JP5016556B2 - Scene change detection apparatus, encoding apparatus, and scene change detection method - Google Patents

Description

  The present invention relates to a scene change detection device, an encoding device, and a scene change detection method, and more particularly, to a scene change detection device, an encoding device, and a scene change detection method for detecting a scene change based on a change point of an input sound and an input image.

  In a video recorder that records television broadcasts and the like, it is possible to detect a commercial (CM) portion from recorded information and skip or delete the commercial portion. Conventionally, the detection of a CM section during TV broadcasting has been performed by a combination of audio feature information such as switching of audio signals (for example, switching between stereo and monaural) and video feature information such as changes in video brightness. In addition, the detection of the CM section may use a time feature such as time information. When time feature information is used, the time feature information defines, for example, one unit as 15 seconds and one CM section as an integer multiple of unit time. Various proposals have been made for commercial part detection devices and detection methods in such devices. An example of such an apparatus is disclosed in Patent Documents 1 and 2 (hereinafter referred to as Conventional Examples 1 and 2).

  In Conventional Example 1, a scene change point is detected from an input image and input sound, and a signal section is calculated from the detected scene change point. And it has the specific area judgment means which determines whether a signal area corresponds with the specific characteristic of CM. Here, the specific feature of the CM is the above-described voice feature information, video feature information, or time feature information (hereinafter, these generic names are referred to as feature information).

On the other hand, in the conventional example 2, the first time difference information is obtained based on the detection unit that detects the edge black frame information and the fade black frame information based on the image information, and two adjacent information among the edge black frame information or the fade black frame information. Time difference means for determining and test means for determining whether the first time difference information is included in the first preset time difference information. That is, in Conventional Example 2, when the first time difference information is included in the first preset time difference information, the section is detected as a commercial section. Note that the first preset time difference information is set information of a setting within a range of 0.5 seconds before and after each of a plurality of setting values such as 10 seconds, 20 seconds, and 30 seconds.
JP 2004-147204 A Special table 2003-534757 gazette

  However, it is similar to the case where CM unit time is not specified (for example, broadcasting in Europe and the United States), when there is no change in stereo, monaural, etc., or when switching between CM section and program section during a program section in movie broadcasting etc. There are cases where there is a video / audio feature, and in the case of such a broadcast form, there is a problem that it is difficult to detect the CM section with high accuracy in the conventional examples 1 and 2. That is, in the conventional examples 1 and 2, only those that match the signal interval defined in advance are detected as the CM portion. Therefore, when trying to record a television broadcast having a CM period with an indefinite time (for example, when each CM time is random), the CM portion may not be detected.

  According to one aspect of the present invention, a sound information determination unit that detects a soundless state of a sound signal and outputs silence detection time information, and detects a change in luminance or a decrease in luminance level of the image in the image signal. Alternatively, an image information determination unit that stores scene change time information indicating a time when a decrease in luminance level is detected, and a scene change that outputs a first scene change candidate time based on the silence detection time information and the scene change time information Validity of the first scene change candidate time based on the candidate point detection unit, the amount of change in the audio level of the audio signal at a time later than the first scene change candidate time, and a preset minimum determination value A scene change candidate point determination unit that outputs the first scene change candidate time determined to be effective and the second scene change candidate time as the second scene change candidate time; An output judging unit that outputs a scene change detection information based on the time difference between the detected maximum value of the time that the front and rear of the change candidate time is a scene change detection device including a.

  According to another aspect of the present invention, there is provided a scene change detection method in a scene change detection apparatus to which a video composed of an audio signal and an image signal is input. The silence change time information is detected by detecting a silence state of the audio signal. Generating a scene change time information indicating a time at which the brightness change or the decrease in the brightness level is detected, detecting a brightness change or a decrease in the brightness level of the image in the image signal, and generating the silence detection time information and the A first scene change candidate time is generated based on the scene change time information, a change amount of the audio level of the audio signal at a time later than the first scene change candidate time, and a preset determination minimum value, The first scene change candidate time is determined based on the first scene change candidate time, the first scene change candidate time determined to be valid is output as a second scene change candidate time, and the second scene change candidate time is output. And it outputs a scene change detection information when time difference between successive first scene change candidate time is below detection maximum time.

