A method for realizing dynamic high spatiotemporal resolution noise map

Abstract

The present application relates to the field of electronic digital data processing technology, and in particular to a method for realizing a dynamic high-temporal and spatial resolution noise map. The method comprises: S100, obtaining a noise audio sequence A uploaded by a target vehicle, wherein A m is the noise audio corresponding to the mth acquisition time period; S200, traverse A, obtain A m The corresponding noise characteristic B m ; S300, according to B m Get A m The corresponding location label set C m , C m Middle C m,i(m) A m The corresponding i(m)th location tag; S400, obtain the location tag set D of each geographic hash grid in the target area, D in D n is the location tag set of the nth geo-hash grid in the target area; S500, according to C m and D determine the track of the target vehicle; S600, displaying preset information of noise audio corresponding to the noise audio collection position on the track on a map of the target area. The present invention can generate a dynamic high-temporal and spatial resolution noise map.

Description

Technical Field

[0001] The invention relates to the technical field of electrical digital data processing, and in particular to a method for realizing a dynamic high-temporal and spatial resolution noise map. Background Art

[0002] In the prior art, the positioning of the vehicle is usually achieved based on the GPS positioning system, and then the trajectory of the vehicle is obtained. However, in some scenarios, the vehicle is not equipped with a GPS positioning system. The prior art discloses a technical solution for realizing high-accuracy recognition of sound based on a neural network model. For example, a Chinese patent application with application publication number CN118098270A discloses a noise source tracing method based on feature extraction and feature fusion. The method uses the spectrogram features, mel-spectrogram features, spectrogram integration features and mel-spectrogram integration features corresponding to the audio to train the neural network model, and analyzes the excessive noise based on the trained model to achieve noise source tracing. The audio corresponding to different locations passed by the vehicle during driving is different. How to generate a dynamic high-temporal and spatial resolution noise map based on the external noise audio data collected by the vehicle during driving is an urgent problem to be solved. Summary of the invention

[0003] The object of the present invention is to provide a method for realizing a dynamic high spatiotemporal resolution noise map, so as to generate a dynamic high spatiotemporal resolution noise map based on external noise audio data collected during the driving of a vehicle.

[0004] According to the present invention, a method for realizing a dynamic high spatiotemporal resolution noise map is provided, the method comprising the following steps:

[0005] S100, obtaining the noise audio sequence A uploaded by the target vehicle, A=(A 1 ,A 2 ,…,A m ,…,A M ), A m is the mth collection time period T m The corresponding noise audio, m ranges from 1 to M, M is the number of acquisition time periods; T m =[t m -Δt,t m +Δt],t m is the mth acquisition time point, Δt is the initial acquisition duration, and the time interval between any two adjacent acquisition time points is ΔT.

[0006] S200, traverse A, get A m The corresponding noise characteristic B m .

[0007] S300, according to B m Get Am The corresponding location label set C m , C m =(C m,1 ,C m,2 ,…,C m,i(m) ,…,C m,u(m) ), C m,i(m) A m The corresponding i(m)th position label, i(m) ranges from 1 to u(m), where u(m) is A m The number of corresponding position labels.

[0008] S400, obtaining a location tag set D of each geographic hash grid in the target area, D = (D 1 ,D 2 ,…,D n ,…,D N ), D n is the nth geohash grid g in the target area n The location tag set of n ranges from 1 to N, where N is the number of geo-hash grids in the target area; D n =(D n,1 ,D n,2 ,…,D n,j(n) ,…,D n,p(n) ), D n,j(n) g n The j(n)th position label of , the value range of j(n) is 1 to p(n), and p(n) is g n The target vehicle travels within the target area in each collection time period.

[0009] S500, according to C m and D determine the trajectory of the target vehicle.

[0010] S600: Displaying preset information of noise audio corresponding to the noise audio collection position on the track on a map of the target area.

