CN108872939B - Indoor space geometric outline reconstruction method based on acoustic mirror image model - Google Patents

Indoor space geometric outline reconstruction method based on acoustic mirror image model Download PDF

Info

Publication number
CN108872939B
CN108872939B CN201810405623.5A CN201810405623A CN108872939B CN 108872939 B CN108872939 B CN 108872939B CN 201810405623 A CN201810405623 A CN 201810405623A CN 108872939 B CN108872939 B CN 108872939B
Authority
CN
China
Prior art keywords
sound source
mirror image
sound
indoor
reconstruction
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Active
Application number
CN201810405623.5A
Other languages
Chinese (zh)
Other versions
CN108872939A (en
Inventor
宋浠瑜
王玫
仇洪冰
罗丽燕
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Guilin University of Electronic Technology
Original Assignee
Guilin University of Electronic Technology
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by Guilin University of Electronic Technology filed Critical Guilin University of Electronic Technology
Priority to CN201810405623.5A priority Critical patent/CN108872939B/en
Publication of CN108872939A publication Critical patent/CN108872939A/en
Application granted granted Critical
Publication of CN108872939B publication Critical patent/CN108872939B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Images

Classifications

    • G—PHYSICS
    • G01—MEASURING; TESTING
    • G01S—RADIO DIRECTION-FINDING; RADIO NAVIGATION; DETERMINING DISTANCE OR VELOCITY BY USE OF RADIO WAVES; LOCATING OR PRESENCE-DETECTING BY USE OF THE REFLECTION OR RERADIATION OF RADIO WAVES; ANALOGOUS ARRANGEMENTS USING OTHER WAVES
    • G01S5/00—Position-fixing by co-ordinating two or more direction or position line determinations; Position-fixing by co-ordinating two or more distance determinations
    • G01S5/18—Position-fixing by co-ordinating two or more direction or position line determinations; Position-fixing by co-ordinating two or more distance determinations using ultrasonic, sonic, or infrasonic waves
    • G01S5/22—Position of source determined by co-ordinating a plurality of position lines defined by path-difference measurements

Landscapes

  • Physics & Mathematics (AREA)
  • Engineering & Computer Science (AREA)
  • General Physics & Mathematics (AREA)
  • Radar, Positioning & Navigation (AREA)
  • Remote Sensing (AREA)
  • Stereophonic System (AREA)
  • Measurement Of Velocity Or Position Using Acoustic Or Ultrasonic Waves (AREA)

Abstract

The invention discloses an indoor space geometric outline reconstruction method based on an acoustic mirror image model, which comprises an acoustic receiving device, a phase transformation generalized cross-correlation analysis module, a sound source positioning module, a plane parameter analysis module, an acoustic mirror image model and the like, wherein a voice signal is used as a sound source signal, and a movable M array element star array is used as the acoustic receiving device; firstly, positioning a sound source based on controllable response power phase transformation and a region contraction method; secondly, based on an acoustic mirror model, obtaining estimated distance values of the speaker from indoor reflecting surfaces before and after single-step movement through cross-correlation analysis of sound transmitting and receiving signals; and finally, combining the shape prior of the indoor space, converting the space geometric outline reconstruction problem into a sound source positioning problem, and solving the position of the mirror image sound source to realize the indoor space geometric outline reconstruction. The method can improve the accuracy of indoor space geometric contour reconstruction, meanwhile, gives consideration to the universality of the sound source, and can obtain a robust contour reconstruction effect.

