JP2002333460A - Signal discriminator - Google Patents

Abstract

PROBLEM TO BE SOLVED: To allow exact discrimination even in a signal of which the leading time or the attenuation time is not defined clearly. SOLUTION: In this signal discriminator provided with an input means 31 for taking a signal from a discrimination object in, extraction means 32, 33 for extracting a feature quantity of the signal, and a judging means 34 for judging a condition of the discrimination object based on the feature quantity, a degree of asymmetry and/or a degree of peakedness is used as the feature quantity of the signal.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a method and an apparatus for discriminating a type of a signal, and more particularly to a method for discriminating a signal that changes with time based on a characteristic on a time axis.

[0002]

2. Description of the Related Art Discrimination of non-stationary signals, that is, signals whose quality and quantity change significantly with time, is necessary in various situations. One example is noise discrimination of machinery and equipment. In order to reduce the noise of machinery, it is necessary to identify the sound source that causes noise. For example, in a device having many movable parts such as a copying machine and a printer, a large amount of unsteady sound is generated due to collision of parts. In order to identify the cause of the non-stationary sound, a technique for analyzing the non-stationary signal is required. Another example is device failure determination. It monitors the signal waveform of the sound and vibration generated from the machine, and notifies a failure when a signal that is not expected in normal operation is observed.

[0003] Non-stationary signals are generally distinguished by monitoring the power of the signal and analyzing the frequency. When a state where the fluctuation value of the signal power exceeds a predetermined threshold value is observed, it is determined that an unsteady signal has occurred, and the waveform portion thereof is subjected to frequency analysis to determine whether the sound is a problem sound. That is. Also, Japanese Patent Application Laid-Open No. 2000-275096 discloses a method of dividing a spectrum into subbands and using a neural network when it is difficult to determine by simple signal level and frequency analysis.

[0004] However, since all of the methods are developed on the frequency axis, some of the characteristics of the waveform appearing on the time axis are lost. In order to evaluate the characteristics of the signal, performing analysis directly on the time axis may provide more accurate discrimination in some cases.

On the other hand, JP-A-8-30874 and JP-A-2000-146929 disclose a method of discriminating a signal using a decay time of a waveform.

[0006]

However, the fact that the signal can be determined based on the decay time or the rise time has a drawback that the rising portion and the attenuating portion of the waveform are limited to clear ones. For example, in noise
Sounds such as those involving friction have an irregular rise in sound pressure, and a number of peaks of similar magnitude appear, making it difficult to define the rise / decay time.

The present invention has been made in view of such technical problems, and a purpose thereof is to provide a signal discriminating apparatus capable of accurately discriminating a signal whose rise time and decay time cannot be clearly defined. To provide.

[0008]

SUMMARY OF THE INVENTION The present invention is a signal discriminating apparatus having the following configuration.

The present invention comprises input means for taking in a (time) signal from a discrimination target, extraction means for extracting a characteristic amount of the signal, and judgment means for judging the state of the discrimination target from the characteristic amount. In the signal discriminating apparatus, the degree of asymmetry of the signal is used as the characteristic amount of the signal. The present invention also provides a signal discriminating apparatus including an input unit that captures a signal from a discrimination target, an extraction unit that extracts a feature amount of the signal, and a determination unit that determines a state of the discrimination target from the feature amount. The degree of sharpness of the signal is used as the characteristic amount of the signal. Further, the present invention, input means for capturing a signal from the object of discrimination, extraction means for extracting the characteristic amount of the signal,
In a signal discriminating apparatus including a judgment unit for judging a state of the discrimination target from the feature amount, the degree of asymmetry and / or sharpness of the signal is used as the feature amount of the signal.

Further, the present invention provides the following steps (process).
Can be considered as a signal discrimination method including

The present invention comprises an input step of taking in a (time) signal from a discrimination target, an extraction step of extracting a characteristic amount of the signal, and a judging step of judging a state of the discrimination target from the characteristic amount. In the signal discrimination method, the degree of asymmetry of the signal is used as the characteristic amount of the signal. The present invention also provides a signal discrimination method including an input step of capturing a signal from a discrimination target, an extraction step of extracting a feature amount of the signal, and a determination step of determining a state of the discrimination target from the feature amount. The degree of sharpness of the signal is used as the characteristic amount of the signal. Further, the present invention provides an input step of capturing a signal from a discrimination target, an extraction step of extracting a feature amount of the signal, and a signal discrimination method including a determination step of determining a state of the discrimination target from the feature amount, As the characteristic amount of the signal, a degree of asymmetry and / or sharpness of the signal is used.