[0011] Compared with the prior art, the present invention has at least the following beneficial effects:

[0012] The present invention can generate a dynamic high-temporal-space resolution noise map based on a noise audio sequence uploaded by a target vehicle. Specifically, the noise audio sequence includes noise audios corresponding to several acquisition time periods. Based on the noise audios corresponding to each acquisition time period, the corresponding noise features can be obtained. Based on the noise features corresponding to each acquisition time period, the location tag set corresponding to the corresponding acquisition time period can be obtained. The location tag set corresponding to one acquisition time period includes at least one location tag. The location tag is used to indicate the location of the target vehicle in the acquisition time period. The location tag is used to indicate the type of location. For example, the location tag includes a school, a hospital, a residential area, a river, etc. The present invention also obtains the location tag set of each geographic hash grid in the target area. The location tag set of each geographic hash grid includes at least one location tag. Based on the location tag set corresponding to each acquisition time period and the location tag set corresponding to each geographic hash grid in the target area, the trajectory of the target vehicle in the target area can be determined. On this basis, the preset information of the noise audio corresponding to the noise audio acquisition position on the trajectory is displayed on the map of the target area, so that a dynamic high-temporal-space resolution noise map can be generated. BRIEF DESCRIPTION OF THE DRAWINGS

[0013] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the following briefly introduces the drawings required for use in the description of the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.

[0014] Figure 1 A flowchart of a method for implementing a dynamic high spatiotemporal resolution noise map provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0015] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative work are within the scope of protection of the present invention.

[0016] According to this embodiment, Figure 1 As shown, a method for realizing a dynamic high spatiotemporal resolution noise map is provided, the method comprising the following steps:

[0017] S100, obtaining the noise audio sequence A uploaded by the target vehicle, A=(A 1 ,A 2 ,…,A m ,…,A M ), Am is the mth collection time period T m The corresponding noise audio, m ranges from 1 to M, M is the number of acquisition time periods; T m =[t m -Δt,t m +Δt],t m is the mth acquisition time point, Δt is the initial acquisition duration, and the time interval between any two adjacent acquisition time points is ΔT.

[0018] In this embodiment, a sound collection device is used to obtain external sounds during vehicle driving, and the external sounds during vehicle driving are also noise audio. Those skilled in the art know that any sound collection device in the prior art falls within the protection scope of the present invention.

[0019] In this embodiment, Δt and ΔT are both empirical values, and Δt≤ΔT / 2.

[0020] S200, traverse A, get A m The corresponding noise characteristic B m .

[0021] Optionally, the noise feature is a Mel-spectrogram feature. Those skilled in the art know that any method for obtaining a Mel-spectrogram feature corresponding to an audio segment in the prior art falls within the protection scope of the present invention.

[0022] S300, according to B m Get A m The corresponding location label set C m , C m =(C m,1 ,C m,2 ,…,C m,i(m) ,…,C m,u(m) ), C m,i(m) A m The corresponding i(m)th position label, i(m) ranges from 1 to u(m), where u(m) is A m The number of corresponding position labels.

[0023] In this embodiment, A m The corresponding location tag is the target vehicle in the mth collection time period T mThe label of the location, the label of the location is used to indicate the type of location, for example, the labels of the location include school, hospital, residential area, river, etc. The noise characteristics of the noise audio corresponding to different location labels are different, and the noise characteristics of the noise audio corresponding to the same location label are similar; when the types of two locations passed by the target vehicle are different, the noise characteristics of the noise audio uploaded by the target vehicle are also different; when the types of two locations passed by the target vehicle are the same, the noise characteristics of the noise audio uploaded by the target vehicle are also similar; this embodiment can be based on the mth collection time period T uploaded by the target vehicle m Noise Audio A m The noise characteristics of B m Get A m The corresponding location tag set, as an optional specific implementation, S300 includes:

[0024] S310, B m Input the trained target neural network model and obtain the output result L of the trained target neural network model m , L m =(L m,1 ,L m,2 ,…,L m,r ,…,L m,R ), L m,r B is the output of the trained target neural network m The probability corresponding to the rth preset position label, the value range of r is 1 to R, and R is the number of preset position labels; the trained target neural network model is used to obtain the probability that the input noise feature corresponds to each preset position label.

[0025] In this embodiment, the task performed by the target neural network model is a classification task. The output layer of the target neural network model uses the softmax function as the activation function. The output of the trained target neural network model is the probability corresponding to R preset position labels, and the sum of the probabilities corresponding to the R preset position labels is 1.

[0026] In this embodiment, L m,r The larger the value is, the more likely the target vehicle is in the mth collection time period T m The more likely the location is to be the type of location corresponding to the rth preset location tag; for example, the type of location corresponding to the rth preset location tag is a school, then L m,r The larger the value is, the more likely the target vehicle is in the mth collection time period T m The more likely the location is a school; as an optional specific implementation, the training process of the target neural network model includes:

[0027] S311, obtaining a noise feature sample set, where the noise feature sample set includes a plurality of noise feature samples.