Description

Indoor space geometric outline reconstruction method based on acoustic mirror image model
Technical Field
The invention belongs to the field of research of sound field environment identification, relates to reconstruction of an indoor space geometric outline, and particularly relates to an indoor space geometric outline reconstruction method for positioning a sound source by adopting a controllable response power phase transformation combined with a region contraction method (SRP-PHAT-SRC) based on an acoustic mirror image model, which has important application value in scenes such as synthetic impulse response, auditory analysis, sound field reconstruction and the like.
Background
The propagation speed of the sound wave (20Hz-20KHz) in an indoor environment at room temperature (150C) is 340m/s, and the propagation speed is lower than that of an electromagnetic wave, so that the sound wave can carry room layout information in the propagation process. Through the cross-correlation analysis between the sound receiving and transmitting signals, the layout information carried in the received sound signals is recovered, and the method is an effective method for realizing the reconstruction of the indoor space geometric outline.
Some methods for realizing reconstruction of indoor space geometric outline by adopting acoustic multipath propagation inverse mapping include a scheme for realizing reconstruction by adopting combination of acoustic transceiver devices such as a single sound source and microphone array, a virtual multi-sound source and microphone array, a single sound source and a single microphone and the like to perform direct sound identification. However, in such a cross-correlation analysis mode, the accuracy in an environment with reverberation and noise cannot be guaranteed, and the inversion process of the reflection wall surface parameters is relatively complex, so that the accuracy and the real-time performance of the contour reconstruction are affected. And through indoor impulse response analysis, low-order reflected sound time delay is extracted, distance information is recovered, and the method is also an effective method for realizing indoor space geometric contour reconstruction.
The method for reconstructing the geometric contour of the indoor space by adopting single indoor impulse response analysis is available, but the time delay prior of first-order reflection and second-order reflection of each indoor reflection surface to sound propagation is needed, and the prior is very difficult to obtain in a real sound field environment due to the influence of reverberation and noise.
In addition, in the analysis research adopting a plurality of indoor impulse responses, the relation of the euclidean matrix rank between each array element of the microphone array and the mirror image sound source needs to be analyzed under the condition that a line-of-sight signal is available, and therefore the time delay of the first-order reflection and the second-order reflection of each reflecting surface to sound propagation needs to be identified. However, the decision method of the euclidean matrix rank fails in the case of non-line-of-sight. The best choice to solve this type of problem is a subspace-based complex search algorithm at the cost of real-time performance of the reconstruction scheme.
Disclosure of Invention
The invention provides a method for realizing indoor space geometric outline reconstruction based on an acoustic mirror image model, aiming at the problem that the reconstruction effect of an indoor space geometric outline reconstruction method is poor in severe environment due to large noise interference and serious reverberation influence in an indoor complex sound field environment. The method has the advantages that the influence of strong reverberation and noise on the time delay estimation precision of the received sound signal in the indoor complex sound field environment is better solved, the accuracy of indoor space geometric contour reconstruction is improved, meanwhile, the universality of a sound source is considered, and a robust contour reconstruction effect can be obtained.
According to the principles of plane geometry, a plane is uniquely defined when a point on the plane and an outward normal vector to the plane are known. Therefore, the movable M array element star-shaped array can be placed at the central position of the indoor reflection wall surface to serve as an acoustic receiving device, and the voice signal of a single speaker in the room serves as a sound source signal.
The invention discloses a method for realizing indoor space geometric outline reconstruction based on an acoustic mirror image model, which comprises a sound receiving device, a phase transformation generalized cross-correlation analysis module, a sound source positioning module, a plane parameter analysis module, a middle vertical plane construction module and the acoustic mirror image model, and comprises the following specific steps:
(1) a movable M array element star-shaped array is arranged at the central position of an indoor reflection wall surface to be used as a sound receiving device, the voice signal of a single indoor speaker is used as a sound source signal,
(2) carrying out sound source positioning based on controllable response power phase transformation and a region contraction method;
specifically, the sound arrival time TOA of a sound signal received by an M array element on a direct path is obtained through a phase transformation generalized cross-correlation analysis module, the SRP-PHA-SRC algorithm of an SRP-PHAT-SRC sound source positioning module is carried out, the SRP-PHA is optimized, and a speaker position coordinate s is given in an iterative mode;
(3) based on the acoustic mirror image model, the plane parameter analysis module obtains estimated values of distances from indoor reflecting surfaces of the speaker before and after single-step movement through cross-correlation analysis of the sound transmitting and receiving signals;