In these signal discriminating apparatuses and signal discriminating methods, signals from a discrimination target include a vibration signal and an acoustic signal. When the signal from the discrimination target is a vibration signal, the input unit may include a vibration pickup, and when the signal is a sound signal, the input unit may include a microphone.

[0013] The input means may include a correction section for correcting an input signal by a predetermined correction filter.
Similarly, the input step may include a correction step of correcting an input signal by a predetermined correction filter.
Here, specific examples of the correction filter include A, B, C, D,
Each characteristic filter of E can be used, a high-pass filter (to remove low frequency dark signal components), or other uniquely defined characteristic filters.

As the input signal, a signal which is particularly suitable for determination by the signal discriminating apparatus according to the present invention is, for example, a collision sound (impact between objects) or an impact sound having a duration of 1000 times.
This is an unsteady sound of less than [msec], and further less than 100 [msec]. Although there is no particular limitation on the source of the signal, the source of the signal is considered in consideration of the fact that there are countless moving parts in the device and that it is often installed in a place where a relatively quiet environment is required. Is an image forming apparatus (copier, printer, facsimile, etc.), the application of the present invention can be expected to have a great effect on noise reduction of the image forming apparatus, which is preferable.

Examples of the signal from the image forming apparatus include a sound signal from collision between components inside the image forming apparatus and a sound signal caused by a recording medium inside the image forming apparatus. More specifically, as the collision sound signal between the parts, a collision sound between metal parts, a collision sound between resin parts, a collision sound between metal parts and a resin part, and the like, and an acoustic signal caused by a recording medium are used. Examples include a sound of collision between a recording medium and a component, a buckling sound of the recording medium, and a sound of a trailing edge of the recording medium being conveyed in the transport direction.

When the degree of asymmetry and the degree of sharpness of the signal are used as the characteristic quantities of the signal, the determining means (or the determining step) performs the discrimination by multivariate analysis using the degree of asymmetry and degree of sharpness as variables. Can be determined. For example, the degree of asymmetry and the degree of sharpness are calculated in advance based on a signal having a clear cause, and a linear discriminant function for discriminating each cause is obtained. The state of the discrimination target can be determined based on the discriminant function. Further, for example, the degree of asymmetry and the degree of sharpness are calculated in advance based on a signal having a clear cause, and the average, variance, and covariance of the degree of asymmetry and the degree of sharpness are obtained for each cause, and are obtained based on a signal to be discriminated. The degree of asymmetry and sharpness and the average,
From the variance and covariance, the state of the discrimination target can be determined based on the Mahalanobis distance.

[0017]

Embodiments of the present invention will be described below with reference to examples.

Embodiment FIG. 1 illustrates a configuration of a signal judgment system (signal discriminating apparatus) 100 according to an embodiment of the present invention. The signal targeted by the signal determination system 100 is a noise (acoustic signal) emitted from the copying machine (image forming apparatus) T. In the signal determination system 100, the cause of the noise is either (1): collision of relatively hard parts (for example, metal parts) or (2): rubbing of a recording medium (paper). Is to judge.

The entire signal determination system 100 is a copying machine
A microphone (input means) 1 as a sound pickup unit for capturing noise from T, a DAT (digital audio tape) recorder (input means) 2 as an analog / digital conversion unit connected to the microphone 1, and a DAT recorder 2 connected to a personal computer system C. Further, the personal computer system C includes a computer main body 3, a keyboard 4a and a mouse 4b as input means, a display device 5 as output means, and the like.

The hardware resources in the computer main body 3 include a CPU as operation control means, a RAM as main storage means, a hard disk as auxiliary storage means, and an input / output control device (all not shown). And
The software resources in the computer main body 3 include an operating system, acoustic analysis software, numerical analysis software, and the like (all not shown). The functions shown in the following FIG. 2 are realized by the joint work of these hardware resources and software resources. Each function can be realized by an electronic circuit.

