JP3377679B2 - Coded interlaced video cut detection method and apparatus - Google Patents

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a method and apparatus for detecting a cut from coded video, and more specifically, to a coded interface suitable for detecting a cut from video data obtained by coding an interlaced video signal. The present invention relates to a race video cut detection method and device.

[0002]

2. Description of the Related Art In order to grasp the outline of a video, it is common to watch the video in time sequence. However, if a cut (change of scene) can be automatically detected from the video data, a list of scenes can be automatically displayed. Since it can be created, you can efficiently grasp the outline of the video and search for the necessary scene.

In recent years, applications of digitized video have been spreading in various fields such as communication, broadcasting and entertainment, and there is a technique for detecting a cut directly (without decoding) from encoded video data. is necessary.

Most of the conventional cut detection methods are intended for non-encoded video data. In order to handle the encoded video data, it is necessary to perform a decoding process, and it is necessary to implement it by software. Had a problem that it took a long time to process. Further, even if it is realized by hardware, there is a problem that the hardware scale becomes large.

"Scene change detecting device" (Japanese Patent Laid-Open No. 6-
No. 22304), a feature amount calculated at the time of encoding / decoding of a moving image, such as a cumulative value of prediction errors for each frame, a data amount of encoded data, or the number of pixels encoded in a frame, is described. It is used to detect the cut automatically. The principle of the cut detection will be described with reference to FIG. Figure 11
In the frame sequence of, since there is a cut between the continuous frames 22 and 23 (the scene switches from white to black), the correlation between the frames 22 and 23 becomes small, and The amount is large. Therefore, the cut can be automatically detected by comparing the above feature amount with the threshold value.

[0006]

By the way, video data (hereinafter, coded interlaced video) encoded by interlaced video represented by NTSC analog signal (video signal system generally used in television broadcasting) is used. Data) will be described with respect to problems when the above-described conventional cut detection technique is applied.

Consider the frame sequence shown in FIG. 12 as an example. In FIG. 12, the odd fields of the frame 31 and the frame 32 form one scene (white), and the even field of the frame 32 and the frame 33 form another scene (black). Such a situation where there is a cut between the even and odd fields frequently occurs when a film image (24 frames per second) is converted into an NTSC signal (30 frames per second) (telecine conversion). When the above-described conventional cut detection technique is applied to such a frame sequence, the consecutive two frames (32, 33) are cut because the feature amount has a high value in both the frame 32 and the frame 33. Duplicately detected. In order to avoid this, a continuous cut of two frames can be regarded as one cut, but another problem as described below occurs.

Consider the frame sequence shown in FIG. 13 as an example. This frame sequence schematically shows the situation in which the flash light is reflected at the time of shooting. The image that was completely dark in the frame 41 was momentarily brightened in the even field of the frame 42 and then in the next frame 43. Return to the original. When the above method is applied to this frame sequence, the feature amount is the same as that of the frame sequence of FIG.
And the frame 43 both have high values. Therefore, if an attempt is made to detect a cut between the even field and the odd field, noise such as flash light will be erroneously detected as a cut in the situation shown in FIG.

In the prior art, when trying to detect a cut generated between an odd field and an even field, instantaneous noise such as flash light is erroneously detected as a cut, and conversely erroneous detection of flash light is detected. However, there is a problem that the cut generated between the fields cannot be detected if the above problem is avoided.

The present invention has been made in order to solve the above-mentioned problems, and cuts video data encoded using a field prediction method such as MPEG2 at high speed and accurately without performing a decoding process. It is an object of the present invention to provide a cut detection method and apparatus that can detect (the flash light is not erroneously detected as a cut and a cut generated between an odd field and an even field can be correctly detected).

[0011]

In order to achieve the above object, the invention according to claim 1 of the present invention is a code for detecting a cut from video data coded by a coding method including a field prediction method. Interlaced video cut detection method for calculating the inter-field similarity for each field based on reference field selection information for selecting whether to reference an odd field or an even field A similarity calculation step,
The method further comprises a calculation step of performing a predetermined calculation corresponding to the inter-field similarity calculated from a plurality of fields, and a comparison step of determining that a cut exists when the calculation result is larger than a predetermined value. It is characterized by

The invention according to claim 2 of the present invention is
In the inter-field similarity calculation step, the difference data is subjected to a predetermined operation, and a value obtained by accumulating the difference data for each reference field is set as the inter-field similarity.