specifically, the M array element receiving array is moved to the central positions of different reflecting surfaces, the TOA minimum value of the M array elements is taken as the direct distance estimation from a sound source to the reflecting surface and is used as the input of a plane parameter analysis module to obtain the reflecting parameters of the reflecting surface
Figure GDA0002549261700000023
The position coordinates s and the parameters of the reflecting surface of the sound source
Figure GDA0002549261700000024
And normal vector n of reflecting surface l1, 2., 6 as input to the acoustic mirror model, the first order mirror sound source positions are estimated
Figure GDA0002549261700000021
(4) Combining the shape prior of the indoor space, converting the spatial geometric contour reconstruction problem into a sound source positioning problem, and quickly and effectively realizing the reconstruction of the indoor spatial geometric contour by solving the position of the mirror image sound source;
specifically, a sound source position s and each first-order mirror image sound source position are constructed through a middle vertical plane construction module
Figure GDA0002549261700000022
And obtaining the indoor geometric outline reconstruction result.
The plane parameter analysis module is the prior art, and the vertical plane construction module is provided by the inventor according to the inter-chamber geometric principle.
The specific signal processing process of each module related by the method of the invention is as follows:
the SRP-PHA-SRC sound source positioning module in the step (2): the method for positioning the sound source by the phase transformation weighted controllable response power (SRP-PHAT) combines the inherent robustness and the short-time analysis characteristic of the controllable response power method with the insensitivity of the phase transformation method to the sound field environment in the time delay estimation, so that the sound source positioning effect is robust under the influence of reverberation and noise. However, the accuracy and robustness of the localization effect comes at the expense of system real-time. Aiming at the problem, in order to ensure the accuracy and real-time performance of the reconstruction of the geometric outline of the indoor space, it is necessary to optimize the full-grid search method of the SRP-PHAT in the sound source space. The random area Shrinkage (SRC) is a coarse-grained parallel processing method, which is used to solve the computational problem of finding the global optimal solution of the objective function. Therefore, the sound source positioning model adopts an SRP-PHA-SRC algorithm which is the existing algorithm, the sound acquisition channel of the sound source positioning model is changed, the SRP-PHA is optimized, the matching rate of the SRC is improved, the total calculated amount can be greatly reduced, and the real-time performance of a contour reconstruction system is ensured.
The optimization in the step (2) comprises four steps:
(2.1) initializing sound source space parameters;
(2.2) the sound source search area is contracted;
(2.3) shrinkage parameter evaluation;
(2.4) the decision stops or continues to the global optimum;
the present invention makes changes to the decision phase, others remain the same as in the traditional.
The optimization of the step (2) comprises the following specific steps:
(2.1) initialization of spatial parameters of Sound Source
s is the position of the sound source,
Figure GDA0002549261700000031
are the possible sound source positions in the sound source space,
Figure GDA0002549261700000038
is the controllable response power of the phase transformation weights,
Figure GDA0002549261700000032
Figure GDA0002549261700000033
thus, the sound source position
Figure GDA0002549261700000034
Setting a spatial initial search rectangular region volume V0Number of random points in region J0Order ∈0＝E(J0) Then there is
Figure GDA0002549261700000035
Wherein N is0Is the number of valid points for the next search area volume;
(2.2) Sound Source search area contraction
For the object function in the current search region volume ViCarrying out random search internally to obtain random sampling point JiAnd calculating the effective point Ni，Ni＜＜JiFrom NiDistribution determination of a new search area volume Vi+1And the boundary thereof
Figure GDA0002549261700000036
Figure GDA0002549261700000037
(2.3) shrinkage parameter evaluation
Setting a threshold value:
① volume of minimum search area Vu；
② current search area calculation amount (FE)i) Maximum value of (d): phi;
③ maximum number of iteration steps: i;
when V isi+1≤VuThen, the search space representing the next step is smaller than the volume V of the minimum search areauAt this moment, the algorithm stops searching and saves the result;
when V isi+1＞VuThen, the volume of the search space representing the next step is larger than the volume V of the minimum search areauAt this time, considering the influence of the computing resource loss on the algorithm efficiency, V needs to be consideredi+1Further range judgment is carried out;
let T be a positive integer, andi+1＞Vuthe method is divided into two parts: vu＜Vi+1≤TVuAnd TVu＜Vi+1，Vu＜Vi+1≤TVuThe search space representing the next step is very close to the minimum search space, and the second contraction parameter is respectively introduced into the two partial areas, namely the calculated amount FE of the current search areaiThe maximum value Φ, the execution of the algorithm is evaluated:
when V isu＜Vi+1＜TVuWhen, if FE isiIf phi is satisfied, the algorithm generates excessive computing resource loss for finding the optimal solution in the search range of the section close to the minimum search space, therefore, the algorithm should stop searching and save the result, otherwise, the algorithm runs;