FIG. 2 is a basic functional block diagram of the signal determination system 100. This signal judgment system 100
The functions of the microphone 1, the DAT recorder 2, the waveform input unit (input means) 31,
It has an asymmetry calculating section (extracting means) 32, a sharpness calculating section (extracting means) 33, a judging section (judging means) 34, and a display device (display means) 5. The signals input and output between these basic functional blocks include an analog electric signal AS, a digital signal DS, and a time waveform f.
(T), degree of asymmetry α3, degree of sharpness α4, and determination result R.

FIG. 3 is a detailed functional block diagram for explaining the waveform input section 31 shown in FIG. 2 in more detail. The waveform input unit 31 includes a squaring unit 311, a smoothing unit 312, and a cutout unit 313. Further, the waveform input section 31 may include a storage section 310 for temporarily storing the digital signal DS. The signals input and output between these detailed functional blocks are a digital signal DS, a squared digital signal DS ² , a smoothed digital signal “DS ² ”,
This is a digital signal “ΔDS ² ” that has been cut out. Note that the cut-out processed digital signal “ΔDS ² ” corresponds to the time waveform f (t).

FIG. 4 is a flow chart for explaining the method of use and operation of the signal judgment system 100. Less than,
The operation of the signal determination system 100 according to the present embodiment will be described based on this flowchart.

First, prior to using the signal determination system 100, preparation processing is performed (step S in FIG. 4).
1). In this preparation process, the cause is clear (cause (1)
Or cause (2)) measure the sample noise (step S
2), digitization (step S3), and time waveform f
(T) is obtained (step S4), and the asymmetry α of the waveform is obtained.
₃ and the degree of sharpness α ₄ are calculated (steps S 5 and S 6), and a correlation map between the cause of the noise, the degree of asymmetry α ₃ and the degree of sharpness α ₄ is obtained and stored in the determination unit 34 in advance.

FIG. 5 schematically shows a correlation map stored in the judgment section 34 by this preparation processing. The vertical axis of the correlation map indicates the degree alpha ₄ kurtosis, the horizontal axis represents the degree of asymmetry alpha _3. Then, the cause (1): the values of the sharpness α ₄ and the asymmetry α ₃ obtained from the sample noise caused by the collision of the hard parts are plotted with black circles, and the cause (2) is the rubbing of the recording medium. The values of the degree of sharpness α ₄ and the degree of asymmetry α ₃ obtained from the sample noise
This is plotted with.

In this embodiment, the discriminating section 34 calculates a linear discriminant function J (α ₃ , α ₄ ) for discriminating the group of black circles and the group of white circles from the data shown in FIG. I remember. This linear discriminant function J (α ₃ , α ₄ ) is calculated by the degree of asymmetry α obtained from the acoustic signal due to the cause (1)
₃ (1) and the degree of sharpness α ₄ (1), J (α ₃
(1), α ₄ (1)) ≧ 0, degree of asymmetry α ₃ (2), degree of sharpness α ₄ (2) obtained from an acoustic signal due to cause (2)
Is calculated so that J (α ₃ (2), α ₄ (2)) <0.

When the preparatory process (step S1 in FIG. 4) is completed, discrimination of noise from the copying machine T whose cause is unknown is performed (steps S2 to S7 in FIG. 4).

First, noise to be determined (the cause of which is unknown) is measured by the microphone 1 (step S2 in FIG. 4). The noise from the copying machine T is measured by the microphone 1 installed close to the copying machine T. The noise from the copying machine T is a sound wave, which vibrates the vibrator in the microphone 1 and the vibration is converted into an electric signal AS (see FIG. 2). The electric signal AS is an analog signal.

Next, the electric signal AS is digitized by the DAT recorder 2 (step S3 in FIG. 4).
This is performed by converting the electric signal AS into the digital signal DS by the A / D conversion circuit in the DAT recorder 2 connected to the microphone 1 (see FIG. 2). In this embodiment, the resolution (sampling frequency) of the A / D conversion circuit of the DAT recorder 2 is 48 [kHz].
z], the obtained digital signal DS is obtained as 48,000 sampling data per second.

Next, the time waveform f is input by the waveform input section 31.
(T) is generated (step S4 in FIG. 4).

More specifically, first, the storage section 310 stores the digital signal DS recorded and output to the DAT recorder 2 (see FIG. 3). Specifically, it is stored in the hard disk of the personal computer main body 3 connected to the DAT recorder 2. FIG.
5 is a graph showing a digital signal DS stored in 0.
In this graph, the vertical axis represents sound pressure, and the horizontal axis represents the number of sampling data. In this embodiment, the storage unit 310 includes a collision sound whose absolute value of the sound pressure indicated by a circle in FIG. 6 has a maximum value, and stores about 6000 sample data before and after the collision sound in time.

