JP2015163988A - Device, method and program for determining presence or absence of section including common signal - Google Patents

Abstract

PROBLEM TO BE SOLVED: To enable high-speed determination of presence or absence of a section including a corresponding signal between an input signal and an accumulation signal.SOLUTION: An input signal characteristic vector extracting unit 101 extracts a characteristic vector about a partial signal of an input signal, and an input signal characteristic vector integrating unit 102 generates an input signal integrated characteristic vector in which characteristic vectors of all sections are integrated. An accumulation signal characteristic vector extracting unit 103 extracts a characteristic vector about a partial signal of an accumulation signal, and an accumulation signal characteristic vector integrating unit 104 generates an accumulation signal integrated characteristic vector in which characteristic vectors of all the sections are integrated. An integrated characteristic vector similarity calculation unit 105 calculates integrated similarity which is similarity of the input signal integrated characteristic vector and the accumulation signal integrated characteristic vector corresponding to a section taken out according to a prescribed criterion, and a determination unit 106 determines presence or absence of a section including the same signal between the input signal and the accumulation signal on the basis of magnitude of the integrated similarity.

Description

  The present invention provides a common signal containing section presence / absence determination method and apparatus for determining the presence or absence of a section including a common signal when a time series signal called an input signal and a time series signal called an accumulation signal are given. And the program.

  The presence / absence determination of the common signal containing section includes the same signal between the input signal and the accumulated signal when a time series signal called an input signal and a time series signal called an accumulated signal are given as shown in FIG. The presence or absence of a certain section is determined.

  This common signal containing section presence / absence determination is conventionally performed by cutting out a signal of a small section called a reference signal from the input signal while shifting the same signal by detecting the section including the reference signal from the accumulated signal for each cut out reference signal. This can be realized by determining whether or not there is a section including the symbol (hereinafter, this method is called the brute force method).

  When the time-series signal is an acoustic signal, the technique of Patent Document 1 (hereinafter, the technique of Patent Document 1 is referred to as the division match search method 2) for detection from the accumulated signal of the extracted reference signal in the brute force method. Can be used.

Japanese Patent No. 4327202

  However, in the brute force method, it is necessary to cut out reference signals from a plurality of locations of the input signal, the division match search method 2 must be executed for each cut out reference signal, and the accumulated signal is In the case of a long signal, there is a problem that it takes time to execute the brute force method because it takes time to execute the division match search method 2 once.

  The present invention has been made in consideration of the above circumstances, and it is possible to determine the presence / absence of a common signal-containing section that can quickly determine the presence / absence of a section including a corresponding signal between an input signal and an accumulated signal. It is an object to provide a method, an apparatus, and a program.

  In order to achieve the above object, the common signal containing section presence / absence determination method according to the present invention includes a section including a corresponding signal between an input signal that is a time-series signal and an accumulated signal that is a time-series signal. A common signal containing section presence / absence determination method for determining presence / absence, wherein an input signal feature vector extraction unit extracts a feature vector for each partial signal extracted from the input signal based on a predetermined reference, and an input signal feature The vector integration unit generates an input signal integrated feature vector that integrates the feature vectors of all sections extracted by the input signal feature vector extraction unit, and the stored signal feature vector extraction unit extracts from the stored signal on a predetermined basis The feature vectors are extracted from the partial signals of all the sections, and the stored signal feature vector integration unit extracts the stored signals. An accumulated signal integrated feature vector that integrates feature vectors of all sections extracted by the feature vector extraction unit is generated, and the predetermined reference generated by the input signal feature vector integration unit by an integrated feature vector similarity calculation unit is generated. The similarity between the input signal integrated feature vector corresponding to the section extracted in step (1) and the accumulated signal integrated feature vector corresponding to the section extracted based on the predetermined reference generated by the stored signal feature vector integration unit. Is calculated as an integrated similarity that is a similarity for determining whether or not there is a section including a corresponding signal between the input signal and the accumulated signal, and the integrated feature vector similarity is calculated by a determination unit. Depending on the degree of integrated similarity calculated by the calculation unit, a corresponding signal is included between the input signal and the accumulated signal. It determines the presence or absence of that section.