when TV is runningu＜Vi+1When, if FE isiIf < phi is established, the algorithm is not converged and continues to operate, otherwise, the algorithm stops searching and stores the result;
if PEi＜Φ，Vi+1＞VuFrom the current valid point set NiMiddle extracted subset Gi
Gi＝{g|gi＞E(Ni)} (5)
(2.4) decision stop or go on to global optimum
If subset GiCan be effectively extracted, then V is showni+1Effective from Vi+1In calculating a new random appearance Ji+1And make it possible to
Figure GDA0002549261700000041
To extract the best Ni+1
Figure GDA0002549261700000042
Repeating the above steps to obtain the average value sequence ∈ of each iteration step0≤∈1≤…≤∈i≤∈i+1Not more than …, showing monotone rising,
Figure GDA0002549261700000051
wherein the content of the first and second substances,
Figure GDA0002549261700000052
a sufficient requirement for becoming a global optimum is that when i → ∞ is present, the domain active point set N is searched foriIs/are as follows
Figure GDA0002549261700000053
Value mi+1→∈i+1
Figure GDA0002549261700000054
The plane parameter analysis template in the step (3): we define
Figure GDA0002549261700000055
Reflecting the wall surface parameters for the room,
Figure GDA0002549261700000056
also sound source s and respective first order mirror image sound sources
Figure GDA0002549261700000057
The center point of (a) is,
Figure GDA0002549261700000058
wherein the content of the first and second substances,
Figure GDA0002549261700000059
snand sn+1Is the position coordinate of the speaker at different times of n and n +1, dn+1＝dn+dssinα,dn＝cτnC is the speed of sound, dsIs the speaker from position snTo a position sn+1α is the speaker from position snTo a position sn+1The steering of (2).
Since the speaker is from position snTo a position sn+1Distance d ofsAnd the turn α may be preset, and therefore,
Figure GDA00025492617000000510
conversion of problem into τnAnd (4) solving, namely, estimating the time delay of the direct sound from the sound source to the reflecting wall surface.
The acoustic mirror image model in the step (3): because the prior of the shape of the room is set, a geometric coordinate system of the room is set by taking one corner in the room as an origin,combining the geographic orientation (e.g. upper, lower, northeast, west, east, west) with the coordinate system to obtain the external normal vector n of each reflector l1, 2, 6. the exact value,
according to the acoustic mirror image principle, the sound source position s and the reflecting surface parameter can be obtained
Figure GDA00025492617000000511
And normal vector n of reflecting surfacelThe relationship is as follows:
Figure GDA00025492617000000512
the method for realizing the reconstruction of the indoor space geometric outline based on the acoustic mirror image model only takes the indoor space shape as prior, obtains the first-order mirror image sound source distribution based on the acoustic mirror image model by accurately positioning the position of the speaker, converts the outline reconstruction problem into the SRP-PHAT-SRC sound source positioning problem according to the geometric plane principle, and effectively and reliably realizes the reconstruction of the indoor space geometric outline. The problem of huge operation amount in a traditional SRP-PHAT sound source positioning algorithm is solved, and the reconstruction scheme has certain noise immunity, reverberation resistance and robustness by means of insensitivity of phase transformation in time delay estimation to the surrounding environment of a signal.
Drawings
FIG. 1 is a block diagram of a method for reconstructing an indoor spatial contour according to the present invention.
FIG. 2 is a plane parameter model in the example.
Fig. 3 is a schematic diagram of a geometric coordinate system of a room in the embodiment.
Detailed Description
The present invention will be described in further detail with reference to the following examples and drawings, but the present invention is not limited thereto.
Referring to fig. 1, the method for realizing indoor space geometric contour reconstruction based on an acoustic mirror image model of the present invention includes an acoustic receiving device, a phase transformation generalized cross-correlation analysis module, a sound source localization module, a plane parameter analysis module, a mid-plane construction module, and an acoustic mirror image model, and specifically includes the following steps:
(1) a movable M-array element star-shaped array is arranged at the central position of an indoor reflection wall surface to serve as a sound receiving device, and the voice signal of a single indoor speaker is a sound source signal.
(2) Carrying out sound source positioning based on controllable response power phase transformation and a region contraction method;
specifically, the sound arrival time TOA of a sound signal received by the M array elements on a direct path is obtained through a phase transformation generalized cross-correlation analysis module, the SRP-PHA-SRC algorithm of an SRP-PHAT-SRC sound source positioning module is carried out, the SRP-PHA is optimized, and the position coordinates of a speaker are given in an iterative mode;
the optimization comprises the following specific steps:
(2.1) initialization of spatial parameters of Sound Source
Figure GDA0002549261700000061
Is the controllable response power of the phase transformation weights,
Figure GDA0002549261700000062
are the possible sound source positions in the sound source space,
Figure GDA0002549261700000063
Figure GDA0002549261700000064
thus, the sound source position
Figure GDA0002549261700000065
Setting an initial search rectangular region volume V0Number of random points in region J0Order ∈0＝E(J0) Then there is
Figure GDA0002549261700000066
Wherein N is0Is used forThe effective point number of the next search area body;
(2.2) Sound Source search area contraction