  The scene change detection device, the scene change detection method, and the encoding device including these according to the present invention, when detecting a scene change, are based on the silence change time and the scene change time information indicating that the luminance of the image indicates the change time. The candidate scene change time is output. In the present invention, the effectiveness of the first scene change candidate time is determined based on the amount of change in the audio level of the audio signal after the first scene change candidate time. That is, it is determined whether or not the first scene change candidate time is between the program section and the CM section. Then, the effective first scene change candidate time is set as the second scene change candidate time, and the scene change is detected from the time difference between the second scene change candidate time at the current time and the second scene change candidate time at the previous time. Output information. That is, the scene change detection device, the scene change detection method, and the encoding device including these according to the present invention can switch between programs whose section length is indefinite based on the second scene change candidate time detected with high accuracy. Even if this occurs, it is possible to detect each section with high accuracy without having a plurality of settings.

  Further, another aspect of the present invention detects a silence state of an audio signal, outputs an audio information determination unit that outputs silence detection time information, and detects a change in luminance or a decrease in luminance level of the image in the image signal. When the scene change time information is included in an image information determination unit that stores scene change time information indicating a time at which the change in luminance or a decrease in luminance level is detected, and a silence detection period set based on the silence detection time information In addition, the audio signal and the image signal are included in response to detection of a change amount of the audio level of the audio signal that is equal to or greater than a predetermined minimum determination value after the time indicated by the scene change time information. A scene change detection apparatus including a section determination unit that outputs scene change detection information indicating a change point of a video signal.

  According to the scene change detection device of the present invention, when the scene change time information is included in the silence detection period set based on the silence detection time information, it is set in advance after the time indicated by the scene change time information. Scene change detection information is output in response to detection of a change in the sound level of the sound signal that is equal to or greater than the minimum determination value. That is, in the generation of the scene change detection information, it is possible to discriminate between the program section and the CM section with higher accuracy by making a determination based on the amount of change in the audio level that is a feature of the section switching. Therefore, the scene change detection information output by the scene change detection apparatus of the present invention can indicate the time that becomes the boundary between the program section and the CM section with higher accuracy.

  According to the scene change detection device and the scene change detection method of the present invention, it is possible to accurately detect a section to be detected even when the length of the section to be detected is indefinite.

Embodiment 1
Embodiments of the present invention will be described below with reference to the drawings. FIG. 1 shows a block diagram of an encoding apparatus 1 according to the first embodiment. As shown in FIG. 1, an encoding apparatus 1 includes an analog / digital converter (A / D converter) 10, a delay circuit (Time Base Corrector Unit: TBC unit) 20, an encoder 30, a storage device 40, and an application execution unit. (CPU) 50 and a scene change detection device 60 are provided.

  The A / D converter 10 converts an analog image signal and an analog audio signal into a digital image signal and a digital audio signal, and outputs them. In the following description, the digital image signal and the digital audio signal are simply referred to as an image signal and an audio signal. The TBC unit 20 is one of delay circuits, and for example, by adjusting the delay amount of the image signal output from the A / D converter 10, the time fluctuation of the image signal is reduced. The encoder 30 encodes the image signal output from the TBC unit 20 and the audio signal output from the A / D converter 10 according to a predetermined format (for example, MPEG2 (Moving Picture Experts Group 2)). The storage device 40 stores data encoded by the encoder.

  The application execution unit 50 executes application software such as video recording software, and uses, for example, a central processing unit (CPU). The application execution unit 50 receives scene change detection information from a scene change detection device 60 to be described later, generates an information file describing a CM (Commercial Message) section, for example, and stores it in the storage device 40 together with the recorded data.

  The scene change detection device 60 outputs scene change detection information based on the image signal output from the TBC unit 20 and the audio signal output from the A / D converter 10. The scene change detection device 60 includes an audio information determination unit 70, an image information determination unit 80, and a section determination unit 90. The voice information determination unit 70 analyzes the voice level of the voice signal and outputs silence detection time information A. The audio information determination unit 70 has a silent period determination unit 71. The silence period determination unit 71 uses a time when the sound level of the sound signal is equal to or lower than a preset silence determination level as a start time, and the sound level is set to a threshold value during a silence determination period set in advance from the start time. If not, the end time of the silence determination period is output as silence detection time information A.

  The image information determination unit 80 detects a change in luminance or a decrease in luminance level in the image signal output from the TBC unit 20, and stores the time when the change in luminance or the decrease in luminance level is detected as scene change time information. The image information determination unit 80 includes a scene change detection unit 81, a black image detection unit 82, an OR circuit 83, and a scene change time information storage unit 84.