The invention according to claim 3 of the present invention is
A first condition that the inter-field similarity between the field of interest and the past field is small,
It is characterized by including an operation for outputting a large value only when both the second condition that the inter-field similarity between the field of interest and the future field is large are satisfied.

The invention according to claim 4 of the present invention is
The calculation step includes a calculation that outputs a large value when the inter-field similarities between the field of interest and the past and future fields are both small, and the comparison step includes a case where the result of the calculation is larger than a predetermined value. It is characterized by determining that there is a temporary image change.

Further, the invention according to claim 5 of the present invention is a coded interlaced video cut detection apparatus for detecting a cut from video data coded by a coding method including a field prediction method, wherein: Inter-field similarity calculation means for calculating inter-field similarity based on a reference field selection signal; inter-field similarity holding means for holding the calculated inter-field similarity across a plurality of fields; Further, it is characterized in that it comprises: arithmetic means for performing a predetermined arithmetic operation on a plurality of inter-field similarities; and comparing means for comparing the result of the arithmetic operation with a predetermined value and outputting a signal indicating the presence or absence of a cut.

In the invention according to claim 1 of the present invention, the inter-field similarity calculation step calculates a plurality of similarities with each field based on the reference field selection information, and the calculation step is performed. Each time the inter-field similarity for one field is calculated, a predetermined calculation is performed on the inter-field similarity calculated for several fields up to that point. The comparison step determines that a cut exists when the calculation result is larger than a predetermined value. In this way, the cut can be detected at high speed by calculating the inter-field similarity based on the reference field selection information that can be extracted from the encoded video data without performing the decoding process. Further, by calculating the similarity in field units rather than in frame units, it is possible to correctly detect cuts in encoded interlaced video data.

Further, in the invention according to claim 2 of the present invention, after the inter-field similarity calculation step performs a predetermined calculation on the difference data obtained without performing the decoding process from the encoded video data, Cuts can be detected at high speed by using the value obtained by accumulating them for each reference field as the inter-field similarity.

Further, in the invention according to claim 3 of the present invention, the first condition that the calculation step has a small inter-field similarity between the field of interest and the past field (as viewed from the field of interest) and the field of interest. By executing the operation that outputs a large value only when the second condition that the inter-field similarity between the future field and the future field is large is satisfied, a large value is obtained only when the operation result is a cut, and the flash light By taking a small value for noise such as, the cut can be correctly detected even for the encoded interlaced video data.

Further, in the invention according to claim 4 of the present invention, the calculation step executes a calculation for outputting a large value when the inter-field similarities between the field of interest and the past and future fields are both small, When the comparison step determines that there is a temporary image change when the calculation result is larger than a predetermined value, the temporary image change (flash light or the like) can be detected separately from the cut.

Further, in the invention according to claim 5, the inter-field similarity calculation means calculates the inter-field similarity based on the reference field selection signal obtained from the encoded video data without undergoing the decoding process. The inter-field similarity holding means holds the value of the similarity over a plurality of fields. The calculation means performs a predetermined calculation on the plurality of inter-field similarities input from the inter-field similarity holding means, compares the calculation result with a predetermined value by the comparison means, and outputs a signal indicating the presence or absence of the cut. As described above, by calculating the inter-field similarity based on the reference field selection signal obtained directly from the encoded data without performing decoding, hardware for decoding is unnecessary. In addition, by calculating the degree of similarity in units of fields rather than in units of frames, it is possible to correctly detect a cut in encoded interlaced video data.

[0021]

BEST MODE FOR CARRYING OUT THE INVENTION Hereinafter, four embodiments of the present invention will be described with reference to the drawings.

In the following embodiments, MPEG2 (frame structure, field prediction) forward prediction pictures (p
(Picture) will be described as an example of encoded video data.
First, the outline of the MPEG2 encoding mechanism will be described with reference to FIG.

In MPEG2, the frame 51 is 16 × 16
It is divided into macroblocks 52 of pixels and each macroblock is encoded as follows. Either (1) non-intra mode or (2) intra mode is selected depending on whether or not the motion prediction of the macro block can be performed from the previous frame. In the intra mode of (1), closed coding is performed within a frame without using correlation between frames. Specifically, for a 16 × 16 pixel macroblock, in the case of luminance,
Further, it is divided into four 8 × 8 pixel blocks, DCT conversion is performed for each block, and DCT coefficients are output as encoded data. In the non-intra mode (2), a frame structure or a field structure can be selected, but only the frame structure will be described here. In the field structure field prediction method, a macro block of 16 × 16 pixels is divided into an odd field and an even field and divided into two 16 × 8 fields.
The pixel blocks 53 and 54 are formed, and
Motion estimation is performed on a × 8 pixel block, and (A) two motion vectors, (B) two sets of prediction error (difference) DCT coefficients, and (C) two reference field selections for each macroblock. The flag is output as encoded data.
The reference field selection flag in (C) is a flag for enabling selection of whether to perform motion compensation from an even field or motion compensation from an odd field during decoding.

