A kind of audio frequency method for detecting
Technical field
The invention belongs to surveillance equipment field, particularly a kind of audio frequency method for detecting.
Background technology
For simple video monitoring mode, poor effect under night and overcast sky, voice signal is not then by the impact of light and weather.Voice Surveillance product has acoustic pickup, player, voice mail system, sound recorded book etc., and core is acoustic pickup.For audio frequency and video integrated monitoring, video camera is eyes, and acoustic pickup is ear.But simple with sound size for whether reporting to the police according to detecting, blowing a whistle the natural phonation such as sound, a sound firecrackers sound as thunder, car also can trigger alarm, causes wrong report.Meanwhile, it can not use in a noisy environment, is very easy to fail to report and report by mistake.
Summary of the invention
In order to overcome the deficiencies in the prior art, the invention provides a kind of audio frequency method for detecting, determine whether to there is illegal invasion with audio frequency, carry out the detection of abnormal sound, the method can get rid of the noise or paroxysmal noise effect or interference that there is regularity around, can not to blow a whistle the natural sound of sound etc. and trigger alarm because of neighbouring thunder, car, more can not be ineffective near the business district of background noise complexity, be mainly used in the sound detection of safety defense monitoring system under comparatively complex environment in this way.
Technical scheme provided by the invention is:
A kind of audio frequency method for detecting, comprises the following steps:
After step one, Voice Surveillance are opened, sample with the environmental background noise of the sampling rate of 8k ~ 44.1kHz to guarded region and calculate energy value corresponding to each sampled point, sampling time length S1 is 20 ~ 40 seconds, by the energy value that all sampled points in this time period of comparison are corresponding, obtain maximal value E
maxwith minimum value E
min;
After step 2, the environmental background noise to guarded region described in step one carry out sampling in 20 ~ 40 seconds, then sample with the sound of sampling time interval Δ t to guarded region of 1/64 ~ 1/2 second and calculate energy value corresponding to each sampled point, sampling time length S2 is the time span that exists of abnormal sound that expectation that user sets listens to 5 ~ 60 seconds, the energy value that each sampled point obtained in sampling time length S2 is corresponding all compares with a warning reference threshold, and the computing method of described warning reference threshold are E
max+ (E
max-E
min) * x, wherein, x is the noise sensitivity that user is arranged, if when the sampled point number being greater than described warning reference threshold exceedes numerical value N/2, then trigger alarm, if when the sampled point number being greater than described warning reference threshold is less than numerical value of N/2, then not trigger alarm, N is the sampled point number S2/ Δ t in sampling time length S2;
When the sampled point number being greater than described warning reference threshold in sampling time length S2 described in step 3, step 2 is less than numerical value of N/2, repeat the step in step 2;
Step 4, every 10-60 minute, repeat the implementation of step one, step 2 and step 3.
Preferably, described audio frequency method for detecting, in step one, sample with the environmental background noise of the sampling rate of 44.1kHz to guarded region, sampling time length S1 is 30 seconds.
Preferably, described audio frequency method for detecting, the sampling time interval Δ t described in step 2 is 1/16.
Preferably, described audio frequency method for detecting, the noise sensitivity x numerical value that the user described in step 2 is arranged is 0.1 ~ 10.0.
Preferably, described audio frequency method for detecting, after the sampled point number being greater than described warning reference threshold is greater than numerical value of N/2 trigger alarm, repeats the implementation in step one and step 2 in the length S2 of sampling time described in step 2.
Wherein, the mode of integration is used to carry out cumulatively obtaining energy value to sound wave area during the energy value that calculating sampling point is corresponding in sampling time length S1 or sampling time length S2 described in step one neutralization procedure two.
Audio frequency method for detecting of the present invention, compared with prior art, one is can run under more constant noisy environment, prevent from failing to report alert, the method can get rid of the noise that there is regularity around, the such as sound of the long time running such as air-conditioning, motor, the brouhaha near business district, Che Sheng, the sound etc. of the lathe running of the environment of plant.Two is to get rid of thunder, and car is blown a whistle the interference of the natural causes such as sound, prevents from making a false report alert.Three is the modes adopting integration, carries out the detection of abnormal sound, more reliable than the compare threshold of pure.
Accompanying drawing explanation
Fig. 1 is the process flow diagram of audio frequency method for detecting of the present invention.
Embodiment
Below in conjunction with accompanying drawing, the present invention is described in further detail, can implement according to this with reference to instructions word to make those skilled in the art.
As shown in Figure 1, the invention provides a kind of audio frequency method for detecting, comprising:
After step one, Voice Surveillance are opened, with the sampling rate of 44.1kHz, the environmental background noise of sampling depth to guarded region of 8/16bit is sampled and is calculated energy value corresponding to each sampled point, sampling time length S1 is 30 seconds, the mode of integration is adopted to carry out the cumulative energy value obtaining sound to sound wave area during energy value corresponding to calculating sampling point, obtain one group of energy value like this, by the energy value in this group of comparison, obtain maximal value E
maxwith minimum value E
min;
After step 2, the environmental background noise to guarded region described in step one carry out sampling in 30 seconds, then sample with the sound of sampling time interval Δ t to guarded region of 1/16 second and calculate energy value corresponding to each sampled point, sampling time length S2 is the time span that exists of abnormal sound that expectation that user sets listens to 5 ~ 60 seconds, the energy value that each sampled point obtained in sampling time length S2 is corresponding all compares with a warning reference threshold, and the computing method of described warning reference threshold are E
max+ (E
max-E
min) * x, wherein, x is the noise sensitivity that user is arranged, if when the sampled point number being greater than described warning reference threshold exceedes numerical value N/2, then trigger alarm, if when the sampled point number being greater than described warning reference threshold is less than numerical value of N/2, then not trigger alarm, N is the sampled point number S2/ Δ t in sampling time length S2;
When the sampled point number being greater than described warning reference threshold in sampling time length S2 described in step 3, step 2 is less than numerical value of N/2, repeat the step in step 2;
Step 4, E
maxand E
minwarning reference threshold according to the actual needs of user, can upgrade once every 10-60 minute, to adapt to the change of background noise environment, prevents failing to report and making a false report of alert, so every 10-60 minute, repeated the implementation of step one, step 2 and step 3.
Described audio frequency method for detecting, the noise sensitivity x numerical value that the user described in step 2 is arranged is 0.1 ~ 10.0.
Described audio frequency method for detecting, after the sampled point number being greater than described warning reference threshold is greater than numerical value of N/2 trigger alarm, repeats the implementation in step one and step 2 in the length S2 of sampling time described in step 2.
Although embodiment of the present invention are open as above, but it is not restricted to listed in instructions and embodiment utilization, it can be applied to various applicable the field of the invention completely, for those skilled in the art, can easily realize other amendment, therefore do not deviating under the universal that claim and equivalency range limit, the present invention is not limited to specific details and illustrates here and the legend described.