  The scene change detection unit 81 generates a scene change at the current time when the difference between the luminance level of the image signal at the previous time and the luminance level of the image signal at the current time is greater than or equal to a preset luminance change width threshold. The current time is output as the scene change detection result B. When the absolute value of the luminance level of the image signal at the current time is equal to or less than a preset luminance level threshold, the black image detection unit 82 determines that a change to the black image has occurred at the current time, Is output as a black image detection result C.

  The OR circuit 83 outputs the logical sum of the scene change detection result B and the black image detection result C as the scene change time. That is, when the scene change detection result B and the black image detection result C indicate the same time, the OR circuit 83 sets the time as the scene change time, and the scene change detection result B and the black image detection result C indicate different times. In each case, each is output as a scene change time. The scene change time information storage unit 84 stores the scene change time output from the OR circuit 83 as scene change time information D.

  The section determination unit 90 outputs scene change detection information based on the silence detection time information A output from the audio information determination unit 70 and the scene change time information D stored in the image information determination unit 80. The section determination unit 90 includes a scene change candidate point detection unit 91, a scene change candidate point determination unit 92, an output determination unit 93, a determination minimum value storage unit 94, and a detection maximum value storage unit 95. The scene change candidate point detection unit 91 sets a predetermined period including the silence detection time information A as the silence detection period. If the scene change time extracted from the scene change time information D is included in the silence detection period, the scene change candidate point detection unit 91 outputs the scene change time as the scene change candidate time E. The scene change candidate point detection unit 91 outputs the scene change time at the earliest time as the scene change candidate time E when a plurality of scene change times are included in the silence detection period.

  The scene change candidate point determination unit 92 is stored in the audio level change amount (in this embodiment, the audio level increase ΔAUDIO) that occurs immediately after the scene change candidate time E, and the determination minimum value storage unit 94. Is compared with the determination minimum value A2_th. More specifically, the scene change candidate point determination unit 92 sets a predetermined period (for example, the candidate point determination period Δt) after the scene change candidate time E, and is equal to or greater than the determination minimum value A2_th during the candidate point determination period Δt. Is detected, the scene change candidate time E at that time is determined to be valid. Then, the scene change candidate point determination unit 92 outputs the scene change candidate time E determined to be valid as the valid scene change candidate time F. On the other hand, the scene change candidate point determination unit 92 determines that the scene change candidate time E at that time is invalid when the increase amount ΔAUDIO of the audio level equal to or greater than the determination minimum value A2_th is not detected during the candidate point determination period Δt. To do. Then, the scene change candidate point determination unit 92 discards the scene change candidate time E determined to be invalid. The determination minimum value storage unit 94 stores a minimum determination value. This minimum determination value may be transmitted from the application execution unit 50 or may be a preset value.

  The output determination unit 93 calculates a detection interval based on the difference between the effective scene change candidate times F and the preceding and following times, and calculates the detection interval time and the detection maximum value CMmax stored in the detection maximum value storage unit 95. The scene change detection information is output after comparison. Details of the scene change detection information determination process will be described later. The maximum detection value storage unit 93 stores the maximum detection value CMmax. This detected maximum value CMmax may be transmitted from the application execution unit 50 or may be a preset value. The detected maximum value CMmax is defined as, for example, the maximum time of one CM. In the following description, a section shorter than the detected maximum value is a CM section, and a long section is a main section.

  Here, the operation of the scene change candidate point detection unit 91 will be described in detail. A flowchart including the operation of the scene change candidate point detection unit 91 is shown in FIG. In FIG. 2, the scene change candidate point detection unit 91 operates based on the operation of the audio information determination unit 70, and then the scene change candidate point detection unit 91 further operates based on the operation of the image information determination unit 80. .

  According to the flowchart shown in FIG. 2, when the detection of the scene change candidate time is started, as step S1, the audio information determination unit 70 first analyzes the audio level of the audio signal to obtain the silence detection time information A. (Step S1). Subsequently, the scene change candidate point detection unit 91 adds the pre-detection period value before the time indicated by the silence detection time information A, and further adds the post-detection period value after this time. The pre-detection period value and the post-detection period The total value with the value is set as a silence detection period (step S2).