[0028] Optionally, each noise feature sample is a noise feature corresponding to noise audio uploaded by other vehicles, and the duration of the noise audio corresponding to each noise feature sample is equal or unequal.

[0029] S312, obtaining a noise feature label set, wherein the noise feature label set includes a position label set for each noise feature sample in the noise feature sample set, and the position label set for each noise feature sample includes a plurality of position labels.

[0030] In this embodiment, the location label set of each noise feature sample includes at least one location label. If the location label set of a noise feature sample includes more than two location labels, it means that there are more than two types of locations for obtaining the noise audio corresponding to the noise feature sample. For example, the location label set of a noise feature sample includes two location labels, namely a location label corresponding to a school and a location label corresponding to a residential area. Then, the locations for obtaining the noise audio corresponding to the noise feature sample include both schools and residential areas.

[0031] Optionally, a manual labeling method is used to obtain a position label set for each noise feature sample.

[0032] S313: Use the noise feature sample set and the noise feature label set to train the target neural network.

[0033] Those skilled in the art know that any supervised training method with labels in the prior art falls within the protection scope of the present invention. The trained target neural network model has the function of obtaining the probability that the input noise feature corresponds to each preset position label.

[0034] S320, traverse L m , if L m,r ≥e 2 , then L m,r The corresponding preset position labels are added to the preset initial label set H, which is initialized to an empty set; e 2 is the preset probability threshold.

[0035] In this embodiment, e 2 is the experience value, 0 <e 2 <0.5.

[0036] In this embodiment, after traversing L, H includes all the 2 L m,r The corresponding preset position label.

[0037] S330, if H is not an empty set, determine H to be C m .

[0038] In this embodiment, if H is an empty set, L m The set of preset position labels corresponding to the first ζ probability with the largest probability is determined as C m , ζ is the preset number of tags, optional, ζ is 1 or 2 or 3.

[0039] S400, obtaining a location tag set D of each geographic hash grid in the target area, D = (D 1 ,D 2 ,…,D n ,…,D N ), D n is the nth geohash grid g in the target area n The location tag set of n ranges from 1 to N, where N is the number of geo-hash grids in the target area; D n =(D n,1 ,D n,2 ,…,D n,j(n) ,…,D n,p(n) ), D n,j(n) g n The j(n)th position label of , the value range of j(n) is 1 to p(n), and p(n) is g n The target vehicle travels within the target area in each collection time period.

[0040] Those skilled in the art know that the number of bits of a geohash grid (Geohash) is related to the size of the geographic area represented by each geohash grid; in this embodiment, the number of bits of a geohash grid is an empirical value, and optionally, the number of bits of a geohash grid (Geohash) is 6, 7 or 8.

[0041] S500, according to C m and D determine the trajectory of the target vehicle.

[0042] In this embodiment, C m and one of D n The more similar the corresponding location labels are, the more likely the target vehicle is in the mth collection time period T m The more likely the location is to be the nth geohash grid g in the target area. n As a specific implementation, S500 includes:

[0043] S510, traverse D, obtain D n With C 1 The similarity f n,1 , f n,1 =sn,1 / p(n),s n,1 C 1 Appears in D n The number of position labels in .

[0044] In this embodiment, if C 1 No position label appears in D n In, then s n,1 =0; if C 1 Only one position label appears in D n In, then s n,1 =1; if C 1 Only 2 position labels appear in D n In, then s n,1 =2; and so on.

[0045] In this embodiment, f n,1 The larger the value of g n The target vehicle in the first collection time period T 1 The greater the probability of the location.

[0046] S520, if f n,1 ≥e 1 , then determine g n is the suspected first position of the target vehicle, and g n Append to the suspected first position set G 1 , and get G 1 =(G 1,1 ,G 1,2 ,…,G 1,q ,…,G 1,Q ), G 1,q To be added to G 1 The qth geohash grid of G, where q ranges from 1 to Q, and Q is the value appended to G 1 The number of geo-hash grids; G 1 is initialized to an empty set, e 1 is the first preset similarity threshold, 0 <e 1 <1.

[0047] In this embodiment, e 1 For experience value, optional, e 1 is 0.5.