For the object function (1) in the current search region volume ViCarrying out random search internally to obtain random sampling point JiAnd calculating the effective point Ni，Ni＜＜Ji. From NiDistribution determination of a new search area volume Vi+1And the boundary thereof
Figure GDA0002549261700000071
Figure GDA0002549261700000072
(2.3) shrinkage parameter evaluation
Setting a threshold value:
① volume of minimum search area Vu；
② current search area calculation amount (FE)i) Maximum value of (d): phi;
③ maximum number of iteration steps: i;
when V isi+1≤VuThen, the search space representing the next step is smaller than the volume V of the minimum search areauAt this moment, the algorithm stops searching and saves the result;
when V isi+1＞VuThen, the volume of the search space representing the next step is larger than the volume V of the minimum search areauAt this time, considering the influence of the computing resource loss on the algorithm efficiency, V needs to be consideredi+1Further range judgment is carried out;
let T be a positive integer, andi+1＞Vuthe method is divided into two parts: vu＜Vi+1≤TVuAnd TVu＜Vi+1，Vu＜Vi+1≤TVuThe search space representing the next step is very close to the minimum search space, and the second contraction parameter is respectively introduced into the two partial areas, namely the calculated amount FE of the current search areaiThe maximum value Φ, the execution of the algorithm is evaluated:
when V isu＜Vi+1＜TVuWhen, if FE isiIf phi is satisfied, the algorithm generates excessive computing resource loss for finding the optimal solution in the search range of the section close to the minimum search space, therefore, the algorithm should stop searching and save the result, otherwise, the algorithm runs;
when TV is runningu＜Vi+1When, if FE isiIf < phi is established, the algorithm is not converged and continues to operate, otherwise, the algorithm stops searching and stores the result;
if FEi＜Φ，Vi+1＞VuFrom the current valid point set NiMiddle extracted subset Gi
Gi＝{g|gi＞E(Ni)} (5)
(2.4) decision stop or go on to global optimum
If subset GiCan be effectively extracted, then V is showni+1Effective from Vi+1In calculating a new random appearance Ji+1And make it possible to
Figure GDA0002549261700000081
To extract the best Ni+1
Figure GDA0002549261700000082
Repeating the above steps to obtain the average value sequence ∈ of each iteration step0≤∈1≤…≤∈i≤∈i+1Not more than …, showing monotone rising,
Figure GDA0002549261700000083
wherein the content of the first and second substances,
Figure GDA0002549261700000084
a sufficient requirement for becoming a global optimum is that when i → ∞ is present, the domain active point set N is searched foriIs/are as follows
Figure GDA0002549261700000085
Value mi+1→∈i+1
Figure GDA0002549261700000086
(3) Based on the acoustic mirror image model, obtaining estimated distance values of the speaker from indoor reflecting surfaces before and after single-step movement through cross-correlation analysis of sound transmitting and receiving signals;
specifically, the receiving array is moved to the central positions of different reflecting surfaces, the TOA minimum value of M array elements is taken as the direct distance estimation from the sound source to the reflecting surface, and the TOA minimum value is taken as the input of a plane parameter analysis module to obtain the reflecting parameters of the reflecting surface
Figure GDA0002549261700000087
The position coordinates s and the parameters of the reflecting surface of the sound source
Figure GDA0002549261700000088
And normal vector n of reflecting surface l1, 2., 6 as input to the acoustic mirror model, the first order mirror sound source positions are estimated
Figure GDA0002549261700000089
The planar parameter analysis template is defined firstly
Figure GDA00025492617000000810
Reflecting the wall parameters for the room, as shown in figure 2,
Figure GDA00025492617000000811
also sound source s and respective first order mirror image sound sources
Figure GDA00025492617000000812
The center point of (a) is,
Figure GDA00025492617000000813
wherein the content of the first and second substances,
Figure GDA00025492617000000814
snand sn+1Is the position coordinate of the speaker at different times of n and n +1, dn+1＝dn+dssinα,dn＝cτnAnd/2, c is the speed of sound.
(4) The reconstruction of the indoor space reflecting surface is quickly and effectively realized by solving the position of the mirror image sound source;
specifically, a sound source position s and each first-order mirror image sound source position are constructed through a middle vertical plane construction module
Figure GDA0002549261700000091
And obtaining the indoor geometric outline reconstruction result.
(4) Combining the shape prior of the indoor space, converting the spatial geometric contour reconstruction problem into a sound source positioning problem, and quickly and effectively realizing the reconstruction of the indoor spatial geometric contour by solving the position of the mirror image sound source;
specifically, a sound source position s and each first-order mirror image sound source position are constructed through a middle vertical plane construction module
Figure GDA0002549261700000092
Obtaining an indoor geometric outline reconstruction result;
the acoustic mirror model sets a room geometric coordinate system with a corner in the room as an origin due to the fact that the room shape is set a priori, as shown in fig. 3. Combining the geographic orientation (e.g. upper, lower, northeast, west, east, west) with the coordinate system to obtain the external normal vector n of each reflector l1, 2, 6. the value is known;
according to the acoustic mirror image principle, the sound source position s and the reflecting surface parameter can be obtained
Figure GDA0002549261700000093
And normal vector n of reflecting surfacelThe relationship is as follows:
Figure GDA0002549261700000094

