlaced image.

FIG. 9 is an explanatory diagram for explaining another correction mode of the 3: 2 pull-down information correction circuit 6. FIG. 9 corresponds to FIG. 8, FIG. 9A shows 3: 2 pull-down information before correction, FIG. 9B shows a plan of correction information,
FIG. 9C shows the correction information.

The correction mode shown in FIG. 9 is a modification of the correction mode shown in FIG. In FIG. 8, considering that one cycle of the telecine pattern is 10 fields of NTSC, when a telecine pattern continuous by multiples of 5 from the reference image is obtained, the multiple NTSC frames of 5 are used. The correction is performed as a unit. On the other hand, in FIG. 9, if the continuity of the telecine pattern is maintained, the telecine pattern becomes 1
The correction is performed without going around.

Now, it is assumed that the 3: 2 pull-down information generated by the 3: 2 pull-down information generating circuit 5 is as shown in FIG. An operator specifies a division boundary of the correction range. 3: 2 pulldown information generation circuit 5
Sets, for example, the image immediately before the correction range as the reference image B1.

The 3: 2 pull-down information correction circuit 6 sets a pull-down pattern according to the telecine pattern with reference to the reference image B1. When the 3: 2 pull-down information correction circuit 6 determines that the set pull-down pattern is correct when the pull-down pattern does not cross the division boundary at the division boundary of the correction range, it is the same as the example of FIG. is there.

However, when the reference image B1 is used as a reference, as shown in FIG. 9B, the pull-down pattern becomes a pattern P4 near the division boundary, and this pattern P4 is composed of the field data F1 and F2 that cross the division boundary. And Therefore, the pull-down pattern set on the basis of the reference image B1 is discontinuous at the division boundary.

In this case, as shown in FIG. 9C, the 3: 2 pull-down information correction circuit 6 makes the change of the pull-down pattern P2 near the division boundary effective, and the 3: 2 pull-down information correction circuit 6 from the pattern P2 to the division boundary. A pattern P4 is created by combining the two field data F5, F6 and F7.
In the example of FIG. 8, two patterns P near the division boundary
Although 1 has been corrected to be continuous, in FIG. 9, only the last pull-down pattern is set to a pattern different from the telecine pattern.

Thus, the 3: 2 pull-down information correction circuit 6 corrects the division correction range of the input 3: 2 pull-down information using the correction information shown in FIG. 9C. FIG.
By using the correction mode of FIG.
It is possible to expand more than.

In the pattern P4 immediately before the division boundary, the field data F5 and F6 are based on the same image, and the field data F6 and F7 are based on different images. However, in the actual compression encoding, the field data F7 immediately before the division boundary is not used, and only the first two field data F5 and F6 are used. Therefore, the video stream created by the compression encoding is decoded. Images are of little concern to the appearance.

FIG. 10 shows a 3: 2 pull-down information correction circuit 6.
FIG. 9 is an explanatory diagram for describing another correction mode. FIG.
0 corresponds to FIG. 8, and FIG. 10A shows 3:
10 (b) shows the reference image, FIG. 10 (c) shows the correction information, and FIG. 10 (d) shows the 3: 2 pull-down information after the correction.

The correction mode shown in FIG. 10 is a combination of the correction modes shown in FIGS. 4 and 8. In FIG. 8, a part of the correction information is used even when the telecine pattern is not continuous at the boundary of the divided correction range. However, when a video stream obtained by compression-encoding again using the corrected 3: 2 pull-down information is decoded, a normal image may not be obtained at a division boundary. FIG. 10 does not correct the vicinity of the division boundary in this case.

Now, the 3: 2 pull-down information generated by the 3: 2 pull-down information generating circuit 5 is shown in FIG.
It is assumed to be as shown in FIG. An operator specifies a division boundary of the correction range. The 3: 2 pull-down information generation circuit 5 sets, for example, the image immediately before the correction range as the reference image B1 (FIG. 10B).

The 3: 2 pull-down information correction circuit 6 sets a pull-down pattern according to the telecine pattern with reference to the reference image B1. When the 3: 2 pull-down information correction circuit 6 determines that the set pull-down pattern is correct when the pull-down pattern does not cross the division boundary at the division boundary of the correction range, it is the same as the example of FIG. is there.

However, when the reference image B1 is used as a reference, as shown in FIG. 10C, the pull-down pattern is set across the division boundary according to the telecine pattern. Therefore, the pull-down pattern set based on the reference image B1 is discontinuous at the division boundary.

In this case, as shown in FIG. 10D, the 3: 2 pull-down information correction circuit 6 invalidates the change of the pull-down patterns P2 and P3 in the vicinity of the division boundary, and changes the pattern up to the immediately preceding pattern P1. Valid.

Thus, the 3: 2 pull-down information correction circuit 6 corrects the division correction range of the input 3: 2 pull-down information using the correction information shown in FIG. That is, the correction according to the telecine pattern is performed on the reference image B1 side, and the correction is invalidated near the boundary of the correction range opposite to the reference image B1.