[0048] S530, traverse G 1 , get G 1,q The corresponding first geographic hash grid set H 1,q , H 1,q =(H 1 1,q ,H 21,q ,…,H k (q) 1,q ,…,H p(q) 1,q ), H k(q) 1,q G 1,q The corresponding k(q)th first geographic hash grid, k(q) ranges from 1 to p(q), p(q) is G 1,q The number of the corresponding first geo-hash grid; G 1,q The corresponding first geographic hash grid is the first geographic hash grid in the target area corresponding to G 1,q The distance falls into the first preset distance interval dis 1 The geo-hash grid, dis 1 =[ΔT×v q,min ,ΔT×v q,max ],v q,min The default G 1,q The corresponding minimum speed, v q,max The default G 1,q The corresponding maximum speed.

[0049] In this embodiment, the distance between two geo-hash grids is not the straight-line distance between the centers of the two geo-hash grids, but the distance along the extension direction of the roads in the geo-hash grids, that is, the distance from the center of one geo-hash grid to the center of the other geo-hash grid when the target vehicle travels along the extension direction of the roads in the geo-hash grids.

[0050] In this embodiment, the minimum speeds corresponding to different geographic hash grids are the same or different, and the maximum speeds corresponding to different geographic hash grids are the same or different. Optionally, the minimum speed and maximum speed corresponding to a geographic hash grid are obtained based on experience or based on the speed limit information of the roads in the geographic hash grid.

[0051] S540, traverse H 1,q , get H k(q) 1,q With C 2 The similarity f k(q) q,2 , f k(q) q,2 =qua k(q) 1,q / num k(q) 1,q ,qua k(q) 1,q C 2 Appears in H k(q) 1,q The number of position labels in , num k(q)1,q H k(q) 1,q The number of location tags.

[0052] S550, if f k(q) q,2 ≥e 1 , then determine H k(q) 1,q is the suspected second position of the target vehicle; otherwise, it is determined that H k (q) 1,q The second location is not the target vehicle.

[0053] In this embodiment, H 1,q There may be more than one H k(q) 1,q The second suspected position of the target vehicle, H 1,q H may not exist in k(q) 1,q The suspected second position of the target vehicle is determined as follows; as a specific implementation, S550 further includes:

[0054] S560, if H 1,q Each H k(q) 1,q If both are determined not to be the second position of the target vehicle, it is determined that G 1,q is not the first position of the target vehicle; if H 1,q There is H k(q) 1,q If it is determined to be the suspected second position of the target vehicle, the process proceeds to S570.

[0055] S570, get H 1,q The geographical hash grid set H of the suspected second location determined to be the target vehicle 2,q , H 2,q =(H 1 2,q ,H 2 2,q ,…,H x(q) 2,q ,…,H y(q) 2,q ), H x(q) 2,q H 1,q The x(q)th geohash grid that is determined to be the suspected second location of the target vehicle, where the value of x(q) ranges from 1 to y(q), and y(q) is H 1,q The number of geo-hash grids determined to be the suspected second location of the target vehicle.

[0056] S580, traverse H 2,q, get H x(q) 2,q The corresponding first geographical hash grid set, if H x(q) 2,q If each hash grid in the corresponding first geographic hash grid set is determined to be not the third position of the target vehicle, then H x(q) 2,q The second position that is not the target vehicle; H x(q) 2,q The corresponding first geographical hash grid set is the set of nodes in the target area that are related to H x(q) 2,q The distance falls into the second preset distance interval dis 2 The geo-hash grid, dis 2 =[ΔT×v x(q) q,min ,ΔT×v x(q) q,max ],v x (q) q,min The default H x(q) 2,q The corresponding minimum speed, v x(q) q,max The default H x(q) 2,q The corresponding maximum speed.

[0057] In this embodiment, it is determined that H x(q) 2,q The method for determining whether a certain geographic hash grid in the corresponding first geographic hash grid set is the suspected third location of the target vehicle is the same as the above-mentioned method for determining whether a certain geographic hash grid in the first geographic hash grid set is the suspected third location of the target vehicle. 1,q Medium H k(q) 1,q The method of determining whether it is the suspected second position of the target vehicle is similar, that is, obtaining H x(q) 2,q The location label set of the geo-hash grid in the corresponding first geo-hash grid set is the same as C 3 If the similarity is greater than or equal to e 1 , then determine H x(q) 2,q The corresponding first geographic hash grid set determines that the geographic hash grid is the suspected third location of the target vehicle; otherwise, it is determined that H x(q) 2,q The geo-hash grid in the corresponding first geo-hash grid set is not a third location of the target vehicle.