Claims (3)

1. the method for realizing indoor space geometric outline reconstruction based on the acoustic mirror image model is characterized by comprising an acoustic receiving device, a phase transformation generalized cross-correlation analysis module, a sound source positioning module, a plane parameter analysis module, a middle vertical plane construction module and the acoustic mirror image model, and specifically comprises the following steps:
(1) a movable M array element star-shaped array is placed at the central position of an indoor reflection wall surface to serve as a sound receiving device, and the voice signal of a single indoor speaker is a sound source signal;
(2) carrying out sound source positioning based on controllable response power phase transformation and a region contraction method;
specifically, the sound arrival time TOA of a sound signal received by an M array element on a direct path is obtained through a phase transformation generalized cross-correlation analysis module, the SRP-PHA-SRC algorithm of an SRP-PHAT-SRC sound source positioning module is carried out, the SRP-PHA is optimized, and a speaker position coordinate s is given in an iterative mode;
the optimization comprises the following specific steps:
(2.1) initialization of spatial parameters of Sound Source
s is the position of the sound source,
Figure FDA0002549261690000011
are the possible sound source positions in the sound source space,
Figure FDA0002549261690000012
is the controllable response power of the phase transformation weights,
Figure FDA0002549261690000013
Figure FDA0002549261690000014
thus, the sound source position
Figure FDA0002549261690000015
Setting a spatial initial search rectangular region volume V0Number of random points in region J0Order ∈0＝E(J0) Then there is
Figure FDA0002549261690000016
Wherein N is0Is the number of valid points for the next search area volume;
(2.2) Sound Source search area contraction
For the object function in the current search region volume ViCarrying out random search internally to obtain random sampling point JiAnd calculating the effective point Ni，Ni＜＜JiFrom NiDistribution determination of a new search area volume Vi+1And the boundary thereof
Figure FDA0002549261690000017
Figure FDA0002549261690000018
(2.3) shrinkage parameter evaluation
Setting a threshold value:
① volume of minimum search area Vu；
② calculation amount FE of current search areaiMaximum value of (d): phi;
③ maximum number of iteration steps: i;
when V isi+1≤VuThen, the search space representing the next step is smaller than the volume V of the minimum search areauAt this moment, the algorithm stops searching and saves the result;
when V isi+1＞VuThen, the volume of the search space representing the next step is larger than the volume V of the minimum search areauAt this time, the calculation is performed in consideration of the consumption of the calculation resourcesInfluence of process efficiency, on Vi+1Further range judgment is carried out;
let T be a positive integer, andi+1＞Vuthe method is divided into two parts: vu＜Vi+1≤TVuAnd TVu＜Vi+1，Vu＜Vi+1≤TVuThe search space representing the next step is very close to the minimum search space, and the second contraction parameter is respectively introduced into the two partial areas, namely the calculated amount FE of the current search areaiThe maximum value Φ, the execution of the algorithm is evaluated:
when V isu＜Vi+1＜TVuWhen, if FE isiIf phi is satisfied, the algorithm generates excessive computing resource loss for finding the optimal solution in the search range of the section close to the minimum search space, therefore, the algorithm should stop searching and save the result, otherwise, the algorithm runs;
when TV is runningu＜Vi+1When, if FE isiIf < phi is established, the algorithm is not converged and continues to operate, otherwise, the algorithm stops searching and stores the result;
if FEi＜Φ，Vi+1＞VuFrom the current valid point set NiMiddle extracted subset Gi
Gi＝{g|gi＞E(Ni)} (5)
(2.4) decision stop or go on to global optimum
If subset GiCan be effectively extracted, then V is showni+1Effective from Vi+1In calculating a new random appearance Ji+1And make it possible to
Figure FDA0002549261690000021
To extract the best Ni+1
Figure FDA0002549261690000022
Repeating the above steps to obtain the average value sequence of each iteration step∈0≤∈1≤…≤∈i≤∈i+1Not more than …, showing monotone rising,
Figure FDA0002549261690000031
wherein the content of the first and second substances,
Figure FDA0002549261690000032
a sufficient requirement for becoming a global optimum is that when i → ∞ is present, the domain active point set N is searched foriIs/are as follows
Figure FDA0002549261690000033
Value of
mi+1→∈i+1
Figure FDA0002549261690000034
(3) Based on the acoustic mirror image model, obtaining estimated distance values of the speaker from indoor reflecting surfaces before and after single-step movement through cross-correlation analysis of sound transmitting and receiving signals;
specifically, the receiving array is moved to the central positions of different reflecting surfaces, the TOA minimum value of M array elements is taken as the direct distance estimation from the sound source to the reflecting surface, and the TOA minimum value is taken as the input of a plane parameter analysis module to obtain the reflecting parameter p of the reflecting surfacel1, 2, ·, 6; the position coordinates s and the reflection surface parameters p of the sound source are calculatedlAnd normal vector n of reflecting surfacel1, 2., 6 as input to the acoustic mirror model, the first order mirror sound source positions are estimated
Figure FDA0002549261690000035
(4) Combining the shape prior of the indoor space, converting the spatial geometric contour reconstruction problem into a sound source positioning problem, and quickly and effectively realizing the reconstruction of the indoor spatial geometric contour by solving the position of the mirror image sound source;
concretely, it isA cross-middle vertical plane construction module for constructing a sound source position s and first-order mirror image sound source positions
Figure FDA0002549261690000036
And obtaining the indoor geometric outline reconstruction result.
2. The method of claim 1, further comprising: the planar parameter analysis template in the step (3) is defined by firstly defining pl1, 2, 6 is a room reflection wall parameter, plAlso sound source s and respective first order mirror image sound sources
Figure FDA0002549261690000037
The center point of (a) is,
Figure FDA0002549261690000038
wherein the content of the first and second substances,
Figure FDA0002549261690000039
snand sn+1Is the position coordinate of the speaker at different times of n and n +1, dn+1＝dn+dssinα,dn＝cτnAnd/2, c is the speed of sound.
3. The method of claim 1, further comprising: setting a room geometric coordinate system by taking a corner in the room as an origin due to the fact that the room shape is set in the acoustic mirror image model in the step (3), and combining the geographic orientation with the coordinate system to obtain an external normal vector n of each reflecting surfacel1, 2, 6. the value is known;
according to the acoustic mirror image principle, the sound source position s and the reflecting surface parameter p can be obtainedlAnd normal vector n of reflecting surfacelThe relationship is as follows:
Figure FDA00025492616900000310
CN201810405623.5A 2018-04-29 2018-04-29 Indoor space geometric outline reconstruction method based on acoustic mirror image model Active CN108872939B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN201810405623.5A CN108872939B (en) 2018-04-29 2018-04-29 Indoor space geometric outline reconstruction method based on acoustic mirror image model