  The common signal containing section presence / absence determination device according to the present invention determines a presence / absence of a section including a corresponding signal between an input signal that is a time series signal and an accumulated signal that is a time series signal. A presence / absence determination device, wherein an input signal feature vector extraction unit that extracts a feature vector for each partial signal extracted from the input signal according to a predetermined reference, and all the signals extracted by the input signal feature vector extraction unit An input signal feature vector integration unit that generates an input signal integrated feature vector that integrates feature vectors of sections, and a stored signal feature vector that extracts feature vectors for each partial signal extracted from the stored signal according to a predetermined reference Integration unit integrating the extraction unit and the feature vectors of all sections extracted by the storage signal feature vector extraction unit An accumulated signal feature vector integrating unit for generating a collection vector; and the input signal integrated feature vector corresponding to the section extracted by the predetermined reference, generated by the input signal feature vector integrating unit; and the accumulated signal feature vector A signal corresponding to the interval between the input signal and the stored signal, which is similar to the stored signal integrated feature vector corresponding to the section extracted by the predetermined reference, is generated by the integration unit. An integrated feature vector similarity calculating unit that calculates as an integrated similarity that is a similarity for determining the presence or absence of a section, and the input signal based on the size of the integrated similarity calculated by the integrated feature vector similarity calculating unit And a determination unit that determines the presence or absence of a section in which a corresponding signal is included between the accumulated signals.

  The program which concerns on this invention is a program for functioning a computer as each part of said common signal containing area presence determination apparatus.

  As described above, according to the common signal containing section presence / absence determination method, apparatus, and program of the present invention, an input signal integrated feature vector obtained by integrating feature vectors extracted for each section of the input signal is generated, Generating a stored signal integrated feature vector by integrating the feature vectors extracted for each section of the stored signal, and calculating an integrated similarity using the similarity between the input signal integrated feature vector and the stored signal integrated feature vector As a result, it is possible to determine at high speed whether or not there is a section including a corresponding signal between the input signal and the accumulated signal.

It is a block diagram which shows the example of 1 structure of the common signal containing area presence determination apparatus which concerns on the 1st Embodiment of this invention. It is a conceptual diagram explaining the process which extracts a feature vector from a video signal. It is a conceptual diagram explaining the process which extracts a feature vector from an acoustic signal. It is a conceptual diagram explaining the process which produces | generates an input signal integrated feature vector. It is a conceptual diagram explaining the process which calculates integrated feature vector similarity. It is a flowchart which shows the content of the common signal containing area presence determination processing routine in the common signal containing area presence determination apparatus which concerns on the 1st Embodiment of this invention. It is a conceptual diagram explaining the process which produces | generates an input signal integrated feature vector. It is a conceptual diagram explaining the process which calculates integrated feature vector similarity. It is a conceptual diagram explaining the process which integrates a feature vector. It is a block diagram which shows the example of 1 structure of the integrated feature vector similarity calculation part of the common signal containing area presence determination apparatus which concerns on the 4th Embodiment of this invention. It is a flowchart which shows the content of the common signal containing area presence determination processing routine in the common signal containing area presence determination apparatus which concerns on the 4th Embodiment of this invention. It is a conceptual diagram explaining the process which determines the presence or absence of the area where the same signal is contained between an input signal and the accumulation | storage signal.

  Hereinafter, embodiments of the present invention will be described in detail with reference to the drawings.

[First Embodiment]
<System configuration>
FIG. 1 is a block diagram showing an embodiment of a common signal containing section presence / absence determining apparatus according to the present invention. The common signal containing section presence / absence determination apparatus shown in FIG. 1 is an apparatus that determines the presence / absence of a section including the same signal between a specific video signal called an input signal and a video signal called an accumulated signal. Yes, specifically, a computer comprising a CPU (Central Processing Unit), a RAM, and a ROM storing a program for executing a common signal containing section presence / absence determination routine described later, and functionally The configuration is as follows.

  The common signal containing section presence / absence determining device includes an input signal feature vector extracting unit 101, an input signal feature vector integrating unit 102, an accumulated signal feature vector extracting unit 103, an accumulated signal feature vector integrating unit 104, an integrated feature vector similarity calculating unit 105, And a determination unit 106.