<Embodiment 1> FIG. 1 shows a first embodiment corresponding to the invention of claim 3 including the inventions of claims 1 and 2 of the present invention.
FIG. 6 shows a flow diagram of an example embodiment of FIG. This flow chart shows a procedure for processing one frame of encoded video data (t-th frame) and determining whether or not a cut exists in the previous frame (t-1). The main steps include an inter-field similarity calculation step, a calculation step, and a comparison step.

(A) Inter-field similarity calculation step (12
0) First, an array sim that stores values of inter-field similarity
(T, k, m) (k = 0,1; m = 0,1) are all 0
(Step 101). However, variables k and m
The value 0/1 taken by corresponds to the odd field / even field, respectively. One frame of encoded video data is read (step 102). The coding mode flag (intra / non-intra) is extracted from the coded video data for each macroblock (step 103). Step 11 if the encoding mode flag indicates intra mode
The process is moved to 3 (step 104). In the non-intra mode, the encoded video data is further analyzed, and the reference field selection flag field corresponding to the odd field is
Extract d_sel (= 0 or 1) (step 1
05). If field_sel = 0, it indicates that the motion compensation is performed by referring to the odd field of the previous frame, and if field_sel = 1, it indicates that the even field is referenced. Subsequently, only the DC component of the DCT coefficient representing the prediction error (difference data) is extracted from the encoded video data (step 106). The absolute value of the DC component is compared with the threshold value TH1 (step 107), and only when it is smaller than the threshold value, sim (t, 0, field_se).
1 is added to l) (step 108). The procedure equivalent to the odd field (steps 105 to 108) is executed for the even field (steps 109 to 112). The processing of steps 103 to 112
The process is repeated until the final macroblock is reached (step 113), and the resulting sim (t, k, m) is used as the inter-field similarity (step 114). The characteristics of the inter-field similarity sim (t, k, m) will be described later.

In the present embodiment, the difference data DC
Although the component is extracted (step 106) and the absolute value thereof is thresholded (step 107), the sum of squares of the DCT count of the difference data may be thresholded. Further, it is also possible to omit steps 106 to 108 and omit the processing on the difference data.

(B) Calculation step and comparison step Next, the following calculation is performed on the inter-field similarity sim (t, k, m) to obtain D _frame (t-1) and D _field (t).
Two values of -1) are obtained (step 115).

[0028]

[Equation 1]

[0029]

[Equation 2]

However, MB represents the number of macroblocks included in the frame. The meaning of this expression will be described later.

D _frame (t-1) and D _field (t-1) are compared with two thresholds TH2 and TH3 (step 11).
6) If either of them holds, it is judged that a cut exists in the frame (t-1) (step 117). Otherwise, it is determined that there was no cut (step 11).
8). The process moves to the next frame (step 119).

The calculation performed in the calculation step is expressed by equation (1),
It is not limited to (2). For example,

[0033]

[Equation 3]

[0034]

[Equation 4]

Although the inter-field similarity for two frames is used in this embodiment, the inter-field similarity obtained from more frames may be used.

(C) Property of inter-field similarity degree The property of the inter-field similarity degree sim (t, k, m) will be described with reference to FIG. FIG. 2 shows continuous three frames divided into fields, and a line connecting the three frames corresponds to the inter-field similarity. When the procedure described above is executed for the frame t, the four inter-field similarities sim (t, k, m) (k = 0,1; m
= 0, 1) is obtained. The lower limit of sim (t, k, m) is 0, and the upper limit is the number of macroblocks MB. "Odd field of frame t" 62 and "frame (t-1)"
The field-to-field similarity between the odd fields "61
im (t, 0,0), and takes a large value if the patterns in the fields 61 and 62 are similar. This is because if the pattern of the field 61 is similar to that of the field 62, the number of times the field 62 is referred to by the field 62 during motion compensation increases, and the number of times step 108 is executed increases. Similarly, sim (t, 0,1) is between fields 62 and 63, sim (t, 1,0) is between fields 61 and 64, and sim (t, 1,1) is fields 63 and 64. Represents the similarity between fields.