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN201810405623.5A CN108872939B (en) 2018-04-29 2018-04-29 Indoor space geometric outline reconstruction method based on acoustic mirror image model

Publications (2)

Publication Number Publication Date
CN108872939A CN108872939A (en) 2018-11-23
CN108872939B true CN108872939B (en) 2020-09-29

Family

ID=64327183

Family Applications (1)

Application Number Title Priority Date Filing Date
CN201810405623.5A Active CN108872939B (en) 2018-04-29 2018-04-29 Indoor space geometric outline reconstruction method based on acoustic mirror image model

Country Status (1)

Country Link
CN (1) CN108872939B (en)

Families Citing this family (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN109901112B (en) * 2019-03-29 2022-10-04 桂林电子科技大学 Acoustic simultaneous positioning and mapping method based on multi-channel sound acquisition
CN111157952B (en) * 2020-01-06 2022-04-01 北京大学 Room boundary estimation method based on mobile microphone array

Citations (12)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US5224382A (en) * 1990-07-12 1993-07-06 Olympus Optical Co., Ltd. Transmission type scanning acoustic microscope
CN102074236A (en) * 2010-11-29 2011-05-25 清华大学 Speaker clustering method for distributed microphone
CN102707262A (en) * 2012-06-20 2012-10-03 太仓博天网络科技有限公司 Sound localization system based on microphone array
CN103426434A (en) * 2012-05-04 2013-12-04 索尼电脑娱乐公司 Source separation by independent component analysis in conjunction with source direction information
CN103813260A (en) * 2014-01-28 2014-05-21 哈尔滨工业大学 Mirror image source method based long spacing reverberation time obtaining method
CN104142492A (en) * 2014-07-29 2014-11-12 佛山科学技术学院 SRP-PHAT multi-source spatial positioning method
CN105976827A (en) * 2016-05-26 2016-09-28 南京邮电大学 Integrated-learning-based indoor sound source positioning method
CN106093864A (en) * 2016-06-03 2016-11-09 清华大学 A kind of microphone array sound source space real-time location method
CN107271963A (en) * 2017-06-22 2017-10-20 广东美的制冷设备有限公司 The method and apparatus and air conditioner of auditory localization
CN107329114A (en) * 2017-06-21 2017-11-07 歌尔股份有限公司 Sound localization method and device
CN107621625A (en) * 2017-06-23 2018-01-23 桂林电子科技大学 Sound localization method based on double micro-microphone battle arrays
CN107872752A (en) * 2016-09-28 2018-04-03 本田技研工业株式会社 Acoustic characteristic bearing calibration, acoustic characteristic means for correcting and FFT loops