Input signal feature vector extraction unit 101, the partial signals of each section of the input signal (the start time and t interval), as shown in FIG. 2, to extract each feature vector V ^I _t.

Specifically, each frame of the focused section is divided into small areas of a certain size, encoded for each small area, and a predetermined small area of interest (for example, a user-specified small area, etc.) Each code of the small area is extracted, and a feature vector in which the extracted codes are arranged as elements is generated. Here, the time of the frame is t _i , the code obtained from the w-th small area is written as C _{ti, w,} and the element of the feature vector V ^I _t whose start time is t is V ^I _{t, 1} , V ^I _{t, 2} , ..., V ^I _{t, d} .

Note that the case where the input signal and the accumulation signal are video signals has been described as an example, but the present invention is not limited to this. For example, the input signal and the accumulation signal may be acoustic signals. In this case, a power spectrum is extracted from an acoustic signal that is an input signal to obtain a spectrogram. Then, from the spectrogram of the input signal, for each section of the input signal (the start time of the interval to t), as shown in FIG. 3, extracts a feature vector V ^I _t.

  Specifically, the spectrumgram of the focused section is divided into small areas of a certain size, encoded for each small area, and a predetermined small area of interest (for example, a user-specified small area or an equal interval) Are extracted, and a feature vector in which the extracted codes are arranged as elements is generated.

Input signal feature vector integration section 102 generates an input signal integration feature vector which integrates feature vector V ^I _t of each section extracted by the input signal feature vector extraction unit 101.

  Specifically, as shown in FIG. 4, the feature vector of each section extracted by the input signal feature vector extraction unit 101

L ^I sections, and for each dimension of the feature vector, for each dimension of the feature vector, the set of all the elements that appeared in the dimension of the feature vector of each section

Elements of the corresponding dimension of

As a result, an input signal integrated feature vector is generated.

Similar to the input signal feature vector extraction unit 101, the accumulated signal feature vector extraction unit 103 extracts each of the feature vectors V ^S _t from the partial signals in each section (the start time of the section is t) of the accumulated signal. To do.

Stored signal feature vector integration section 104 generates an accumulation signal integration feature vector which integrates feature vector V ^S _t of each section extracted by the stored signal feature vector extraction unit 103.

  Specifically, similar to the input signal feature vector integration unit 102, the feature vectors of each section extracted by the accumulated signal feature vector extraction unit 103

Are collected for l ^S sections, and for each dimension of the feature vector, a set of all elements appearing in the dimension of the feature vector of each section is stored as the accumulated signal integrated feature vector.

Elements of the corresponding dimension of

As a result, an accumulated signal integrated feature vector is generated.

The integrated feature vector similarity calculating unit 105 uses the similarity between the input signal integrated feature vector generated by the input signal feature vector integrating unit 102 and the accumulated signal integrated feature vector generated by the accumulated signal feature vector integrating unit. As shown in FIG. 5, according to the following formula, the integrated similarity between the section summed up by l ^I and the section summed up by l ^S is calculated.

Here, s (u ^I , u ^S ) is a similarity between codes u ^I and u ^S defined in advance.

  Based on the integrated similarity calculated by the integrated feature vector similarity calculation unit 105, the determination unit 106 determines whether there is a section including the same signal between the input signal and the accumulated signal.

<Operation of common signal containing section presence / absence judging device>
Next, the operation of the common signal containing section presence / absence determining apparatus according to the present embodiment will be described.

  In the common signal containing section presence / absence determination device, the common signal containing section presence / absence determination processing routine shown in FIG. 6 is executed.

  First, in step S101, the input signal feature vector extraction unit 101 reads an input signal. Then, the input signal feature vector extraction unit 101 extracts a feature vector of each section from the input signal.

  In the next step S102, the input signal feature vector integration unit 102 integrates the feature vectors of each section extracted in step S101 to generate an input signal integrated feature vector.

  In step S103, the accumulated signal feature vector extraction unit 103 reads the accumulated signal. Then, the accumulated signal feature vector extraction unit 103 extracts the feature vector of each section from the accumulated signal.