When there is no cut, that is, when the patterns in the fields 61, 62, 63, 64 of FIG. 2 are all similar, sim (t, k, m) takes a large value. Cut is frame (t-2) and frame (t-1)
, The inter-field similarity indicated by the solid line in FIG. 3 has a large value, while the inter-field similarity indicated by the dotted line has a small value. When there is a cut between the odd field and the even field of frame (t-1), the inter-field similarity indicated by the solid line has a large value and the inter-field similarity indicated by the dotted line is large as shown in FIG. Has a small value. FIG. 5 shows the case where the flash light is reflected in the even field of the frame (t-1). It has a pattern of inter-field similarity different from that of FIG. In the conventional technique, since the feature amount (cumulative value of prediction errors, etc.), which is the amount for evaluating the difference in the pattern between the frames, is used, it is impossible to distinguish the situations of FIG. 4 and FIG. According to the invention, the similarity between fields is calculated instead of between frames, so that FIG. 4 and FIG. 5 can be distinguished.

(D) Role of Calculation Step Next, the equations (1) and (2) calculated in the calculation step will be described. Expression (1) is for detecting the cut in the state shown in FIG. In the situation shown in FIG. 3, sim (t, k, m) / MB has a value close to 1, and sim (t-1, k, m) / MB has a value close to 0. Therefore, D _frame (t
-1) takes a value close to 1, and in other situations (FIGS. 4 and 5) takes a value close to 0. Therefore, by comparing with the threshold value TH2 in step 116, the cut of the situation shown in FIG. 3 can be detected separately from the others. Equation (2) is
It is designed to take a value close to 1 only when the situation shown in FIG. 4 occurs, and take a value close to 0 in other situations (FIGS. 3 and 5). Therefore, in step 116, the threshold T
By comparing with H3, the cut in the situation of FIG. 4 can be detected. In the situation of FIG. 5, D _frame (t-1) and D _field (t-
Since both 1) have a small value, noise such as flash light will not be erroneously detected as a cut.

<Embodiment 2> FIG. 6 shows a flow chart of a second embodiment of the present invention, which corresponds to the invention of claim 4 of the present invention. Since the inter-field similarity calculation step is the same as that of the first embodiment, it will be omitted. Further, of the calculation process and the comparison process, the same parts as those of the first embodiment are denoted by the same reference numerals as those in FIG.

In the present embodiment, the value calculated by the following equation is D when it is determined in step 118 that there is no cut.
_Flash (t-1) is set (step 1001).

[0041]

[Equation 5]

If _Dflash (t-1) is larger than the threshold TH4 (step 1002), it is determined that there is a temporary image change in frame (t-1) (step 10).
03). According to this method, it is possible to detect the frame in which the flash is cooked in news or the like. Further, it is possible to obtain a device that detects an instantaneous (one frame) image inserted for the subliminal effect in a video.

<Embodiment 3> Hereinafter, claim 5 of the present invention
An embodiment of the cut detection device according to the invention will be described with reference to the block diagram of FIG.

The signal separating means 1101 separates and extracts (A) non-intra / intra mode selection signal, (B) difference signal, and (C) reference field selection signal from the input encoded video data. . The comparing means 1102 takes the sum of absolute values of the difference signal (B) and outputs 1 when it is smaller than the threshold value TH1. The signal is switch 11
It is input to 03. The switch 1103 switches according to the non-intra / intra mode selection signal (A),
In the non-intra mode, the signal from the comparing means 1102 is sent to the switch 1104, and in the intra mode, 0 (ground) is sent to the switch 1104. The switch 1104 is adapted to switch in accordance with the reference field selection signal (C), and if the reference field selection signal indicates that it refers to an odd field, then 11
A signal is input to the counting means 05, and a signal is input to the counting means 1106 if it indicates that the even field is referred to. The counting means 1105 and 1106 count the number of inputs of 1 as a signal. When the counting for one field is completed, the count values of 1105 and 1106 are output as the inter-field similarity, respectively. Figure 7
A portion 1113 surrounded by a dotted line is a portion corresponding to the inter-field similarity calculation means in the invention of claim 5. 1
The inter-field similarity holding means 107 holds N fields, that is, 2N inter-field similarities (overwrites a new value in the area in which the oldest inter-field similarity is stored). The calculation unit 1108 executes the calculation of Expression (1) on the inter-field similarity input from the inter-field similarity holding unit. The calculation unit 1110 executes the calculation of Expression (2). The calculation result is 11
09 and 1111 are respectively inputted to the comparing means, and the threshold value (T
H2 and TH3), and if they have a value larger than the threshold value, the comparison means 1109, 1111 outputs a signal. The OR circuit 1112 takes the OR of the two signals and outputs it as a cut detection signal.

