JP2003281680A - Accident sound detecting device - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To discriminate quick brake sounds from siren sounds, which have been hitherto impossible, and to efficiently detect only accident sounds to improve the ability of detecting such accident sounds. <P>SOLUTION: The accident sound detecting device for detecting accident sounds using an sound signal from the external comprises an accident sound acquiring means 20 acquiring an accident sound, an accident sound detecting means 30 detecting the accident sound from an output signal of the accident sound acquiring means 20, an periodically accidental sound acquiring means 40 acquiring accident sounds from sound signals at predetermined intervals, a pitch variation determining/computing means 50 determining the magnitude of the pitch variation of an accident sound from an output signal from this periodically accident sound acquiring means 40, and a pitch variation sound removing means 60 removing a pitch variation sound having a large pitch variation from the accident sound in the case where the pitch variation determined by the pitch variation determining/computing means 50 is larger than a predetermined value. <P>COPYRIGHT: (C)2004,JPO

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to an accident for efficiently detecting only an accident sound from traffic sounds at an intersection or the like, accident sounds (including accident sounds such as collisions and sudden braking sounds leading to accidents). The present invention relates to a sound detecting device.

[0002]

2. Description of the Related Art Generally, since many traffic accidents occur at intersections, a traffic accident status recording device for recording the status of accident occurrence has been conventionally considered. FIG. 8 is a block diagram showing a conventional traffic accident situation recording device disclosed in, for example, Japanese Patent Laid-Open No. 4-338900. In FIG. 8, 1,
4 is a traffic signal, 1B, 1Y, 1R, 4B, 4Y, 4R
Is a signal light, 2B, 2Y, 2R is a signal line, 3 is a signal light blinking device, 5 is a signal light symbol generator, 6 is a time character generator,
7 is an adder, 8 is an imaging device, 9 is an endless recording device,
Reference numeral 10 is an ambient sound detection device, 11 is a collision sound detection circuit, 12 is a recording control device, and 13 is a magnetic recording / reproducing device.

The parts that detect accident sounds are the ambient sound detection device 10 and the collision sound detection circuit 11. That is, the acoustic signal is detected by the ambient sound detection device 10 and sent to the collision sound detection circuit 11. The collision sound detection circuit 11 determines whether or not the detected acoustic signal is a collision sound of the vehicle, and when it is determined to be a collision sound, a collision sound detection signal is sent to the recording control device 12. In the collision sound detection circuit 11, it is determined whether the detected traffic sound is an accident (collision) sound by comparing the level of the input acoustic signal with a predetermined threshold value, and the level is equal to or higher than the threshold value. If it is, it is judged as an accident.
As a measure against malfunction, the ambient sound detecting device 10 uses a directional microphone and a band elimination filter in the path of the acoustic signal.

[0004]

By the way, in the collision sound detection circuit described in the above publication, the level comparison between the acoustic signal detected in the discrimination of the collision sound and a preset fixed threshold value is carried out. However, in general traffic sounds, there are many high-level sounds other than accident sounds, such as sudden braking sounds, horn sounds, and runaway sounds. Therefore, in the method of performing the level comparison between the detected acoustic signal and the fixed threshold value, there is a problem that many other traffic sounds are also identified as accident sounds in addition to the actual accident sound.
Further, there is a problem in that no information about the type of the detected accident sound such as a collision sound or a sudden braking sound can be obtained.

The present invention has been made in order to solve the above problems. Instead of using a signal level threshold value for determining an accident sound, the accident sound is captured at regular intervals and The difference between the sudden braking sound and the siren sound, which could not be distinguished until now, is calculated by calculating the fluctuation of the fundamental frequency (pitch fluctuation), judging that it is a siren sound when the pitch fluctuation is large, and excluding it from the accident sound. Is possible and
It is an object of the present invention to provide an accident sound detection device capable of improving the accident sound detection capability.

[0006]

An accident sound detecting apparatus according to the invention of claim 1 is an accident sound detecting apparatus for detecting an accident sound using an external acoustic signal, and takes in the accident sound. Accident sound capturing means, accident sound detecting means for detecting an accident sound from an output signal of the accident sound capturing means, and always capturing a signal at regular intervals regardless of the presence or absence of the accident sound, and each of them has a capturing time. A periodic accident sound capturing means for outputting a plurality of equal output signals at regular time intervals, and a pitch fluctuation determining calculating means for determining a pitch fluctuation of the accident sound from an output signal from the periodic accident sound capturing means, When the pitch fluctuation judged by the pitch fluctuation judgment calculating means is larger than a predetermined value, pitch fluctuation sound removing means for excluding a pitch fluctuation sound having a large pitch fluctuation from the accident sound is provided.