  In the next step S104, the accumulated signal feature vector integration unit 104 integrates the feature vectors of each section extracted in step S103, and generates an accumulated signal integrated feature vector.

  In step S105, the integrated feature vector similarity calculation unit 105 calculates the similarity between the input signal integrated feature vector generated in step S102 and the accumulated signal integrated feature vector generated in step S104. Calculate as degrees. In step S106, the determination unit 106 includes a common signal between the input signal and the accumulated signal based on whether or not the integrated similarity calculated in step S105 is greater than or equal to a threshold value. The presence / absence of the section is determined, the determination result is output, and the common signal containing section presence / absence determination processing routine is terminated.

  As described above, according to the common signal containing section presence / absence determination device according to the embodiment of the present invention, an input signal integrated feature vector that integrates feature vectors extracted for each section of the input signal is generated, Generating a stored signal integrated feature vector by integrating the feature vectors extracted for each section of the stored signal, and calculating an integrated similarity using the similarity between the input signal integrated feature vector and the stored signal integrated feature vector Thus, since the number of comparisons can be reduced by comparing the integrated feature vectors, the presence / absence of a section in which the same signal is included between the input signal and the accumulated signal can be determined at high speed. This reduces the number of executions of the division match search method 2 in the brute force method, and also reduces the number of sections to be searched in the division match search method 2 in the accumulated signal, thereby quickly determining whether or not there is a common signal containing section. Equivalent to doing.

[Second Embodiment]
<System configuration>
Next, a second embodiment will be described. Note that the configuration of the common signal containing section presence / absence determination device according to the second embodiment is the same as that of the first embodiment, and thus the same reference numerals are given and description thereof is omitted.

  In the second embodiment, a method for extracting feature vectors, a method for integrating feature vectors, and a method for calculating integrated similarity are different from those in the first embodiment.

The input signal feature vector extraction unit 101 of the common signal containing section presence / absence determining device according to the second embodiment represents a histogram for the partial signal in each section (the start time of the section is t) of the input signal. extracting a characteristic vector ^{V I} _t.

  Specifically, each frame of the focused section is divided into small areas of a certain size, encoded for each small area, and a predetermined small area of interest (for example, a user-specified small area, etc.) Each code of the small area is extracted, the appearance frequency is counted for each extracted code, and a feature vector representing a histogram in which the appearance frequencies for each code are arranged as elements is generated.

  Note that the case where the input signal and the accumulation signal are video signals has been described as an example, but the present invention is not limited to this. For example, the input signal and the accumulation signal may be acoustic signals.

Input signal feature vector integration section 102 generates an input signal integration feature vector which integrates feature vector V ^I _t of each section extracted by the input signal feature vector extraction unit 101.

  Specifically, as shown in FIG. 7, the feature vector of each section extracted by the input signal feature vector extraction unit 101

Are collected for l ^I sections, and the histograms of the feature vectors are overlapped to obtain the maximum value of the appearance frequency for each code, and the input signal integrated feature vector

Is generated.

Similar to the input signal feature vector extraction unit 101, the accumulated signal feature vector extraction unit 103 represents a feature vector V ^S _t representing a histogram for each partial signal of the interval (starting time of the interval is t) of the accumulation signal. To extract.

Stored signal feature vector integration section 104 generates an accumulation signal integration feature vector which integrates feature vector V ^S _t of each section extracted by the stored signal feature vector extraction unit 103.

  Specifically, similar to the input signal feature vector integration unit 102, the feature vectors of each section extracted by the accumulated signal feature vector extraction unit 103

Are collected for l ^S sections, the histograms of feature vectors are overlaid, the maximum value of appearance frequency is obtained for each code, and the accumulated signal integrated feature vector

Is generated.

As shown in FIG. 8, the integrated feature vector similarity calculation unit 105 includes an input signal integrated feature vector generated by the input signal feature vector integration unit 102, and an accumulated signal integrated feature vector generated by the accumulated signal feature vector integration unit. the similarity between, calculated as integrated similarity between the l ^I min summarized section and l ^S content summary interval.

  However, the similarity between the input signal integrated feature vector and the accumulated signal integrated feature vector may be based on a normally considered distance between vectors (for example, cosine similarity).