[0058] In this embodiment, according to whether the similarity is greater than or equal to e 1 The method can only determine H x(q) 2,qSome of the corresponding first geo-hash grids are not the third location of the target vehicle, even if the similarity is greater than or equal to e 1 The method of H x(q) 2,q Some of the corresponding first geographic hash grid sets are tentatively determined as the suspected third location of the target vehicle. These geographic hash grids are not necessarily the suspected third location of the target vehicle. It is necessary to make a judgment based on the first geographic hash grid set corresponding to the geographic hash grid determined as the third location in a different place; for example, based on whether the above similarity is greater than or equal to e 1 The method of H x(q) 2,q If a certain geo-hash grid in the corresponding first geo-hash grid set is determined to be the suspected third position of the target vehicle, but all the geo-hash grids in the first geo-hash grid set corresponding to the geo-hash grid determined to be the suspected third position are determined not to be the suspected fourth position of the target vehicle, then the geo-hash grid determined to be the suspected third position will also be determined not to be the third position of the target vehicle; only if there is a geo-hash grid determined to be the suspected fourth position of the target vehicle in the first geo-hash grid set corresponding to the geo-hash grid determined to be the suspected third position, then the geo-hash grid determined to be the suspected third position will be determined as the suspected third position of the target vehicle.

[0059] S590, if H 2,q Each H x(q) 2,q If both are determined not to be the second position of the target vehicle, it is determined that G 1,q Not the first position for the target vehicle.

[0060] In this embodiment, the mth position of the target vehicle refers to the position of the target vehicle in the mth collection time period T m If all the geo-hash grids in the first geo-hash grid set corresponding to a certain suspected m-th position of the target vehicle determined according to the similarity are determined not to be the m+1-th position of the target vehicle, then the suspected m-th position of the target vehicle determined according to the similarity is determined not to be the m-th position of the target vehicle. It should be understood that only when there is a geo-hash grid determined to be the suspected m+1-th position of the target vehicle in the first geo-hash grid set corresponding to a certain suspected m-th position of the target vehicle determined according to the similarity, the suspected m-th position of the target vehicle determined according to the similarity is determined to be the suspected m-th position of the target vehicle.

[0061] In this embodiment, in addition to determining whether the geo-hash grid is the suspected 1st and Mth positions of the target vehicle, the method for determining whether the geo-hash grid is the suspected 2nd to M-1st positions of the target vehicle includes three determination conditions. For example, the value range of λ is 2 to M-1, and the determination conditions for determining whether the geo-hash grid is the suspected λth position of the target vehicle are: (1) whether the geo-hash grid belongs to the first geo-hash grid set corresponding to a geo-hash grid determined to be the suspected λth position of the target vehicle; (2) whether the location label set corresponding to the geo-hash grid is consistent with C λ Is the similarity greater than or equal to e? 1 ; (3) Whether there is a geo-hash grid in the first geo-hash grid set corresponding to the geo-hash grid that is determined to be the suspected λ+1th position of the target vehicle. Only when the judgment results of the three judgment conditions are all yes, the geo-hash grid is finally determined to be the suspected λth position of the target vehicle. The method for determining whether the geo-hash grid is the suspected 1st position of the target vehicle only includes the above conditions (2) and (3), and the method for determining whether the geo-hash grid is the suspected Mth position of the target vehicle only includes the above conditions (1) and (2).

[0062] As a specific implementation manner, after S590, the following steps are further included:

[0063] S591: If all M suspected positions of the target vehicle exist, and there is a track in the target area that satisfies the first preset condition, the track is determined to be the track of the target vehicle; the first preset condition includes: the starting position of the track is the suspected first position of the target vehicle, the ending position of the track is the suspected Mth position of the target vehicle, the track passes through the suspected βth position of the target vehicle, and the value range of β is 2 to M-1; the distance between the suspected ηth position and the suspected η+1th position on the track falls within the preset distance interval corresponding to the suspected ηth position, and the preset distance interval corresponding to the suspected ηth position is [ΔT×vmin η ,ΔT×vmax η ],vmin η is the preset minimum speed corresponding to the suspected ηth position, vmax η is the maximum speed corresponding to the preset suspected ηth position, and the value range of η is 1 to M-1.