Next, the squaring unit 311 stores the data in the storage unit 31.
0 squares the digital signals DS which stores, obtain two-squared processed digital signal DS ² (see FIG. 3). Next, the smoothing unit 12 smoothes (moves and averages) the squared digital signal DS ² with a predetermined parameter (for example, a 160-sample section), and obtains the smoothed digital signal “DS”.
² ”(see FIG. 3).

FIG. 7 shows a digital signal "DS
^TwoThe vertical axis represents the square of the sound pressure value,
The horizontal axis indicates the number of sampling data. And off
The extraction unit 312 outputs the smoothed digital signal.
"DS^Two”Is cut out and cut out digital
The signal “ΔDS^Two, Ie, obtain the time waveform f (t).
(See FIG. 3). The cutout unit 312 performs the obtained smoothing process.
Digital signal "DS" ^TwoOf the square of the sound pressure value
Large value P^Two(MAX) and its maximum value P^Two(MAX)
Is multiplied by a predetermined coefficient k to obtain a threshold value TH (= k × P^Two(MA
X)) and the smoothed digital signal “DS^TwoIs the threshold
From the position (t = 0) that is larger than the value TH, the smoothing process is performed.
Digital signal "DS"^TwoIs smaller than the threshold value TH.
To a position (t = T) where the time waveform f (t)
(T = 0 to t = T: the same applies hereinafter).

Next, based on the following equations (1) to (5) (more precisely, equations (1), (2), (3) and (4)), the degree of asymmetry α ₃ and the degree of sharpness α _{4 are} calculated. Is calculated.

[0035]

(Equation 1)

That is, the asymmetry calculating section 32 calculates the asymmetry α ₃ as the characteristic amount of the time waveform f (t) (see step S5 in FIG. 4). The degree of asymmetry α ₃ is different from the time distribution of the obtained time waveform f (t) in comparison with the waveform of the normal distribution corresponding to the equation (3) (see the graph of α ₃ = 0 in FIG. 8A). Indicates whether it is asymmetric.

First, the center of gravity μ of the time waveform f (t) is obtained by equation (1). Next, using the obtained center of gravity μ, the degree of spread σ of the time waveform f (t) is further obtained by Expression (2). Further, the degree of asymmetry α _{3 of the} time waveform f (t)
Is obtained from the center of gravity μ, the degree of spread σ, and Expression (4). The degree of asymmetry α ₃ is equivalent to the degree of skewness in the probability distribution, and the waveform of the normal distribution in which the obtained center of gravity μ is equal to the degree of spread σ can be obtained by Expression (5).

FIG. 8A illustrates the relationship between the sign of the degree of asymmetry α ₃ and the shape of the waveform. In the figure,
The waveform shown by the solid line (the waveform of the normal distribution shown by the equation (4)) is symmetric on the left and right, and the asymmetry α ₃ is expressed as “asymmetry α ₃ =
0 ". On the other hand, the waveform shown by the dashed line has a long tail at the right side, and the degree of asymmetry α ₃ is “asymmetry α ₃ <0”. On the other hand, the waveform shown by the two-dot chain line has a long tail on the left side,
The degree of asymmetry α ₃ is “asymmetry degree α ₃ > 0”.

Next, the sharpness calculating section 33 calculates the sharpness α ₄ as the characteristic amount of the time waveform f (t) (see step S6 in FIG. 4). As for the sharpness α _4, the obtained time waveform f (t) has a normal distribution waveform (FIG. 8).
(Refer to the graph of (b) in which α ₄ = 0).

This degree of sharpness α ₄ is equivalent to the degree of kurtosis in the probability distribution, and is determined in the previous step (step S5 in FIG. 4).
) And the degree of spread σ of the time waveform obtained in
It can be determined by equation (5).