  In addition, about the other structure and effect | action of the common signal containing area presence determination apparatus concerning 2nd Embodiment, since it is the same as that of 1st Embodiment, description is abbreviate | omitted.

  In this way, an input signal integrated feature vector is generated by integrating the feature vectors representing the histogram extracted for each section of the input signal, and the feature vector representing the histogram extracted for each section of the accumulated signal is integrated. The integrated feature vector is generated, and the integrated similarity is calculated by using the similarity between the input signal integrated feature vector and the stored signal integrated feature vector. Since it can be reduced, the presence or absence of a section in which the same signal is included between the input signal and the accumulated signal can be determined at high speed.

[Third Embodiment]
<System configuration>
Next, a third embodiment will be described. Note that the configuration of the common signal containing section presence / absence determining device according to the third embodiment is the same as that of the first embodiment, and thus the same reference numerals are given and description thereof is omitted.

  In the third embodiment, a method for integrating feature vectors is different from that in the second embodiment.

Similar to the method described in the second embodiment, the input signal feature vector extraction unit 101 of the common signal containing section presence / absence determination device according to the third embodiment the partial signal of the start time and t), extracts a feature vector V ^I _t representing the histogram.

Input signal feature vector integration section 102 generates an input signal integration feature vector which integrates feature vector V ^I _t of each section extracted by the input signal feature vector extraction unit 101.

  Specifically, the feature vector of each section extracted by the input signal feature vector extraction unit 101

Are combined for l ^I sections, and as shown in FIG. 9, the histograms of the feature vectors are summed, and the sum of the appearance frequencies is obtained for each code, and the input signal integrated feature vector

Is generated.

Similar to the method described in the second embodiment, the accumulated signal feature vector extraction unit 103 represents a histogram for a partial signal of each section (the start time of the section is t) of the accumulated signal. to extract the vector V ^S _t.

Stored signal feature vector integration section 104 generates an accumulation signal integration feature vector which integrates feature vector V ^S _t of each section extracted by the stored signal feature vector extraction unit 103.

  Specifically, similar to the input signal feature vector integration unit 102, the feature vectors of each section extracted by the accumulated signal feature vector extraction unit 103

Are combined for l ^S sections, and as shown in FIG. 9 above, the histograms of the feature vectors are summed up, the sum of the appearance frequencies is obtained for each code, and the accumulated signal integrated feature vector

Is generated.

As in the second embodiment, the integrated feature vector similarity calculation unit 105 includes the input signal integrated feature vector generated by the input signal feature vector integration unit 102 and the storage generated by the storage signal feature vector integration unit. the similarity between the signal integration feature vectors is calculated as the integrated similarity between the l ^I min summarized section and l ^S content summary interval.

  In addition, about the other structure and effect | action of the common signal containing area presence determination apparatus which concern on 3rd Embodiment, since it is the same as that of 1st Embodiment, description is abbreviate | omitted.

  In this way, the integrated input signal feature vector is generated so as to add the feature vectors representing the histogram extracted for each section of the input signal, and the histogram extracted for each section of the accumulated signal is represented. By generating the accumulated signal integrated feature vector that is integrated so as to add the feature vectors and calculating the integrated similarity using the similarity between the input signal integrated feature vector and the accumulated signal integrated feature vector, Since the number of comparisons can be reduced by performing the comparison, it is possible to determine at high speed whether there is a section in which the same signal is included between the input signal and the accumulated signal.

[Fourth Embodiment]
<System configuration>
Next, a fourth embodiment will be described. Note that the configuration of the common signal containing section presence / absence determining device according to the fourth embodiment is the same as that of the first embodiment, and therefore the same reference numerals are given and description thereof is omitted.

  In the fourth embodiment, when the similarity between the input signal integrated feature vector and the accumulated signal integrated feature vector is greater than or equal to the threshold value, the above-described division match search method 2 is executed to input the input signal integrated feature vector. The difference between the first embodiment and the second embodiment is that the similarity between the feature vector of the section from which is extracted and the feature vector of the section from which the accumulated signal integrated feature vector is extracted is used as the integrated similarity.