Another embodiment of the inter-field similarity calculation means will be described with reference to FIG. The switch 1301 switches according to the intra / non-intra mode selection signal and outputs 0 in the intra mode and 1 in the non-intra mode. The switch 1302 switches according to the reference field selection signal, and sends a signal to the counting means 1303 when the reference field is an odd field and a signal to the counting means 1304 when the reference field is an even field. The counting means of 1303 and 1304 are
Each counts the number of macroblocks that referred to the “odd field” and the number of macroblocks that referred to the “even field”. When the counting for one field is completed, the count value holds the inter-field similarity holding means (1107).
Is output as the inter-field similarity.

Still another embodiment of the inter-field similarity calculation means will be described with reference to FIG. This structure corresponds to the frame structure of MPEG2. M
When the encoded video data having the frame structure of PEG2 is viewed in order, the encoded data of the odd field and the even field appear alternately in the frame. FIG. 9 shows a configuration suitable for handling such encoded video data. The signal separating means outputs a field selection signal indicating which field the data currently being processed is for decoding, and the switch 1104 is used by using the signal.
The two switches 1401 and 1402 connected to the two outputs are switched. The four counting units 1403 to 1406 connected to the respective outputs of this switch output the count value as the inter-field similarity degree each time the counting for one frame is completed. Two new switches 1401 and 1402
By providing the four inter-field similarities (sim (t, k, m) (k = 0,
1; m = 0, 1)) is obtained at the same time.

<Embodiment 4> FIG. 10 shows a fourth embodiment of the present invention.
2 shows a block diagram of an example embodiment of FIG. This device realizes a cut detection device with a general-purpose computer and software. In this block diagram, a CPU 1201 performs overall control, and data transmission / reception is performed via a bus 1208. A hard disk 1202 stores encoded video data, and the inter-field similarity holding unit 1205 is realized as a memory in the computer. Further, each of the signal separating unit 1203, the inter-field similarity calculating unit 1204, the calculating unit 1206, and the comparing unit 1207 is realized as a program stored in a computer memory and executed by the CPU 1201. Each means executes the corresponding procedure shown in the first embodiment as a computer program.

The present invention is not limited to the above embodiment. The present invention can be implemented not as an apparatus and method for detecting a cut from encoded data but as an encoding apparatus or a part of an encoding apparatus. For example, a configuration is possible in which a differential signal, a reference field selection signal, an intra / non-intra selection signal, and the like, which are obtained at the time of motion prediction in the encoding device, are input to the cut detection device described above. In that case, the above-mentioned signal separation means becomes unnecessary.

[0049]

According to the invention described in claim 1 of the present invention, based on the reference field selection information which can be extracted from the video data encoded by the encoding method including the field prediction method without performing the decoding process. Since the inter-field similarity is calculated, a predetermined operation is performed on the inter-field similarity calculated from a plurality of fields, and it is determined that a cut exists when the operation result is larger than a predetermined value, A time-consuming decoding process is unnecessary and a high-speed cut detection method can be obtained. Also, since the cuts are detected based on the similarity between the fields, not between the frames, the cuts between the odd field and the even field can be accurately detected, and the flash light is not detected as a cut by mistake. ) The effect that the cut can be detected is obtained.

According to the invention of claim 2 of the present invention,
In the inter-field similarity calculation step, the difference data obtained without undergoing the decoding process is subjected to a processing operation, and then the value obtained by accumulating it for each reference field is taken as the inter-field similarity degree, so that it is possible to cut at high speed. Is obtained.

According to the invention of claim 3 of the present invention,
Only when the calculation process satisfies both the first condition that the inter-field similarity between the field of interest and the past field is small and the second condition that the inter-field similarity between the field of interest and the future field is large Since the calculation process is designed to output a large value, it is possible to obtain the effect that the cut can be correctly detected even in the encoded interlaced video.

According to the invention of claim 4 of the present invention,
Since the calculation process is designed to output a large value when the field similarity between the field of interest and the past and future fields is small, the effect that the temporary image change can be detected separately from the cut is obtained. To be

According to the invention of claim 5 of the present invention,
Since the inter-field similarity is calculated based on the reference field selection signal directly obtained from the encoded data without performing the decoding process, hardware for decoding becomes unnecessary. Further, since the cuts are detected by using the similarity between the fields instead of between the frames, it is possible to obtain the effect that the cuts can be correctly detected even in the encoded interlaced video.