The accident sound detecting device according to the invention of claim 2 is
The accident sound capturing means includes a waveform shaping circuit that shapes the waveform of the external acoustic signal, an A / D converter that converts an analog signal of the waveform shaping circuit into a digital signal, and the A
The difference calculating means for extracting the difference signal from the output signal from the / D converter is compared with the difference signal from the difference calculating means and the reference difference value having a preset value, and the trigger signal is generated according to the comparison result. It has a comparing circuit which is generated, and a switch which takes in the output signal from the A / D converter into the pitch variation judging calculating means on the basis of a trigger signal from the comparing circuit.

The accident sound detecting device according to the invention of claim 3 is
A plurality of autocorrelation calculating means for calculating the autocorrelation function from the plurality of output signals fetched from the periodic accident sound capturing means, respectively; A plurality of basic frequency calculating means for calculating a corresponding basic frequency based on each autocorrelation function, and pitch fluctuation of the output signal from the A / D converter based on each basic frequency from the plurality of basic frequency calculating means. And pitch fluctuation determining means for determining.

The accident sound detecting device according to the invention of claim 4 is
The pitch variation determining means obtains a pitch variation rate based on each fundamental frequency from the plurality of fundamental frequency calculating means, and determines that the pitch variation rate is outside the accident sound only when the pitch variation rate is larger than a predetermined value. To do.

[0010]

BEST MODE FOR CARRYING OUT THE INVENTION An embodiment of the present invention will be described below with reference to the drawings. Embodiment 1. 1 is a functional block diagram showing a first embodiment of the present invention. In FIG. 1, 20 is an accident sound capturing means, 100 is an acoustic input signal from the ambient sound detecting device 10 (see FIG. 8), 110 is an output signal of the accident sound capturing means 20, and 30 is an accident sound capturing means 20. From the output signal 110 from the accident sound detecting means, 120 is an output signal of the accident sound detecting means 30, and 40 is a periodic accident sound capturing which constantly takes in a signal at regular intervals regardless of the presence or absence of the accident sound. Inclusive means,
Reference numerals 130 and 140 denote two output signals output from the periodic accident sound capturing means 40 with a constant time difference. In addition,
There are two or more outputs of the periodic accident sound capturing means 40, and each is output at a constant time equal to the capturing time. Further, 50 is an output signal 130,
A pitch fluctuation judgment calculation means for judging the magnitude of the pitch fluctuation of 140, 150 an output signal of the pitch fluctuation judgment calculation means 50, and 60 an output signal 15 of the pitch fluctuation judgment calculation means 50.
Based on 0, pitch fluctuation sound removing means for removing pitch fluctuation sound such as siren sound from the output signal 120 of the accident sound detecting means 30, and 160 is an output signal of the pitch fluctuation sound removing means 60.

FIG. 2 is a circuit configuration diagram showing an example of a specific circuit configuration of the accident sound capturing means 20. In FIG. 2, 2
1 is a waveform shaping circuit for shaping the waveform of an acoustic signal 100 from the outside, 101 is an output signal from the waveform shaping circuit 21, 2
2 is an A / D converter for converting the output signal 101 from the waveform shaping circuit 21 from an analog signal to a digital signal, and 102 is a digital output signal of the A / D converter 22.
Reference numeral 23 is a difference calculation means for extracting a difference signal from the digital output signal 102, 103 is a difference signal from the difference calculation means 23, and 25 is a reference difference value generation circuit for generating a difference signal (reference difference value) of a reference level. , 24 compares the difference signal 103 from the difference calculating means 23 with the reference difference value from the reference difference value generating circuit 25, and the difference calculating means 23
Is a comparison circuit that generates an output signal 104 when the difference signal 103 from is larger than the set reference difference value. Two
6 is a switch for taking in the output signal 102 of the A / D converter 22 by using the output signal 104 from the comparison circuit 24 as a trigger signal, and 110 is an output signal (OUT) taken in by the switch 26.