[0064] In this embodiment, the M suspected positions of the target vehicle are a suspected first position, a suspected second position, ..., a suspected mth position, ..., a suspected Mth position of the target vehicle.

[0065] In this embodiment, the trajectories within the target area are all trajectories along the extension direction of the road within the target area. The method for determining other suspected m positions of the target vehicle except the suspected first position is similar to the method for determining the suspected first position of the target vehicle mentioned above, and will not be repeated here.

[0066] In this embodiment, if there are more than two suspected m-th positions of the target vehicle, as long as a certain trajectory passes through one of the suspected m-th positions, it is determined that the trajectory passes through the suspected m-th position of the target vehicle.

[0067] S592, if all M suspected positions of the target vehicle exist, and there are more than two trajectories in the target area that meet the first preset condition, the trajectory with the greatest corresponding similarity is determined as the trajectory of the target vehicle; the similarity corresponding to the wth trajectory in the target area that meets the first preset condition is ∑ M m＝1 ε w,m , ε w,m The position label set of the suspected mth position of the target vehicle passing by the wth trajectory that meets the first preset condition and C m The similarity.

[0068] In this embodiment, the position label set of the suspected m-th position of the target vehicle is m The method for obtaining the similarity of f is the same as that for obtaining f n,1 The method is similar and will not be repeated here.

[0069] In this embodiment, if the M suspected positions of the target vehicle do not exist, the above ΔT is updated to 2×ΔT, that is, the noise audio in the noise audio sequence A uploaded by the target vehicle is sampled equidistantly, and A is updated to 2×ΔT. 1 As the starting point, sample one noise audio at intervals to form a new noise audio sequence, and perform steps similar to S200-S500; if the trajectory of the target vehicle cannot be determined, update the above ΔT to 3×ΔT, that is, sample the noise audio in the noise audio sequence A uploaded by the target vehicle at equal intervals, and replace A 1 As a starting point, sample 2 noise audios at intervals to form a new noise audio sequence, and execute steps similar to S200-S500; if the trajectory of the target vehicle cannot be determined, continue to update ΔT to 4×ΔT, and so on, until the trajectory of the target vehicle can be determined.

[0070] S600: Displaying preset information of noise audio corresponding to the noise audio collection position on the track on a map of the target area.

[0071] In this embodiment, when all M suspected positions of the target vehicle exist and there is one track in the target area that satisfies the first preset condition, the number of noise audio collection positions is M, and the mth noise audio collection position is the mth suspected position of the target vehicle. When the above ΔT is updated to 2×ΔT and the track of the target vehicle can be determined, only the preset information of the noise audio corresponding to the first, third, fifth, etc. noise audio collection positions on the track is displayed; when the above ΔT is updated to 3×ΔT and the track of the target vehicle can be determined, only the preset information of the noise audio corresponding to the first, fourth, seventh, etc. noise audio collection positions on the track is displayed; when the above ΔT is updated to 4×ΔT and the track of the target vehicle can be determined, only the preset information of the noise audio corresponding to the first, fifth, ninth, etc. noise audio collection positions on the track is displayed; and so on.

[0072] Optionally, the preset information includes average noise intensity and / or maximum noise intensity; preferably, the preset information also includes acquisition time.

[0073] This embodiment can determine the trajectory of the target vehicle in the target area based on the noise audio sequence uploaded by the target vehicle. Specifically, the noise audio sequence includes noise audio corresponding to several collection time periods. The corresponding noise characteristics can be obtained based on the noise audio corresponding to each collection time period. The location label set corresponding to the corresponding collection time period can be obtained based on the noise characteristics corresponding to each collection time period. The location label set corresponding to one collection time period includes at least one location label. The location label is used to indicate the label of the location of the target vehicle in the collection time period. The location label is used to indicate the type of location. For example, the location label includes a school, a hospital, a residential area, a river, etc.; the present invention also obtains the location label set of each geographic hash grid in the target area, and the location label set of each geographic hash grid includes at least one location label; based on the location label set corresponding to each collection time period and the location label set corresponding to each geographic hash grid in the target area, the trajectory of the target vehicle in the target area can be determined; on this basis, the preset information of the noise audio corresponding to the noise audio collection position on the trajectory is displayed on the map of the target area, so as to generate a dynamic high spatiotemporal resolution noise map.