For example, at the start point and end point of the compression encoding and the break point (chapter point) of the compression encoding,
3: 2 pulldown may not be detected. In this case, one of the portions where 3: 2 pull-down detection has not been performed cannot be used as a reference. FIG. 10 shows an effective process in this case.

FIGS. 11 and 12 are explanatory diagrams for explaining another correction mode of the 3: 2 pull-down information correction circuit 6. FIG. 11 and 12 correspond to FIG.
1 (a) and 12 (a) show 3: 2 pulldown information before correction, FIGS. 11 (b) and 12 (b) show a reference image, and FIGS. 11 (c) and 12 (c). 11 shows correction information, and FIGS. 11D and 12D show 3: 2 pull-down information after correction.

The correction modes shown in FIGS. 11 and 12 are for enabling application to a DVD or the like in the correction mode shown in FIG.

Now, as shown in FIG. 11A, there is a part where the pull-down cannot be detected at the head of the correction range, and the top field first flag of the first picture g1 of the compression coding in this head is "0". ", That is, a picture starting from the bottom field.

In this case, the boundary condition is changed so that the top field first flag of the first picture becomes 1, as shown in FIG. That is, the first picture in the correction range is set to the picture g2. DVD
Etc., it is specified that the video starts from the top field. Therefore, by making corrections as shown in FIG. 11C, it is possible to support DVDs and the like.

Note that the other correction processing in FIG. 11C is the same as that in FIG. Thus, the 3: 2 pull-down information shown in FIG. 11D is obtained by updating the 3: 2 pull-down information based on the correction information in FIG.

Conversely, as shown in FIG. 12 (a), there is a part where the pull-down cannot be detected at the end of the correction range, and the top field first flag of the last picture g3 of the compression encoding at this end is 0. And the repeat first field flag is 0, that is, the picture ends with the top field.

In this case, as shown in FIG. 12 (c), the top field first flag of the last picture is 1 and the repeat first field flag is 0, or the top field first flag is 0.
And the boundary condition is changed so that the repeat first field flag is set to 1. That is, the end of the correction range is set to, for example, the picture g4.

In a DVD or the like, it is specified that an image ends with a bottom field. Therefore, FIG.
By making corrections as shown in (c), it is possible to support DVDs and the like.

The other correction processing in FIG. 12C is the same as that in FIG. Thus, by updating the 3: 2 pull-down information based on the correction information of FIG. 12C, the 3: 2 pull-down information shown in FIG. 12D is obtained.

FIG. 13 is a block diagram showing another embodiment of the present invention. 13, the same components as those in FIG. 1 are denoted by the same reference numerals, and description thereof will be omitted.

This embodiment differs from the embodiment of FIG. 1 in that a video decoding circuit 8 is added. The output of the video compression circuit 4 is provided to the video decoding circuit 8. Video decoding circuit 8
Is configured to decode an input video stream and output a decoded image. Note that the video decoding circuit 8
A display unit (not shown) is provided so that the decoded image can be displayed as it is, as well as displayed by switching between a top field and a bottom field.

Whether the 3: 2 pull-down detection for the inverse telecine processing has been normally performed is ultimately determined by:
Unless the encoded video stream is decoded and viewed, it cannot be confirmed. The video decoding circuit 8 is provided for such a decoding process.

Next, the operation of the embodiment thus configured will be described with reference to the explanatory diagram of FIG. FIG.
FIG. 4A shows a display example by normal inverse telecine processing, and FIG. 14B shows a display example by abnormal inverse telecine processing.

The video stream compressed by the video compression circuit 4 is supplied to a video decoding circuit 8. The video decoding circuit 8 decodes, for example, an NTSC frame in which the repeat first field is 1, that is, three fields, and displays a top field and a bottom field.

Now, it is assumed that 3: 2 pull-down detection is normally performed. In this case, an image is normally displayed on the screen as shown in FIG.

On the other hand, when an error occurs in 3: 2 pull-down detection, for example, when an NTSC frame is constituted by field data of a different field, for example, as shown in FIG. The image above is blurred and not displayed properly. The operator
By looking at the decoded image, it is determined whether 3: 2 pulldown detection has been correctly performed.

As described above, in the present embodiment, it is possible to reliably determine whether 3: 2 pull-down detection has been normally performed.

By using the present embodiment, it is possible to specify a reference picture in the correction modes shown in FIGS. By judging the decoded video with the naked eye, a reference picture can be accurately specified.

Furthermore, the video decoding circuit 8 can display the top field and the bottom field separately, and can also display these images while switching them. This makes it easier to determine the reference image.

The present embodiment is effective as a means for designating a picture as a division boundary in the correction modes shown in FIGS. That is, the picture at the division boundary is decoded by the video decoding circuit 8 to display both fields. It is determined whether the displayed image looks like one normal image as shown in FIG. 14A or two abnormal images as shown in FIG. 14B.

In this determination, if the image is determined to be a normal single image, the first field of the designated picture is set as a boundary for division, and if it is determined that the image is two, the second field of the specified picture is determined. Is the boundary of the division.