FIG. 3 is a circuit configuration diagram showing an example of a specific circuit configuration of the pitch variation determination calculation means 30. In FIG. 3, reference numeral 51 is an autocorrelation calculating means for calculating the autocorrelation function of the output signal 130 from the periodic accident sound capturing means 40, that is, the input signal IN1, and 52 is the fundamental frequency of the output signal from the autocorrelation calculating means 51 ( Pitch) f1 is calculated by a fundamental frequency calculating means, 53 is similarly an autocorrelation calculating means for calculating an autocorrelation function of the output signal 140 from the periodic accident sound capturing means 40, that is, the input signal IN2, and 54 is an autocorrelation calculation. The basic frequency calculation means for calculating the basic frequency (pitch) f2 of the output signal from the means 53, 55 is the basic frequency f1 from the basic frequency calculation means 52 and the basic frequency f2 from the basic frequency calculation means 54 to determine the magnitude of pitch fluctuation. It is a pitch fluctuation determination means for determination.

Next, the operation will be described with reference to FIGS. Now, when an acoustic input signal 100 is generated when an accident occurs and is input to the accident sound capturing means 20, the waveform shaping circuit 21 removes unnecessary signal components, and the output signal 10
1 is generated. The output signal 101 is converted from an analog signal to a digital signal 102 by the A / D converter 22.
The output digital signal 102 is input to the difference calculating means 23, and the digital output signal 1 is output from the difference calculating means 23.
The differential signal 103 of 02 is output. The differential signal 103 generally represents a change in signal level, and a larger signal level is generated as the level changes abruptly. On the contrary, if the signal level is large and the level change is small, the level is small. Therefore, when a sudden phenomenon occurs, for example, when an accident occurs, the differential signal 103 shows a large level change.
Therefore, the reference difference value generation circuit 25 generates a reference difference value as a reference in advance, and the comparison circuit 24 compares the difference signal 103 with the reference difference value. When the difference signal 103 is larger than the reference difference value, it is determined that a traffic accident sound whose signal level changes abruptly is generated, and the output signal 104 is generated as a trigger signal.

The switch 26 is turned on by the output signal 110 from the comparison circuit 24 to take in the output signal 102 of the A / D converter 22 and input to the accident sound detecting means 30 as the output signal 110. The accident sound detecting means 30 detects the accident sound and outputs the output signal 120 as the pitch fluctuation sound removing means 6
Enter 0. Further, the periodic accident sound capturing means 40 captures the acoustic input signal 100 at regular time intervals and outputs it as output signals 130 and 140 to the autocorrelation computing means 51 and 53 of the pitch variation determination computing means 50, respectively.

The autocorrelation calculation means 51 and 53 calculate the autocorrelation function of the input output signals 111 and 112. The spectrumgram of each output signal input to the autocorrelation calculation means 51 and 53 is as shown in FIG. As shown in, the frequency or pitch may change over time. Note that, in FIG. 4, the characteristic indicated by the thick line is the lowest frequency of each output signal, that is, the fundamental frequency f0
(Pitch). Therefore, the autocorrelation calculation means 5
In 1 and 53, the autocorrelation function as shown in FIG. 5 is calculated. Here, the autocorrelation function is symmetric with respect to time 0, and is a function that repeats peaks and valleys with time in positive time. If the time given to the position of the first mountain by this autocorrelation function is t0, its reciprocal 1 / t0 becomes the fundamental frequency (pitch) f0.

The output signals of the autocorrelation calculating means 51 and 53 are supplied to the fundamental frequency calculating means 52 and 54, respectively, and the fundamental frequency (pitch) is obtained from the autocorrelation function as described above. Therefore, the periodic accident sound capturing means 40
A basic frequency fi-1 (= f1) having an acquisition time of ti-1 is calculated for the output signal 130 acquired from the same, and similarly, for the output signal 140 from the periodic accident sound acquisition means 40. A fundamental frequency fi (= f2) with the acquisition time ti is calculated.

FIG. 6 shows the basic frequency calculating means 5 at this time.
The relationship between the fundamental frequencies f1 and f2 calculated in 2 and 54 is shown as, for example, fi-1 and fi, and the time ti-1 is the time when the output signal 130 is captured with respect to the fundamental frequency fi-1. Substantially, the time ti substantially represents the time when the output signal 140 is acquired with respect to the fundamental frequency fi.