[0074] Although some specific embodiments of the present invention have been described in detail by way of example, it will be appreciated by those skilled in the art that the above examples are for illustration only and are not intended to limit the scope of the present invention. It will also be appreciated by those skilled in the art that various modifications may be made to the embodiments without departing from the scope and spirit of the present invention. The scope of the present invention is defined by the appended claims.

Claims

1. A method for realizing a dynamic high spatiotemporal resolution noise map, characterized in that: The method comprises the following steps: S100, obtaining a noise audio sequence A uploaded by the target vehicle, A=(A1, A2, …, A m ,…,A M ), A m is the mth collection time period T m The corresponding noise audio, m ranges from 1 to M, M is the number of acquisition time periods; T m =[t m -Δt,t m +Δt],t m is the mth acquisition time point, Δt is the initial acquisition duration, and the time interval between any two adjacent acquisition time points is ΔT; S200, traverse A, get A m The corresponding noise characteristic B m ; S300, according to B m Get A m The corresponding location label set C m , C m =(C m,1 ,C m,2 ,…,C m,i(m) ,…,C m,u(m) ), C m,i(m) A m The corresponding i(m)th position label, i(m) ranges from 1 to u(m), where u(m) is A m The number of corresponding position tags; S400, obtaining a location tag set D of each geographic hash grid in the target area, D=(D1, D2, ..., D n ,…,D N ), D n is the nth geohash grid g in the target area n The location tag set of n ranges from 1 to N, where N is the number of geo-hash grids in the target area; D n =(D n,1 ,D n,2 ,…,D n,j(n) ,…,D n,p(n) ), D n,j(n) g n The j(n)th position label of , the value range of j(n) is 1 to p(n), and p(n) is g n The target vehicle is traveling in the target area in each collection time period; S500, according to C m and D determine the trajectory of the target vehicle; S600: Displaying preset information of noise audio corresponding to the noise audio collection position on the track on a map of the target area.

2. The method for realizing a dynamic high spatiotemporal resolution noise map according to claim 1, characterized in that: S500 includes: S510, traverse D, obtain D n Similarity f with C1 n,1 , f n,1 =s n,1 / p(n),s n,1 C1 appears in D n The number of position tags in ; S520, if f n,1 ≥e1, then judge g n is the suspected first position of the target vehicle, and g n Append to the suspected first position set G1, and get G1 = (G 1,1 ,G 1,2 ,…,G 1,q ,…,G 1,Q ), G 1,q is the qth geo-hash grid appended to G1, the value range of q is 1 to Q, Q is the number of geo-hash grids appended to G1; G1 is initialized to an empty set, e1 is the first preset similarity threshold, 0 <e1<1; S530, traverse G1, obtain G 1,q The corresponding first geographic hash grid set H 1,q , H 1,q =(H 1 1,q ,H 2 1,q ,…,H k (q) 1,q ,…,H p(q) 1,q ), H k(q) 1,q G 1,q The corresponding k(q)th first geographic hash grid, k(q) ranges from 1 to p(q), p(q) is G 1,q The number of the corresponding first geo-hash grid; G 1,q The corresponding first geographic hash grid is the first geographic hash grid in the target area corresponding to G 1,q The distance falls into the geographical hash grid of the first preset distance interval dis1, dis1 = [ΔT × v q,min ,ΔT×v q,max ],v q,min The default G 1,q The corresponding minimum speed, v q,max The default G 1,q The corresponding maximum speed; S540, traverse H 1,q , get H k(q) 1,q Similarity f with C2 k(q) q,2 , f k(q) q,2 =qua k(q) 1,q / num k(q) 1,q ,qua k(q) 1,q C2 appears in H k(q) 1,q The number of position labels in , num k(q) 1,q H k(q) 1,q The number of location tags; S550, if f k(q) q,2 ≥e1, then determine H k(q) 1,q is the suspected second position of the target vehicle; otherwise, it is determined that H k(q) 1,q The second location is not the target vehicle.