As described above, the boundary can be set accurately and effectively by making a decision by looking at the decoded image with the naked eye, thereby enabling effective correction of 3: 2 pull-down information. .

[0133]

As described above, according to the present invention,
By reliably detecting 3: 2 pull-down, the compression efficiency can be improved, the image quality can be made uniform, and the motion of the decoded image can be prevented from being abnormal. Having.

[Brief description of the drawings]

FIG. 1 is a block diagram showing an embodiment of an image compression encoding apparatus according to the present invention.

FIG. 2 is a flowchart for explaining the operation of the embodiment of FIG. 1;

FIG. 3 is an explanatory diagram for explaining a correction mode of a 3: 2 pull-down information correction circuit 6.

FIG. 4 is an explanatory diagram for explaining a correction mode of a 3: 2 pull-down information correction circuit 6.

FIG. 5 is an explanatory diagram for explaining a correction mode of a 3: 2 pull-down information correction circuit 6.

FIG. 6 is an explanatory diagram for explaining a correction mode of a 3: 2 pull-down information correction circuit 6.

FIG. 7 is an explanatory diagram for explaining a correction mode of a 3: 2 pull-down information correction circuit 6.

FIG. 8 is an explanatory diagram for explaining a correction mode of a 3: 2 pull-down information correction circuit 6.

FIG. 9 is an explanatory diagram for explaining a correction mode of the 3: 2 pull-down information correction circuit 6.

FIG. 10 is an explanatory diagram for explaining a correction mode of the 3: 2 pull-down information correction circuit 6.

FIG. 11 is an explanatory diagram for explaining a correction mode of the 3: 2 pull-down information correction circuit 6.

FIG. 12 is an explanatory diagram for explaining a correction mode of a 3: 2 pull-down information correction circuit 6.

FIG. 13 is a block diagram showing another embodiment of the present invention.

FIG. 14 is an explanatory diagram for explaining the operation of the embodiment in FIG. 13;

FIG. 15 is an explanatory diagram for explaining telecine processing.

FIG. 16 is an explanatory diagram for explaining inverse telecine processing.

FIG. 17 is a block diagram showing a conventional image compression encoding apparatus.

[Explanation of symbols]

DESCRIPTION OF SYMBOLS 1 ... Inverse telecine circuit, 2 ... 3: 2 pull-down detection circuit, 3 ... Switch, 4 ... Video compression circuit, 5 ... 3: 2
Pulldown information generation circuit, 6 ... 3: 2 pulldown information correction circuit, 7 ... 3: 2 pulldown memory

 ──────────────────────────────────────────────────続 き Continued on the front page F term (reference) 5C022 BA13 BA19 5C059 KK36 LA05 LA06 LB07 LB13 MA00 MA14 NN00 PP11 RB14 RC24 RF00 RF01 SS00 SS13 UA02 UA05

Claims (7)

[Claims] 1. An encoded stream obtained by encoding a video signal subjected to inverse telecine processing is input, and pull-down information detecting means for detecting pull-down information based on a pull-down pattern employed in the inverse telecine processing; Correction means for correcting the inverse telecine processing of the video signal before inverse telecine processing based on the output of the correction means, and re-encoding the output of the inverse telecine means to output a coded stream. An image compression encoding apparatus comprising: encoding means. 2. A first inverse telecine unit for performing an inverse telecine processing on the input video signal based on a pull-down pattern detected from the input video signal, and an output of the first inverse telecine unit is encoded to form an encoded stream. Encoding means for outputting, pull-down information detecting means for detecting pull-down information based on a pull-down pattern employed in inverse telecine processing by the first inverse telecine means from an output of the encoding means, and correcting the pull-down information Correction means; second inverse telecine means for performing inverse telecine processing on the input video signal again based on the output of the correction means; and re-encoding the output of the second inverse telecine means to output an encoded stream. And second encoding means. Image compression encoding apparatus characterized by. 3. The pull-down information detecting means is a flag included in the coded stream, and determines whether a first field among a plurality of fields constituting an image subjected to inverse telecine processing is a top field. 2. The pull-down information is detected based on a top field first flag indicating the number of fields constituting an image subjected to inverse telecine processing and a repeat first field flag indicating the number of fields forming an image subjected to inverse telecine processing.
Or the image compression encoding apparatus according to any one of 2. 4. The image according to claim 1, wherein the pull-down information detecting unit detects the pull-down information based on an encode log of an inverse telecine process and an encoding process. Compression encoding device. 5. The image compression encoding apparatus according to claim 1, wherein the correction unit corrects the pull-down information based on a continuity of a telecine pattern. 6. The apparatus according to claim 1, wherein the correction unit corrects the pull-down information in consideration of continuity of a telecine pattern within a set correction range. Image compression encoding device. 7. The image compression code according to claim 1, wherein the correction unit corrects the pull-down information in consideration of continuity of a telecine pattern and a position of an edit point. Device.