In this way, the fundamental frequency calculation means 52
The fundamental frequencies f1 and f2 obtained in steps 54 and 54 are supplied to the pitch variation determining means 55, where the pitch variation rate r
p is required. This pitch variation rate rp is the fundamental frequency f
2 divided by f1, that is, rp = f2 / f1 (rp = fi
/ Fi-1).

Next, in the pitch fluctuation judging means 55, as shown in FIG.
As shown in, it is determined whether the calculated pitch fluctuation rate rp is larger than a predetermined pitch fluctuation rate Δr. If the pitch fluctuation rate rp is larger than a predetermined pitch fluctuation rate Δr, the periodic accident sound capture is performed. The signal fetched from the means 40 to the pitch fluctuation judgment calculation means 50 has pitch fluctuation, that is, it is judged as pitch fluctuation sound, and when the pitch fluctuation rate rp is smaller than a predetermined pitch fluctuation rate Δr, the periodic accident sound fetching is carried out. The signal fetched from the means 40 to the pitch fluctuation judgment calculating means 50 is judged to have no pitch fluctuation, that is, an accident sound, and the judgment result is output to the pitch fluctuation sound removing means 60 as an output signal 15.
Supply as 0.

In the pitch fluctuation sound removing means 40, when the input output signal 150 is a signal indicating that the pitch fluctuation is large, this is a pitch fluctuation sound, for example, a siren sound, so that the signal component is excluded from the accident sound. Output signal 16
Outputs 0. Further, in the pitch fluctuation sound removing means 60,
When the input output signal 150 is a signal indicating that the pitch fluctuation is small, it is an accident sound, and therefore, for example, an output signal 160 for a trigger signal for recording this accident sound without removing the signal component. Output as.

As described above, in this embodiment, in order to distinguish between the sudden braking sound and the siren sound in the input acoustic signal such as accident sound, the fluctuation of the fundamental frequency (pitch) is calculated and the pitch thereof is calculated. If the fluctuation is larger than a predetermined value, it is judged as a siren sound and excluded from the accident sound, and if the pitch fluctuation is smaller than the predetermined value, a trigger signal for recording as an accident sound is output. It is possible to distinguish between the sudden braking sound and the siren sound that could not be distinguished, and it is possible to efficiently detect only the accident sound and improve the ability to detect the accident sound.

[0022]

As described above, according to the first aspect of the invention, there is provided an accident sound detecting device for detecting an accident sound by using an acoustic signal from the outside, wherein the accident sound is taken in. A sound capturing means, an accident sound detecting means for detecting an accident sound from an output signal of the accident sound capturing means, and a signal which is always captured at regular intervals regardless of the presence or absence of the accident sound, and each of which is equal to the capturing time. A periodic accident sound capturing means for outputting a plurality of output signals at regular time intervals; and a pitch fluctuation determining calculating means for determining the pitch fluctuation of the accident sound from the output signal from the periodic accident sound capturing means, When the pitch fluctuation judged by the pitch fluctuation judgment calculating means is larger than a predetermined value, the pitch fluctuation sound removing means for excluding the pitch fluctuation sound having a large pitch fluctuation from the accident sound is provided. Suddenly It is possible to identify the rake sound and siren sound, accident sound only effectively detect and there is an effect that it improve the detection capability of an accident sound.

According to a second aspect of the present invention, the accident sound capturing means shapes a waveform shaping circuit for shaping the waveform of the external acoustic signal, and converts the analog signal of the waveform shaping circuit into a digital signal. And a difference calculation means for extracting a difference signal from an output signal from the A / D converter, a difference signal from the difference calculation means, and a reference difference value having a preset value. Since it has a comparison circuit that generates a trigger signal according to the comparison result and a switch that takes in the output signal from the A / D converter to the pitch variation determination calculation means based on the trigger signal from the comparison circuit, There is an effect that it can contribute to efficiently detecting only the accident sound.

According to the third aspect of the present invention, the pitch fluctuation judgment calculating means calculates a plurality of self-correlation functions from the plurality of output signals taken from the periodic accident sound taking means. Correlation calculating means, and a plurality of basic frequency calculating means for calculating a corresponding basic frequency based on each autocorrelation function from the plurality of autocorrelation calculating means,
Since the pitch fluctuation judging means for judging the pitch fluctuation of the output signal from the A / D converter based on each fundamental frequency from the plurality of fundamental frequency calculating means is provided, it is possible to distinguish between the sudden braking sound and the siren sound. It is possible to contribute to improving the ability to detect accident sounds.