  As shown in FIG. 10, the integrated feature vector similarity calculation unit 105 of the common signal containing section presence / absence determination device according to the fourth embodiment includes a first stage similarity calculation unit 301, an integrated similarity calculation determination unit 302, And an integrated similarity calculation unit 303.

  Similar to the integrated feature vector similarity calculation unit 105 according to the first embodiment, the first stage similarity calculation unit 301 calculates the similarity between the input signal integrated feature vector and the accumulated signal integrated feature vector. To do. This similarity is an integrated similarity in the first to third embodiments, but is called a first-stage similarity in this embodiment.

  The integrated similarity calculation determination unit 302 determines whether or not the first step similarity calculated by the first step similarity calculation unit 301 is larger than a preset threshold value.

  The integrated similarity calculation unit 303 calculates the integrated similarity based on the determination result by the integrated similarity calculation determination unit 302. When the first step similarity calculated by the integrated similarity calculation determination unit 302 is equal to or less than the threshold, the integrated similarity is set to 0. When the integrated similarity calculation determination unit 302 determines that the first-stage similarity exceeds the threshold, the feature vector used to create the input signal integrated feature vector is selected, and the selected feature vector is The maximum similarity among the similarities with the feature vectors used to create the accumulated signal integrated feature vector is obtained by the division match search method 2.

  Each feature vector used to create the input signal integrated feature vector is selected, and the maximum value is obtained from the maximum similarity obtained for each selected feature vector, and is set as the integrated similarity. The selection of the feature vector used to create the input signal integrated feature vector is performed, for example, by sampling from the feature vector used to create the input signal integrated feature vector. In addition, the case has been described as an example where the maximum value is obtained from the maximum similarity obtained for each selected feature vector and is set as the integrated similarity, but the present invention is not limited to this. Alternatively, the top N similarities may be obtained from the maximum similarity obtained for each feature vector, and the average value of the top N similarities may be used as the integrated similarity.

<Operation of common signal containing section presence / absence judging device>
Next, the operation of the common signal containing section presence / absence determining apparatus according to the present embodiment will be described. In addition, about the process similar to 1st Embodiment, the same code | symbol is attached | subjected and detailed description is abbreviate | omitted.

  In the common signal containing section presence / absence determination device, a common signal containing section presence / absence determination processing routine shown in FIG. 11 is executed.

  First, in step S101, the input signal feature vector extraction unit 101 reads an input signal. Then, the input signal feature vector extraction unit 101 extracts a feature vector of each section of the input signal.

  In the next step S102, the input signal feature vector integration unit 102 integrates the feature vectors of each section extracted in step S101 to generate an input signal integrated feature vector.

  In step S103, the accumulated signal feature vector extraction unit 103 reads the accumulated signal. Then, the accumulated signal feature vector extraction unit 103 extracts a feature vector of each section of the accumulated signal.

  In the next step S104, the accumulated signal feature vector integration unit 104 integrates the feature vectors of each section extracted in step S103, and generates an accumulated signal integrated feature vector.

  In step S301, the first-stage similarity calculation unit 301 determines the similarity between the input signal integrated feature vector generated in step S102 and the accumulated signal integrated feature vector generated in step S104 as the first level. Calculate as step similarity.

  In the next step S302, the integrated similarity calculation determination unit 302 determines whether or not the first-stage similarity calculated in step S301 is greater than a threshold value. If it is determined that the first-stage similarity is greater than the threshold, in step S303, the integrated similarity calculation unit 303 calculates the feature vector to be calculated from the feature vector of the input signal extracted in step S101. Select.

  In the next step S304, the integrated similarity calculation unit 303 calculates the similarity between the feature vector to be calculated selected in step S303 and each of the feature vectors of the accumulated signal extracted in step S103. Find the maximum similarity.

  In step S305, it is determined whether or not the processes in steps S303 and S304 have been executed for all the feature vectors of the input signal. If there is a feature vector of the input signal that has not been subjected to the processes of steps S303 and S304, the process returns to step S303, the feature vector is selected as a calculation target, and the processes of steps S303 and S304 are executed. . On the other hand, when the processes of steps S303 and S304 are executed for all the feature vectors of the input signal, in step S306, the integrated similarity calculation unit 303 determines the maximum similarity obtained in step S304. The maximum value is calculated as the integrated similarity.