  In step S3, it is determined whether or not there is a scene change time included in the silence detection period with reference to the scene change time information D stored in the image information determination unit 80. In this step S3, when there is no scene change time included in the silence detection period, the scene change candidate time is not output. On the other hand, if there is a scene change time included in the silence detection period in step S3, the scene change candidate point detection unit 91 acquires information on the scene change time as the nth scene change candidate time Ctime (n) (step S3). S4). Note that n is an integer and indicates the order of the scene change candidate points detected in the scene change detection operation of the scene change detection apparatus according to the present embodiment. The same applies to n in the following description.

  Next, in step S5, the scene change candidate point determination unit 92 refers to the determination minimum value A2_th stored in the determination minimum value storage unit 94, and determines whether or not Ctime (n) is valid. Specifically, it is determined whether or not the audio level increase amount ΔAUDIO generated in the candidate point determination period Δt (for example, a time within 500 msec) immediately after Ctime (n) exceeds the determination minimum value A2_th. In step S5, when the audio level increase amount ΔAUDIO is less than the minimum determination value A2_th, Ctime (n) is discarded as an erroneously detected scene change candidate time (step S6). Then, if there is no candidate scene change time included in the silence detection period in step S3, and if the candidate scene change time is invalidated in step S6, the process ends with no valid scene change candidate time. If the audio level increase amount ΔAUDIO exceeds the minimum determination value A2_th in step S5, the detected scene change candidate time is output as the effective scene change candidate time F (step S7), and the process ends.

  Here, a timing chart of the operations of the scene change candidate point detection unit 91 and the scene change candidate point determination unit 92 is shown in FIG. 3, and the operations of the scene change candidate point detection unit 91 and the scene change candidate point determination unit 92 are described in more detail. explain. First, when the sound level of the sound signal input to the sound information determination unit 70 decreases and falls below the silence determination level A1_th, the sound information determination unit 70 starts silence determination using that time as the start time. If the voice level does not exceed the silence determination level A1_th during the silence determination period t1 set in advance from the silence determination disclosure time, the voice information determination unit 70 sets the time after the silence determination period has elapsed. The silence detection time is determined, and the time is output as silence detection time information A.

  Subsequently, the scene change candidate point detection unit 91 adds the pre-detection period value before the time indicated by the silence detection time information A as the predetermined period, and further adds the post-detection period value after this time. And the post-detection period value are defined as a silence detection period. Even when the sound level of the sound signal exceeds the silence determination level A1_th during the silence detection period, the silence detection period is uniquely set including that section. Then, when there is a scene change time included in the silence detection period, the scene change candidate point detection unit 91 outputs the earliest of the scene change times as the scene change candidate time Ctime (n).

  Then, the scene change candidate point determination unit 92, when the audio level increase ΔAUDIO exceeds the determination minimum value A2_th during the candidate point determination period Δt after the scene change candidate time C_Time (n), The scene change candidate time C_Time (n) at that time is determined to be valid, and is output as the effective scene change candidate time Ctime (n). On the other hand, if the audio level increase amount ΔAUDIO of the audio level does not exceed the determination minimum value during the candidate point determination period Δt after the scene change candidate time C_Time (n), the scene change candidate time C_Time ( Determine n) as invalid and discard. Therefore, the effective scene change candidate time F is not output.

  Next, the operation of the output determination unit 93 will be described in detail. A flowchart showing the operation of the output determination unit 93 is shown in FIG. As illustrated in FIG. 4, the output determination unit 93 firstly includes a scene change candidate time Ctime (n−2) two hours before the scene change candidate time and a scene change candidate time Ctime (n−1) one time before. The second detection interval Ltime (n−1) is calculated from the difference between the two (step S10). Subsequently, the output determination unit 93 determines the first detection interval from the difference between the scene change candidate time Ctime (n−1) one hour before the scene change candidate time and the current scene change candidate time Ctime (n). Ltime (n) is calculated (step S11).

  Then, the time of the first detection interval Ltime (n) is compared with the value of the maximum detection value CMmax (step S12). When it is determined in step S12 that the time of the first detection interval Ltime (n) is equal to or less than the value of the detection maximum value CMmax, the time of the second detection interval Ltime (n−1) and the detection maximum value CMmax Are compared (step S13). If the second detection interval Ltime (n−1) is larger than the maximum detection value CMmax in step S13, the scene change candidate time Ctime (n−1) is output as the CM start time (step S14). If the second detection interval Ltime (n−1) is equal to or less than the maximum detection value CMmax in step S13, the process ends without outputting the scene change detection information.