Patent Citations (12)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US5224382A (en) * 1990-07-12 1993-07-06 Olympus Optical Co., Ltd. Transmission type scanning acoustic microscope
CN102074236A (en) * 2010-11-29 2011-05-25 清华大学 Speaker clustering method for distributed microphone
CN103426434A (en) * 2012-05-04 2013-12-04 索尼电脑娱乐公司 Source separation by independent component analysis in conjunction with source direction information
CN102707262A (en) * 2012-06-20 2012-10-03 太仓博天网络科技有限公司 Sound localization system based on microphone array
CN103813260A (en) * 2014-01-28 2014-05-21 哈尔滨工业大学 Mirror image source method based long spacing reverberation time obtaining method
CN104142492A (en) * 2014-07-29 2014-11-12 佛山科学技术学院 SRP-PHAT multi-source spatial positioning method
CN105976827A (en) * 2016-05-26 2016-09-28 南京邮电大学 Integrated-learning-based indoor sound source positioning method
CN106093864A (en) * 2016-06-03 2016-11-09 清华大学 A kind of microphone array sound source space real-time location method
CN107872752A (en) * 2016-09-28 2018-04-03 本田技研工业株式会社 Acoustic characteristic bearing calibration, acoustic characteristic means for correcting and FFT loops
CN107329114A (en) * 2017-06-21 2017-11-07 歌尔股份有限公司 Sound localization method and device
CN107271963A (en) * 2017-06-22 2017-10-20 广东美的制冷设备有限公司 The method and apparatus and air conditioner of auditory localization
CN107621625A (en) * 2017-06-23 2018-01-23 桂林电子科技大学 Sound localization method based on double micro-microphone battle arrays

Non-Patent Citations (2)

* Cited by examiner, † Cited by third party
Title
A 3D MODEL FOR ROOM BOUNDARY ESTIMATION;Luca Remaggi et al.;《2015 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP)》;20150424;全文 *
基于子带可控响应功率的多声源定位方法;倪志莲 等;《计算机工程与应用》;20120816;全文 *

Also Published As

Publication number Publication date
CN108872939A (en) 2018-11-23

Similar Documents

Publication Publication Date Title
Kreković et al. EchoSLAM: Simultaneous localization and mapping with acoustic echoes
CN106093864B (en) A kind of microphone array sound source space real-time location method
Yook et al. Fast sound source localization using two-level search space clustering
CN109901112B (en) Acoustic simultaneous positioning and mapping method based on multi-channel sound acquisition
CN104142492A (en) SRP-PHAT multi-source spatial positioning method
CN108828501B (en) Method for real-time tracking and positioning of mobile sound source in indoor sound field environment
CN109188362B (en) Microphone array sound source positioning signal processing method
Shi et al. A TDOA technique with super-resolution based on the volume cross-correlation function
Perrodin et al. Design and calibration of large microphone arrays for robotic applications
CN108872939B (en) Indoor space geometric outline reconstruction method based on acoustic mirror image model
CN105607042A (en) Method for locating sound source through microphone array time delay estimation
CN109212481A (en) A method of auditory localization is carried out using microphone array
Traa et al. Robust source localization and enhancement with a probabilistic steered response power model
CN109696657A (en) A kind of coherent sound sources localization method based on vector hydrophone
Rosseel et al. Improved acoustic source localization by time delay estimation with subsample accuracy
KR20090128221A (en) Method for sound source localization and system thereof
Le et al. Robust TDOA-based joint source and microphone localization in a reverberant environment using medians of acceptable recovered TOAs
CN103778288A (en) Ant colony optimization-based near field sound source localization method under non-uniform array noise condition
Jyothi et al. SOLVIT: A reference-free source localization technique using majorization minimization
CN109600711B (en) Indoor positioning method based on channel response frequency domain and spatial domain combined processing
Zhong et al. Spherical convolutional recurrent neural network for real-time sound source tracking
Tachikawa et al. Sound source localization based on sparse estimation and convex clustering
Dang et al. Multiple sound source localization based on a multi-dimensional assignment model
Chen et al. A sound source localization device based on rectangular pyramid structure for mobile robot
Bu et al. TDOA estimation of speech source in noisy reverberant environments

Legal Events

Date Code Title Description
PB01 Publication
PB01 Publication
SE01 Entry into force of request for substantive examination
SE01 Entry into force of request for substantive examination
GR01 Patent grant
GR01 Patent grant
EE01 Entry into force of recordation of patent licensing contract
EE01 Entry into force of recordation of patent licensing contract

Application publication date: 20181123

Assignee: Wuhan xingeno Technology Co.,Ltd.

Assignor: GUILIN University OF ELECTRONIC TECHNOLOGY

Contract record no.: X2022450000387

Denomination of invention: Reconstruction method of interior space geometric contour based on acoustic mirror model

Granted publication date: 20200929

License type: Common License

Record date: 20221226