  On the other hand, if it is determined that the first-stage similarity is equal to or less than the threshold value, the integrated similarity calculation unit 303 sets the integrated similarity to 0 in step S307.

  In step S106, the determination unit 106 determines that a common signal between the input signal and the accumulated signal is based on whether the integrated similarity calculated in step S306 or S307 is equal to or greater than a threshold value. The presence / absence of the included section is determined, the determination result is output, and the common signal containing section presence / absence determination processing routine is terminated.

  As described above, according to the common signal containing section presence / absence determination device according to the embodiment of the present invention, an input signal integrated feature vector that integrates feature vectors extracted for each section of the input signal is generated, Generates a stored signal integrated feature vector that integrates the feature vectors extracted for each section of the stored signal, and calculates the first-stage similarity using the similarity between the input signal integrated feature vector and the stored signal integrated feature vector To do. When the first-stage similarity is equal to or greater than the threshold, the division match search method 2 is executed, and the feature vector of the section from which the input signal integrated feature vector is extracted and the feature vector of the section from which the accumulated signal integrated feature vector is extracted By calculating the similarity and setting the integrated similarity, it is possible to determine at high speed whether there is a section in which the same signal is included between the input signal and the accumulated signal.

  Note that the present invention is not limited to the above-described embodiment, and various modifications and applications are possible without departing from the gist of the present invention.

  For example, although the above-described common signal containing section presence / absence determination apparatus has a computer system therein, the “computer system” is a homepage providing environment (or display environment) if a WWW system is used. Shall also be included.

  In the present specification, the embodiment has been described in which the program is installed in advance. However, the program can be provided by being stored in a computer-readable recording medium.

  For example, when the time-series signal is an acoustic signal, the input signal is a recorded sound of TV broadcast, and the accumulated signal is one music CD, a part of the accumulated signal in the input signal, Alternatively, it can be determined whether or not everything has been used as BGM.

  Also, when the time series signal is a video signal, the input signal is a recorded TV broadcast video, and the accumulated signal is a video signal such as a video clip, a part of the accumulated signal is included in the input signal. Or, it can be determined whether the whole has been used.

101 Input Signal Feature Vector Extraction Unit 102 Input Signal Feature Vector Integration Unit 103 Accumulated Signal Feature Vector Extraction Unit 104 Accumulated Signal Feature Vector Integration Unit 105 Integrated Feature Vector Similarity Calculation Unit 106 Determination Unit 301 First Stage Similarity Calculation Unit 302 Integrated Similarity Degree calculation determination unit 303 Integrated similarity calculation unit

Claims (7)