  On the other hand, when it is determined in step S12 that the time of the first detection interval Ltime (n) is larger than the value of the detection maximum value CMmax, the time of the second detection interval Ltime (n-1) and the detection maximum The value CMmax is compared (step S15). If the second detection interval Ltime (n−1) is equal to or smaller than the maximum detection value CMmax in step S15, the scene change candidate time Ctime (n−1) is output as the CM end time (step S16). . If the second detection interval Ltime (n−1) is larger than the maximum detection value CMmax in step S15, the process ends without outputting the scene change detection information.

  FIGS. 5 to 8 are diagrams schematically showing scene change detection information detected in the scene change detection apparatus. The example shown in FIG. 5 shows a case where the CM start time is output in step S14. In this case, the effective scene change candidate time F is detected based on the silence detection period, the scene change detection result B, the black image detection result C, and the determination result of the sound level increase amount. The first detection interval Ltime (n) obtained from the three scene change candidate times is equal to or less than the detection maximum value CMmax, and the second detection interval Ltime (n−1) is larger than the detection maximum value CMmax. Accordingly, the scene change candidate time Ctime (n−1) one hour before is detected as the CM start time. In the example shown in FIG. 5, even when the scene change detection result B is detected during the silence detection period, the case where the audio level increase amount ΔAUDIO does not exceed the determination minimum value A2_th at that time is also shown. Yes. In such a case, the sea change candidate time E at that time is discarded by the determination in step S6, and the effective scene change candidate time F is not output.

  The example shown in FIG. 6 shows a case where the CM end time is output in step S16. In this case, the scene change candidate time E is detected based on the silence detection period, the scene change detection result B, the black image detection result C, and the determination result of the sound level increase amount. The first detection interval Ltime (n) obtained from the three scene change candidate times is larger than the detection maximum value CMmax, and the second detection interval Ltime (n−1) is less than or equal to the detection maximum value CMmax. Therefore, the scene change candidate time Ctime (n−1) one hour before is detected as the CM end time. In the example shown in FIG. 6, the scene change detection result B is not detected during the silence detection period. In such a case, the audio level increase at that time. Even if ΔAUDIO exceeds the minimum determination value A2_th, the sea change candidate time E is not output from the scene change candidate point detection unit 91, and therefore the valid scene change candidate time F is not output.

  The example shown in FIG. 7 is a case where it is determined in step S13 that the length of the second detection section is smaller than the maximum detection value. In this case, the scene change candidate time E is detected based on the silence detection period, the scene change detection result B, and the black image detection result C. The first detection interval Ltime (n) obtained from the three scene change candidate times E is equal to or less than the detection maximum value CMmax, and the second detection interval Ltime (n−1) is equal to or less than the detection maximum value CMmax. Therefore, the scene change candidate time Ctime (n−1) one hour before is not detected as the CM end time. Then, the scene change candidate time Ctime (n + 1) one time later is detected, and the value of the third detection section obtained from the difference between the scene change candidate time Ctime (n + 1) and the scene change candidate time Ctime (n) is detected maximum. When the value is larger than the value CMmax, the scene change candidate time Ctime (n) before the current time is detected as the CM end time.

  The example shown in FIG. 8 is a case where it is determined in step S15 that the length of the second detection section is larger than the maximum detection value CMmax. In this case, the scene change candidate time E is detected based on the silence detection period, the scene change detection result B, and the black image detection result C. The first detection interval Ltime (n) obtained from the three scene change candidate times E is larger than the detection maximum value CMmax, and the second detection interval Ltime (n−1) is larger than the detection maximum value CMmax. Accordingly, the scene change candidate time Ctime (n−1) one hour before is not detected as the CM start time. Then, the scene change candidate time Ctime (n + 1) one time later is detected, and the value of the third detection section obtained from the difference between the scene change candidate time Ctime (n + 1) and the scene change candidate time Ctime (n) is detected maximum. If the value is less than or equal to the value CMmax, the scene change candidate time Ctime (n) before the current time is detected as the CM start time.