Further, according to the invention of claim 4, the pitch variation judging means obtains a pitch variation rate based on each fundamental frequency from the plurality of fundamental frequency computing means, and the pitch variation rate is less than a predetermined value. Since the pitch fluctuation sound outside the accident sound is determined only when it is large, there is an effect that the pitch fluctuation of the accident sound can be efficiently and reliably determined.

[Brief description of drawings]

FIG. 1 is a functional block diagram showing a first embodiment of the present invention.

FIG. 2 is a block diagram showing an example of accident sound capturing means according to the first embodiment of the present invention.

FIG. 3 is a block diagram showing an example of pitch variation determination calculation means in the first embodiment of the present invention.

FIG. 4 is a diagram for explaining the operation of the autocorrelation calculation means in the first embodiment of the present invention.

FIG. 5 is a diagram for explaining the operation of the autocorrelation calculation means in the first embodiment of the present invention.

FIG. 6 is a diagram for explaining the operation of the fundamental frequency calculating means in the first embodiment of the present invention.

FIG. 7 is a diagram for explaining the operation of the pitch variation determination means in the first embodiment of the present invention.

FIG. 8 is a block diagram showing a conventional traffic accident situation recording device.

[Explanation of symbols]

20 means for capturing accident sound, 21 waveform shaping circuit, 22 A /
D converter, 23 difference calculation means, 24 comparison circuit, 25
Reference difference value generating circuit, 26 switch, 30 accident sound detecting means, 40 periodic accident sound taking means, 50 pitch variation judging calculating means, 51, 53 autocorrelation calculating means, 5
2, 54 basic frequency calculation means, 55 pitch fluctuation determination means, 60 pitch fluctuation sound removing means.

Continued front page    F term (reference) 5C086 AA54 AA60 BA22 CA09 CB26                       DA01 DA18 EA11 EA15 EA38                       EA45 FA17 FA20 GA04                 5H180 CC11 EE15

Claims (4)

[Claims] 1. An accident sound detecting device for detecting an accident sound using an external acoustic signal, comprising: accident sound capturing means for capturing the accident sound; and an output signal of the accident sound capturing means. Accident sound detection means for detecting an accident sound, and a periodic accident sound collection that constantly captures a signal at regular time intervals regardless of the presence or absence of the accident sound and outputs a plurality of output signals equal to the acquisition time at regular time intervals. Input means, pitch fluctuation determination calculation means for determining the pitch fluctuation of the accident sound from the output signal from the periodic accident sound acquisition means, and pitch fluctuation determined by the pitch fluctuation determination calculation means is larger than a predetermined value. In this case, the accident sound detecting device is provided with a pitch fluctuation sound removing means for excluding the pitch fluctuation sound having a large pitch fluctuation from the accident sound. 2. The accident sound capturing means includes a waveform shaping circuit for shaping the waveform of the external acoustic signal, and an A / D for converting an analog signal of the waveform shaping circuit into a digital signal.
A converter, a difference calculation means for extracting a difference signal from an output signal from the A / D converter, a difference signal from the difference calculation means and a reference difference value having a preset value are compared, and the comparison result is obtained. And a comparator circuit for generating a trigger signal according to
The accident sound detecting device according to claim 1, further comprising a switch for taking an output signal from the D converter into the pitch variation determination calculating means. 3. The pitch fluctuation determination calculation means, a plurality of autocorrelation calculation means for calculating respective autocorrelation functions from a plurality of output signals fetched from the periodic accident sound capture means, and the plurality of self correlation calculation means. A plurality of fundamental frequency calculating means for calculating a corresponding fundamental frequency based on each autocorrelation function from the correlation calculating means; and a plurality of fundamental frequency calculating means for calculating the fundamental frequencies from the plurality of fundamental frequency calculating means. 3. The accident sound detecting apparatus according to claim 2, further comprising a pitch fluctuation judging means for judging a pitch fluctuation of the output signal. 4. The pitch variation determining means obtains a pitch variation rate based on each fundamental frequency from the plurality of fundamental frequency calculating means, and only when the pitch variation rate is larger than a predetermined value, the pitch variation outside the accident sound is generated. The accident sound detection device according to claim 3, wherein the accident sound detection device determines that the sound is sound.