FIG. 8B shows the sharpness α._FourPositive and negative and wave
This is to explain the relationship between the shape and the outline. Smell
And the waveform shown by the solid line (the wave of the normal distribution shown by the equation (4))
Shape) is the sharpness α_FourIs "sharpness degree α_Four= 0 ”
Become. On the other hand, the waveform shown by the dashed line has a blunt sharpness.
(Close to a trapezoid), its sharpness α_FourMeans "sharpness
α_Four<0 ”. On the other hand, the waveform shown by the two-dot chain line
The sharpness is the sharpness α _FourIs "sharpness degree α_Four> 0 "
Becomes

Next, the cause of the noise to be determined is determined by the determination unit 34, and a cause determination result R is obtained (step S7 in FIG. 4). This corresponds to the previous step S
5, based on the degree of asymmetry α ₃ and degree of sharpness α ₄ obtained in S6, and the linear discriminant function J (α ₃ , α ₄ ) previously obtained in the preparation stage of step S1 and stored in the determination unit 34. The cause of the noise is determined. That is, the degree of asymmetry α ₃ (i) and the degree of sharpness α ₄ (i) obtained from the noise (i) to be determined through steps S2 to S5 are determined by the linear discriminant function J stored in the determination unit 34. (Α ₃ , α ₄ )
If (α ₃ (i), α ₄ (i)) ≧ 0, the cause of the noise (i) as the determination result R can be determined to be cause (1): a collision between hard parts. (Α ₃ (i),
If α ₄ (i)) <0, it can be determined as the determination result R that the cause of the noise (i) is cause (2): rubbing of the recording medium.

Next, the display device 5 displays the determination result R (step S8 in FIG. 4). That is, when the cause of the noise (i) to be determined is “cause (1): collision between hard parts”, the fact is displayed on the screen of the display device 5 and “cause (2) ):
If it is caused by "rubbing of the recording medium", the fact is displayed on the screen of the display device 5.

When it is determined that the noise is unknown (step S9 in FIG. 4), step S2 is performed again.
And ends if it is to be ended.

According to the signal determination system 100 according to the present embodiment, the cause of noise generation is determined based on the degree of asymmetry α ₃ and the degree of sharpness α ₄ of the time waveform f (t). ) Can be accurately discriminated even when there are a plurality of peaks or when the maximum peak position is likely to fluctuate (see FIG. 7).

Modification The signal judgment system 100 according to the embodiment judges the cause of noise based on a linear discriminant function (α ₃ , α ₄ ). However, the signal judgment system 100 is different from the Mahalanobis distance D ₂ . The cause of the noise can also be determined based on this. Hereinafter will be described as a modification of the signal determination system 100 'to determine the cause of the noise on the basis of the Mahalanobis distance D _2. Note that the basic configuration and basic operation of the system are the same as those of the signal determination system 100 according to the embodiment, and a description thereof will be omitted.

Signal decision system 10 according to this modification
In the preparation process (step S1 in FIG. 4),
Unpaired noise obtained from multiple sample noises due to cause (1)
Degree α _Three, Sharpness α_FourAverage value (average α_Three(1), flat
Average α_Four(1)), dispersion (S^Twoα_Three(1), S^Twoα
_Four(1)), covariance (Sα_Threeα_Four(1))
Sharpness obtained from multiple sample noises due to cause (2)
Degree α_Four, Asymmetry α_ThreeAverage value (average α_Three(2), flat
Average α_Four(2)), dispersion (S^Twoα_Three(2), S^Twoα
_Four(2)), covariance (Sα_Threeα_Four(2))
The information is stored in the cutting unit 34.

When judging the cause of the noise (FIG. 4)
Step S7), based on the degree of asymmetry α ₃ and degree of sharpness α ₄ obtained in the previous steps S5 and S6, respectively, and the average value, variance, and covariance for each cause stored in the determination unit 34. Then, the Mahalanobis distance is calculated to determine the cause of the noise.

First, the degree of asymmetry α ₃ (i) obtained from the noise (i) to be determined through steps S2 to S5,
The sharpness α ₄ (i) and the average value (average α ₃ (1), average α) corresponding to the cause (1) stored in the determination unit 34
₄ (1)), the variance (S ² α ₃ (1), S ² α ₄ (1)), the covariance (Sα ₃ α ₄ (1)), and the equation (6) which is the arithmetic expression of the Mahalanobis distance. Then, the Mahalanobis distance D ² (1) between the noise (i) to be determined and the sample noise group due to the cause (1) is calculated.