  From the above description, the scene change detection device 60 according to the present embodiment calculates the scene change candidate time based on the silence detection period of the audio signal, the change point of the image signal, and the change amount of the audio signal, and a plurality of scene change candidates. The detection interval is calculated from the time. At this time, in the scene change detection device 60 according to the present embodiment, in consideration of the amount of change in the audio level during the silence detection period and the sound period, the detected scene is not greater than the determination minimum value A2_th. Discard the change candidate time. As described above, the scene change detection device 60 according to the present embodiment pays attention to the fact that the difference in the audio level between the CM section and the program section is larger than the difference in the audio level in each section. Time detection accuracy can be improved. Then, based on the scene change candidate time detected with high precision, a detection section indicating a section between the start and end of the program or the start and end of the CM is detected with high precision. Further, by comparing the detection interval and the maximum detection value, it is possible to prevent erroneous detection of the interval to be detected based on one set value even when the length of the detection interval is indefinite. That is, the scene change detection device 60 according to the present embodiment can detect the boundary between the program section and the CM section with high accuracy based on the detection section detected with high accuracy and the maximum detected value.

  Further, the scene change detection device 60 according to the present embodiment compares each detection interval adjacent to the detection maximum value among the detection intervals, and determines a scene change candidate time to be detected based on the result. The detection accuracy can be improved as compared with the case where the determination is made with only one detection section. For example, as shown in FIG. 8, it is possible to perform processing that does not detect the scene change candidate time Ctime (n−1) detected in the main section. Further, by performing determination using such a plurality of detection sections, for example, even when CM sections are continuous, it is possible to determine the start time and end time of the continuous CM sections. For example, as shown in FIG. 7, when a plurality of CM sections are continuous, the start time of the earliest CM section and the end time of the last CM section of the consecutive CM sections are detected, and the consecutive CM sections are detected. It is possible to perform processing without detecting the time between. By performing such processing, the scene change detection apparatus according to the present embodiment can reduce the number of scene change detection information to be detected.

  Furthermore, the scene change detection device 60 according to the present embodiment sets the silence detection period having a predetermined width with respect to the silence detection time determined by the audio information determination unit 70 as silence, thereby determining the scene change candidate time. Prevent false detection. For example, in the encoding apparatus 1 shown in FIG. 1, there is a delay due to the TBC unit 20, which causes a deviation between the audio signal and the image signal as shown in FIG. 9. In the example shown in FIG. 9, the sound level of the sound signal must be lowered in synchronization with the black image of the image signal. The case where the gap has occurred in the period is shown. Even in such a case, the scene change detection device 60 according to the present embodiment uses this deviation as a pre-detection set value in order to set a silence detection period having a predetermined width with respect to the silence detection time. It is possible to accurately detect the scene change candidate time by absorbing the set value after detection.

  Note that the present invention is not limited to the above-described embodiment, and can be changed as appropriate without departing from the spirit of the present invention. For example, the maximum detection value is not limited to the maximum value in the CM section, but may be any value as long as the maximum value in the section to be detected is set.

1 is a block diagram of an encoding device including a scene change detection device according to a first exemplary embodiment. It is a figure which shows the flowchart of the detection process of the scene change candidate time in the scene change detection apparatus concerning Embodiment 1. FIG. 6 is a timing chart of a process for detecting a scene change candidate time in the scene change detection apparatus according to the first embodiment; It is a figure which shows the flowchart of the output determination process in the scene change detection apparatus concerning Embodiment 1. FIG. It is the figure which showed typically the scene change detection information detected in the scene detection apparatus concerning Embodiment 1. FIG. It is the figure which showed typically the scene change detection information detected in the scene detection apparatus concerning Embodiment 1. FIG. It is the figure which showed typically the scene change detection information detected in the scene detection apparatus concerning Embodiment 1. FIG. It is the figure which showed typically the scene change detection information detected in the scene detection apparatus concerning Embodiment 1. FIG. 4 is a timing chart illustrating an example of preventing erroneous detection due to a difference between an audio signal and an image signal in the scene detection device according to the first exemplary embodiment;

Explanation of symbols

1 Encoder 10 A / D Converter 20 TBC Unit 30 Encoder 40 Storage Device 50 Application Execution Unit 60 Scene Change Detection Device 70 Audio Information Determination Unit 71 Silent Period Determination Unit 80 Image Information Determination Unit 81 Scene Change Detection Unit 82 Black Image Detection unit 83 OR circuit 84 Scene change time information storage unit 90 Section determination unit 91 Scene change candidate point detection unit 92 Scene change candidate point determination unit 93 Output determination unit 94 Determination minimum value storage unit 95 Detection maximum value storage unit A Silence detection time Information B Scene change detection result C Black image detection result D Scene change time information E Scene change candidate time F Effective scene change candidate time CMmax Detection maximum value Ctime Scene change candidate time Ltime Detection section