3. The method for realizing a dynamic high spatiotemporal resolution noise map according to claim 2, characterized in that: S550 and later also include: S560, if H 1,q Each H k(q) 1,q If both are determined not to be the second position of the target vehicle, it is determined that G 1,q is not the first position of the target vehicle; if H 1,q There is H k(q) 1,q If it is determined to be the suspected second position of the target vehicle, the process proceeds to S570; S570, get H 1,q The geographical hash grid set H of the suspected second location determined to be the target vehicle 2,q , H 2,q =(H 1 2,q ,H 2 2,q ,…,H x(q) 2,q ,…,H y(q) 2,q ), H x(q) 2,q H 1,q The x(q)th geohash grid that is determined to be the suspected second location of the target vehicle, where the value of x(q) ranges from 1 to y(q), and y(q) is H 1,q The number of geo-hash grids determined to be the suspected second location of the target vehicle; S580, traverse H 2,q , get H x(q) 2,q The corresponding first geographical hash grid set, if H x(q) 2,q If each hash grid in the corresponding first geographic hash grid set is determined to be not the third position of the target vehicle, then H x(q) 2,q The second position that is not the target vehicle; H x(q) 2,q The corresponding first geographic hash grid set is the set of nodes in the target area that are related to H x(q) 2,q The distance falls into the geographical hash grid of the second preset distance interval dis2, dis2 = [ΔT × v x(q) q,min ,ΔT×v x(q) q,max ],v x(q) q,min The default H x(q) 2,q The corresponding minimum speed, v x(q) q,max The default H x(q) 2,q The corresponding maximum speed; S590, if H 2,q Each H x(q) 2,q If both are determined not to be the second position of the target vehicle, it is determined that G 1,q Not the first position for the target vehicle.

4. The method for realizing a dynamic high spatiotemporal resolution noise map according to claim 3, characterized in that: S590 and later also include: S591: If all M suspected positions of the target vehicle exist, and there is a track in the target area that satisfies the first preset condition, the track is determined to be the track of the target vehicle; the first preset condition includes: the starting position of the track is the suspected first position of the target vehicle, the ending position of the track is the suspected Mth position of the target vehicle, the track passes through the suspected βth position of the target vehicle, and the value range of β is 2 to M-1; the distance between the suspected ηth position and the suspected η+1th position on the track falls within the preset distance interval corresponding to the suspected ηth position, and the preset distance interval corresponding to the suspected ηth position is [ΔT×vmin η ,ΔT×vmax η ],vmin η is the preset minimum speed corresponding to the suspected ηth position, vmax η is the preset maximum speed corresponding to the suspected ηth position, and the value range of η is 1 to M-1; S592: If all the M suspected positions of the target vehicle exist, and there are more than two tracks in the target area that satisfy the first preset condition, the track with the greatest corresponding similarity is determined as the track of the target vehicle; the similarity corresponding to the wth track in the target area that satisfies the first preset condition is ε w,m The suspected mth position of the target vehicle passing through the wth trajectory that meets the first preset condition and C m The similarity.

5. The method for realizing a dynamic high spatiotemporal resolution noise map according to claim 1, characterized in that: The noise feature is a Mel-spectrogram feature.

6. The method for realizing a dynamic high spatiotemporal resolution noise map according to claim 1, characterized in that: S300 includes: S310, B m Input the trained target neural network model and obtain the output result L of the trained target neural network model m , L m =(L m,1 ,L m,2 ,…,L m,r ,…,L m,R ), L m,r B is the output of the trained target neural network m The probability corresponding to the rth preset position label, where r ranges from 1 to R, and R is the number of preset position labels; the trained target neural network model is used to obtain the probability that the input noise feature corresponds to each preset position label; S320, traverse L m , if L m,r ≥e2, then L m,r The corresponding preset position tags are added to the preset initial tag set H, which is initialized to an empty set; e2 is the preset probability threshold; S330, if H is not an empty set, determine H to be C m .

7. The method for realizing a dynamic high spatiotemporal resolution noise map according to claim 6, characterized in that: The training process of the target neural network model includes: S311, obtaining a noise feature sample set, where the noise feature sample set includes a plurality of noise feature samples; S312, obtaining a noise feature label set, wherein the noise feature label set includes a position label set of each noise feature sample in the noise feature sample set, and the position label set of each noise feature sample includes a plurality of position labels; S313: Use the noise feature sample set and the noise feature label set to train the target neural network.

8. The method for realizing a dynamic high spatiotemporal resolution noise map according to claim 1, characterized in that: The preset information includes average noise intensity.

9. The method for realizing a dynamic high spatiotemporal resolution noise map according to claim 8, characterized in that: The preset information also includes a maximum noise intensity.

10. The method for realizing a dynamic high spatiotemporal resolution noise map according to claim 9, characterized in that: The preset information also includes the collection time.

Images