[0050]

(Equation 2) Next, each of steps S2 to S2 is performed based on the noise (i) to be determined.
The degree of asymmetry α ₃ (i) obtained through S5 and the degree of sharpness α ₄
(I) and the average value (average α ₃ (2), average α ₄ (2)) corresponding to the cause (2) stored in the determination unit 34, the variance (S ² α ₃ (2), S ² α ₄ (2)), covariance (Sα ₃ α
₄ (2)) and the expression (6) which is the arithmetic expression of the Mahalanobis distance, the noise (i) to be judged and the cause (2)
Mahalanobis distance D ² from sample noise group
Is calculated.

Finally, the obtained Mahalanobis distance D
² (1) is compared with D ² (2), and D ² (1) ≦ D ² (2)
In the case of, the noise (i) can be determined to be the cause (1) due to the collision between the hard parts as the determination result R, and if D ² (1)> D ² (2), the determination is made Result R
It can be determined that the noise (i) is caused by the cause (2): rubbing of the recording medium.

The signal judging system 100 ′ according to the present modified example
To determine the cause of noise generation based on the degree of asymmetry α _{3 of} the time waveform f (t) and the degree of sharpness α ₄ ,
Even when there are a plurality of peaks in the time waveform f (t) or when the maximum peak position is likely to fluctuate (see FIG. 7), it is possible to accurately discriminate.

Test Example FIG. 9 is a graph illustrating the results of a test performed by the present inventors. In this graph, the horizontal axis represents the number of samples the noise measured, and the ordinate indicates the value of the asymmetry alpha ₃ obtained from each sample noise. Here, the number of samples is 1 to 25.
The sample noises up to the first are caused by the collision between the hard parts (1), and the sample noises of the 25th to 51st samples are caused by the rubbing of the recording medium (paper). . As is clear from this graph, the asymmetry α ₃ obtained from the sample noise due to the cause (1) is relatively large, and the asymmetry α ₃ obtained from the sample noise due to the cause (2) is relatively small. That is, there is a significant difference in the degree of asymmetry α ₃ of the time waveform f (t) depending on the cause.

[0054]

As described in detail above, according to the present invention, according to the present invention, when there are a plurality of peaks,
Even when the maximum peak position is likely to fluctuate, accurate discrimination can be performed.

[Brief description of the drawings]

FIG. 1 is an explanatory diagram illustrating an appearance of a signal determination system according to an embodiment.

FIG. 2 is a functional block diagram for explaining a basic function of the signal determination system shown in FIG. 1;

FIG. 3 is a functional block diagram for explaining a part of the functional block diagram shown in FIG. 2 in more detail;

FIG. 4 is a flowchart illustrating a basic operation of the signal determination system shown in FIG. 1;

FIG. 5 is a diagram for explaining the concept of a correlation map stored in advance by a determination unit in a preparation process;

FIG. 6 illustrates a digital signal obtained by measuring noise.

FIG. 7 illustrates a time waveform obtained by measuring noise.

FIG. 8 is a diagram for explaining the relationship between the values of the degree of asymmetry and the degree of sharpness, and the outline of the time waveform.

FIG. 9 illustrates the results of a test.

[Explanation of symbols]

T: copying machine (image forming apparatus), 100: signal determination system (signal discriminating apparatus), 1: macrophone (input means),
2 DAT recorder (input means), C personal computer system, 3 computer body, 5 display device (display means), 31 waveform input section (input means), 32 asymmetry calculation section (extraction means), 33: sharpness degree calculation unit (extraction means), 34: determination unit (judgment means), 310: storage unit, 311: squaring unit, 312: smoothing unit, 313: clipping unit

Claims (3)

[Claims] 1. A signal discriminating apparatus comprising: input means for receiving a signal from a discrimination target; extraction means for extracting a characteristic amount of the signal; and determination means for judging a state of the discrimination target from the characteristic amount. A signal discriminating apparatus characterized in that the degree of asymmetry of the signal is used as the characteristic amount of the signal. 2. A signal discriminating apparatus comprising: input means for receiving a signal from a discrimination target; extraction means for extracting a feature of the signal; and determination means for judging a state of the discrimination target from the feature. A signal discriminating apparatus characterized in that the sharpness of the signal is used as the characteristic amount of the signal. 3. A signal discriminating apparatus comprising: input means for receiving a signal from a discrimination target; extraction means for extracting a feature of the signal; and determination means for judging a state of the discrimination target from the feature. A signal discriminating apparatus characterized by using the degree of asymmetry and / or the degree of sharpness of the signal as a characteristic amount of the signal.