A common signal containing section presence / absence determination method for determining the presence / absence of a section including a corresponding signal between an input signal that is a time series signal and an accumulated signal that is a time series signal,
The input signal feature vector extraction unit extracts a feature vector for each partial signal extracted from the input signal based on a predetermined standard,
An input signal feature vector integration unit generates an input signal integration feature vector that integrates the feature vectors of all sections extracted by the input signal feature vector extraction unit,
The accumulated signal feature vector extracting unit extracts a feature vector for each partial signal extracted from the accumulated signal based on a predetermined standard,
An accumulated signal feature vector integrating unit generates an accumulated signal integrated feature vector obtained by integrating the feature vectors of all sections extracted by the accumulated signal feature vector extracting unit,
Generated by the integrated feature vector similarity calculation unit by the input signal feature vector integration unit, which is generated by the input signal feature vector integration unit, corresponding to the section extracted by the predetermined reference, and the accumulated signal feature vector integration unit The degree of similarity with the accumulated signal integrated feature vector corresponding to the section extracted based on the predetermined criterion is determined as to whether or not there is a section including a corresponding signal between the input signal and the accumulated signal. Calculate as the integrated similarity, which is the similarity for determining,
The determination unit determines whether or not there is a section including a corresponding signal between the input signal and the accumulated signal based on the size of the integrated similarity calculated by the integrated feature vector similarity calculation unit. Method for determining presence / absence of contained section. The generation of the input signal integrated feature vector by the input signal feature vector integration unit is based on the feature vectors of all sections extracted by the input signal feature vector extraction unit, A histogram in which the appearance frequency is counted is generated as the input signal integrated feature vector,
The generation of the accumulated signal integrated feature vector by the accumulated signal feature vector integrating unit is based on the feature vectors of all sections extracted by the accumulated signal feature vector extracting unit, A histogram in which the appearance frequency is counted is generated as the accumulated signal integrated feature vector,
The calculation of the integrated similarity by the integrated feature vector similarity calculating unit calculates the similarity between the histogram of the input signal integrated feature vector and the histogram of the accumulated signal integrated feature vector as the integrated similarity. The common signal containing section presence / absence determination method according to claim 1.   The calculation of the integrated similarity by the integrated feature vector similarity calculating unit includes an input signal integrated feature vector generated by the input signal feature vector integrating unit, and an accumulated signal generated by the accumulated signal feature vector integrating unit. The similarity with the integrated feature vector is calculated, and when the calculated similarity is larger than the threshold value, the feature vector of all sections extracted by the input signal feature vector extraction unit and the accumulated signal feature vector extraction unit are extracted. The common signal containing section presence / absence determination method according to claim 1, wherein the integrated similarity is calculated based on the similarity of each pair with the feature vectors of all sections that have been performed. A common signal containing section presence / absence determination apparatus that determines the presence or absence of a section that includes a corresponding signal between an input signal that is a time series signal and an accumulated signal that is a time series signal,
An input signal feature vector extraction unit that extracts a feature vector for each partial signal of all the sections extracted from the input signal by a predetermined reference;
An input signal feature vector integrating unit that generates an input signal integrated feature vector obtained by integrating the feature vectors of all sections extracted by the input signal feature vector extracting unit;
An accumulated signal feature vector extracting unit that extracts a feature vector for each partial signal of all the sections extracted from the accumulated signal on a predetermined basis;
An accumulated signal feature vector integrating unit for generating an accumulated signal integrated feature vector obtained by integrating the feature vectors of all sections extracted by the accumulated signal feature vector extracting unit;
The input signal feature vector integration unit that is generated by the input signal feature vector integration unit and that corresponds to the section extracted by the predetermined criterion, and the storage signal feature vector integration unit that is extracted by the predetermined criterion. Integration that is a similarity for determining whether or not there is a section that includes a corresponding signal between the input signal and the stored signal, and the similarity to the stored signal integrated feature vector corresponding to the section An integrated feature vector similarity calculator for calculating the similarity;
A determination unit that determines the presence or absence of a section in which a corresponding signal is included between the input signal and the accumulated signal according to the size of the integrated similarity calculated by the integrated feature vector similarity calculation unit;
Common signal containing section presence / absence determination device including The input signal feature vector integration unit, based on the feature vectors of all sections extracted by the input signal feature vector extraction unit, a histogram obtained by counting up the appearance frequency of the elements for each element of the feature vector, Generated as an integrated feature vector,
The accumulated signal feature vector integrating unit, based on the feature vectors of all sections extracted by the accumulated signal feature vector extracting unit, a histogram obtained by counting up the appearance frequency of the elements for each element of the feature vector, Generated as an integrated feature vector,
5. The common signal containing section according to claim 4, wherein the integrated feature vector similarity calculation unit calculates the similarity between the histogram of the input signal integrated feature vector and the histogram of the accumulated signal integrated feature vector as the integrated similarity. Presence / absence judging device.   The integrated feature vector similarity calculating unit calculates a similarity between the input signal integrated feature vector generated by the input signal feature vector integrating unit and the accumulated signal integrated feature vector generated by the accumulated signal feature vector integrating unit. When the calculated similarity is larger than a threshold value, the feature vector of all sections extracted by the input signal feature vector extraction unit and the feature vector of all sections extracted by the accumulated signal feature vector extraction unit 6. The common signal containing section presence / absence determining apparatus according to claim 4, wherein the integrated similarity is calculated based on the similarity of each pair.   The program for functioning a computer as each part of the common signal containing area presence determination apparatus of any one of Claims 4-6.