[Brief description of drawings]

FIG. 1 is a flowchart for explaining a first embodiment example of the present invention.

FIG. 2 is an explanatory diagram of inter-field similarity.

FIG. 3 is an explanatory diagram of inter-field similarity (in the case of a cut that switches between frames).

FIG. 4 is an explanatory diagram of inter-field similarity (in the case of cutting between an odd field and an even field).

FIG. 5 is an explanatory diagram of inter-field similarity (when flash light is reflected)

FIG. 6 is a flowchart for explaining a second embodiment example of the present invention.

FIG. 7 is a block diagram for explaining a third exemplary embodiment of the present invention.

FIG. 8 is a block diagram of another embodiment of inter-field similarity calculation means.

FIG. 9 is a block diagram of yet another embodiment of inter-field similarity calculation means.

FIG. 10 is a block diagram for explaining a fourth exemplary embodiment of the present invention.

FIG. 11 is a diagram for explaining the problems of the conventional technology (in the case of a cut that switches between frames).

FIG. 12 is a diagram for explaining a problem of the conventional technique (in the case of cutting between an odd field and an even field).

FIG. 13 is a diagram for explaining the problems of the prior art (when there is a glare of flash light).

FIG. 14 is a diagram for explaining the mechanism of the MPEG2 encoding system.

[Explanation of symbols]

1101 ... Signal separation means 1102 ... Comparison means 1103, 1104 ... Switch 1105, 1106 ... Counting means 1107 ... Inter-field similarity holding means 1108, 1110 ... Computing means 1109, 1111 ... Comparison means 1112 ... OR circuit 1113 ... Inter-field similarity calculation means 1201 ... CPU 1202 ... magnetic disk 1203 ... Signal separation means 1204 ... Inter-field similarity calculation means 1205 ... Inter-field similarity holding means 1206 ... Computing means 1207 ... Comparison means 1301, 1302 ... Switch 1303, 1304 ... Counting means 1401, 1402 ... switch 1403, 1404, 1405, 1406 ... Counting means

─────────────────────────────────────────────────── ─── Continuation of the front page (56) Reference JP-A-7-288840 (JP, A) JP-A-6-22304 (JP, A) (58) Fields investigated (Int.Cl. ⁷ , DB name) H04N 5/76-5/956 H04N 5/262-5/278 H04N 7/24-7/68

Claims (5)

(57) [Claims] 1. A coded interlaced video cut detection method for detecting a cut from video data coded by a coding method including a field prediction method, which refers to either an odd field or an even field. Based on the reference field selection information for selecting whether or not the field-to-field similarity calculation step is performed for each field, and a predetermined value corresponding to the field-to-field similarity calculated from a plurality of fields. A coded interlaced video cut detection method comprising: a calculation step of performing a calculation; and a comparison step of determining that a cut is present when a result of the calculation is larger than a predetermined value. 2. The coded interlaced video cut detection method according to claim 1, wherein the inter-field similarity calculation step performs a predetermined operation on the difference data, and then sums the accumulated values for each reference field. A method for detecting coded interlaced video cuts, characterized in that the degree of similarity is between fields. 3. The coded interlaced video cut detection method according to claim 1, wherein the calculation step has a first condition that the inter-field similarity between the field of interest and the past field is small. A coded interlaced video cut detection method characterized by including an operation of outputting a large value only when the second condition that the inter-field similarity between the field of interest and the future field is large is both satisfied. 4. Claim 1 or claim 2 or claim 3.
In the coded interlaced video cut detection method described in, the calculation step includes a calculation that outputs a large value when the inter-field similarities between the field of interest and the past and future fields are both small, and the comparison step includes A coded interlaced video cut detection method, characterized in that it is determined that there is a temporary image change when the calculation result is larger than a predetermined value. 5. A coded interlaced video cut detection device for detecting a cut from video data coded by a coding method including a field prediction method, the inter-field similarity based on the reference field selection signal. Inter-field similarity calculation means for calculating, the inter-field similarity retention means for retaining the calculated inter-field similarity across a plurality of fields, and a predetermined value for the retained inter-field similarity. An encoded interlaced video cut detection device comprising: a calculation unit that performs a calculation; and a comparison unit that compares the calculation result with a predetermined value and outputs a signal indicating whether or not there is a cut.