Claims (8)

A voice information determination unit that detects a silence state of the voice signal and outputs silence detection time information;
An image information determination unit that detects a change in luminance or a decrease in luminance level of an image in an image signal and stores scene change time information indicating a time when the luminance change or the decrease in luminance level is detected;
A scene change candidate point detector that outputs a first scene change candidate time based on the silence detection time information and the scene change time information;
The effectiveness of the first scene change candidate time is determined based on the amount of change in the audio level of the audio signal at a time later than the first scene change candidate time and a preset minimum determination value, and is effective A scene change candidate point determination unit that outputs the first scene change candidate time determined as the second scene change candidate time;
An output determination unit that outputs scene change detection information based on the time difference between the preceding and succeeding times of the second scene change candidate time and the detected maximum value ;
The output determination unit
Calculating a first detection interval based on a time difference between a current time of the second scene change candidate times and a time one hour before the current time of the second scene change candidate times;
A second detection interval based on a time difference between a time one hour before the current time of the second scene change candidate time and a time two times before the current time of the second scene change candidate time. To calculate
When the first detection interval is equal to or less than the maximum detection value and the second detection interval is greater than the maximum detection value, the second scene change candidate time one time before is determined as the scene change detection information. Output as the start time,
When the first detection interval is larger than the detection maximum value and the second detection interval is less than or equal to the detection maximum value, the second scene change candidate time one time before is determined as the scene change detection information. A scene change detection device that outputs the end time .   The scene change candidate point determination unit is preset with a candidate point determination period for setting a predetermined period after the first scene change candidate time, and calculates the amount of change in the audio level within the candidate point determination period. The scene change detection apparatus according to claim 1, which is used for determining the effectiveness.   The scene change detection device according to claim 2, wherein the scene change candidate point determination unit discards the first scene change candidate time when the amount of change in the audio level is smaller than the determination minimum value.   The scene change candidate point detection unit sets a predetermined period including the silence detection time information as a silence detection period, and the first scene change candidate when the scene change time information is included in the silence detection period The scene change detection apparatus according to claim 1, which outputs time.   The scene change candidate point detection unit outputs the scene change time information indicating the earliest time as the first scene change candidate time when a plurality of the scene change time information is included in the silence detection period. Item 5. The scene change detection device according to Item 4. The image information determination unit includes a scene change detection unit that detects a luminance change of an image in the image signal, a black image detection unit that detects a decrease in luminance level in the image signal, an output of the scene change detection unit, and the scene change detecting device according to any one of claims 1 to 5 having an oR circuit for calculating the logical sum of the output of the black image detector. And a scene change detecting device according to claim 1 to 6,
An analog / digital converter for converting an analog audio signal and an analog image signal into a digital audio signal and a digital image signal;
A delay circuit for delaying the digital image signal;
An encoder that encodes the digital audio signal and a digital image signal input via the delay circuit,
The audio information determination unit of the scene change detection device outputs the silence detection time information based on the digital audio signal, and the image information determination unit of the scene change detection device inputs the digital image input to the encoder. An encoding device that outputs the scene change time information based on a signal. A scene change detection method in a scene change detection apparatus to which a video composed of an audio signal and an image signal is input,
Detect silence state of audio signal, generate silence detection time information,
Detecting a change in luminance or a decrease in luminance level of an image in an image signal, and generating scene change time information indicating a time at which the luminance change or a decrease in luminance level is detected;
Generating a first scene change candidate time based on the silence detection time information and the scene change time information;
The effectiveness of the first scene change candidate time is determined based on the amount of change in the audio level of the audio signal at a time later than the first scene change candidate time and a preset minimum determination value, and is effective Output the first scene change candidate time determined as the second scene change candidate time,
Calculating a first detection interval based on a time difference between a current time of the second scene change candidate times and a time one hour before the current time of the second scene change candidate times;
A second detection interval based on a time difference between a time one hour before the current time of the second scene change candidate time and a time two times before the current time of the second scene change candidate time. To calculate
If the first detection interval is equal to or less than the maximum detection value and the second detection interval is greater than the maximum detection value, the second scene change candidate time one time before is determined from the scene change detection information. Output as start time,
When the first detection interval is larger than the detection maximum value and the second detection interval is less than or equal to the detection maximum value, the second scene change candidate time one time before is determined as the scene change detection information. A scene change detection method that outputs the end time .

Images