阅读说明：本技术 一种基于人脸检测的音频定位方法及装置 (Audio positioning method and device based on face detection ) 是由 方向阳 李骊 于 2019-10-17 设计创作，主要内容包括：本申请提供的基于人脸检测的音频定位方法及装置,获取使用音频定位算法确定的声源的位置信息,获取对图像进行人脸检测得到的人脸的位置信息,依据声源的位置信息,确定声源在图像中的位置信息,并依据声源在图像中的位置信息和人脸的位置信息,确定目标位置信息,最后依据目标位置信息和声源在图像中的位置信息,得到声源的校正后的位置信息。可见,本技术方案,通过将音频定位算法与对图像人脸检测算法相结合,得到声源的目标位置信息,因为,基于人脸检测算法得到的位置信息的准确度高,所以目标位置信息也是准确度较高的位置信息。同时,依据目标位置信息和声源在图像中的位置信息,对目标位置信息校正,使声源的校正后的位置信息更加准确。(The audio positioning method and device based on the face detection acquire the position information of a sound source determined by using an audio positioning algorithm, acquire the position information of a face obtained by performing face detection on an image, determine the position information of the sound source in the image according to the position information of the sound source, determine the target position information according to the position information of the sound source in the image and the position information of the face, and finally acquire the corrected position information of the sound source according to the target position information and the position information of the sound source in the image. Therefore, according to the technical scheme, the audio positioning algorithm and the image face detection algorithm are combined to obtain the target position information of the sound source, and the position information obtained based on the face detection algorithm is high in accuracy, so that the target position information is also high in accuracy. Meanwhile, the target position information is corrected according to the target position information and the position information of the sound source in the image, so that the corrected position information of the sound source is more accurate.)

1. An audio positioning method based on face detection is characterized by comprising the following steps:

acquiring position information of a sound source determined using an audio positioning algorithm;

acquiring position information of a face obtained by performing face detection on an image, wherein the acquisition range of the image comprises the sound source;

determining the position information of the sound source in the image according to the position information of the sound source;

determining target position information according to the position information of the sound source in the image and the position information of the human face, wherein the target position information is the position information of the human face matched with the position information of the sound source;

and obtaining corrected position information of the sound source according to the target position information and the position information of the sound source in the image.

2. The audio positioning method based on human face detection according to claim 1, wherein the obtaining the position information of the sound source determined by using the audio positioning algorithm comprises:

acquiring a horizontal angle of a sound source determined by using a direction of arrival audio positioning algorithm;

determining the position information of the sound source in the image according to the position information of the sound source, wherein the determining comprises the following steps:

acquiring a mapping relation between an angle and an image coordinate based on sound acquisition equipment and image acquisition equipment;

and determining the corresponding abscissa of the horizontal angle of the sound source in the image as the position information of the sound source in the image by using the mapping relation.

3. The audio positioning method based on human face detection according to claim 2, wherein said determining target position information according to the position information of the sound source in the image and the position information of the human face comprises:

and taking the abscissa in the position information of the human face, which has a difference value with the corresponding abscissa of the sound source in the image not greater than a preset threshold value, as target position information.

4. The audio positioning method based on face detection according to claim 3, wherein the obtaining of the corrected position information of the sound source according to the target position information and the position information of the sound source in the image comprises:

and taking the sum of the target position information and a first numerical value as corrected position information of the sound source, wherein the first numerical value is determined according to the difference between the target position information and the corresponding abscissa of the sound source in the image.

5. The audio positioning method based on human face detection according to claim 1, wherein the determining of the position information of the sound source by using the audio positioning algorithm comprises:

determining cross-correlation values for a plurality of microphone pairs;

determining a peak value of the cross-correlation value and an angle range corresponding to the peak value;

determining an angle corresponding to the peak value from the angle range;

and determining the position information of the sound source according to the angle.

6. An audio positioning device based on face detection, comprising:

a first acquisition unit for acquiring position information of a sound source determined using an audio localization algorithm;

the second acquisition unit is used for acquiring the position information of a human face obtained by detecting the human face of an image, and the acquisition range of the image comprises the sound source;

a first determining unit, configured to determine position information of the sound source in the image according to the position information of the sound source;

a second determining unit, configured to determine target location information according to location information of the sound source in the image and location information of the face, where the target location information is location information of the face that matches the location information of the sound source;

and the third determining unit is used for obtaining the corrected position information of the sound source according to the target position information and the position information of the sound source in the image.

7. The audio positioning apparatus based on human face detection according to claim 6, wherein the first obtaining unit is configured to obtain position information of a sound source determined by using an audio positioning algorithm, and comprises:

the first obtaining unit is specifically configured to obtain a horizontal angle of a sound source determined using a direction of arrival audio localization algorithm;

the first determining unit is configured to determine the position information of the sound source in the image according to the position information of the sound source, and includes:

the first determining unit is specifically configured to obtain a mapping relationship between an angle and an image coordinate based on the sound collecting device and the image collecting device; and determining the corresponding abscissa of the horizontal angle of the sound source in the image as the position information of the sound source in the image by using the mapping relation.

8. The audio positioning apparatus based on human face detection according to claim 7, wherein the second determining unit is configured to determine target position information according to the position information of the sound source in the image and the position information of the human face, and comprises:

the second determining unit is specifically configured to use, as the target position information, an abscissa of the abscissa in the position information of the face, where a difference between the abscissa and the corresponding abscissa of the sound source in the image is not greater than a preset threshold.

9. The audio localization apparatus according to claim 8, wherein the third determining unit is configured to obtain the corrected position information of the sound source according to the target position information and the position information of the sound source in the image, and includes:

the third determining unit is specifically configured to use a sum of the target position information and a first value as corrected position information of the sound source, where the first value is determined according to a difference between the target position information and an abscissa of the sound source corresponding to the image.

10. The audio positioning apparatus based on human face detection according to claim 6, wherein the first obtaining unit is configured to obtain position information of a sound source determined by using an audio positioning algorithm, and comprises:

the first obtaining unit is specifically configured to determine cross-correlation values of a plurality of microphone pairs; determining a peak value of the cross-correlation value and an angle range corresponding to the peak value; determining an angle corresponding to the peak value from the angle range; and determining the position information of the sound source according to the angle.

Technical Field

The present application relates to the field of electronic information, and in particular, to an audio positioning method and apparatus based on face detection.

Background

Audio localization (also known as sound source localization) is widely used in military, civilian, and industrial applications as a microphone array-based localization technique. For example, in the civil field and in many large-scale activity scenes, an audio positioning technology can be used for positioning a speaker in the activity, and the result of the audio positioning is used for adjusting the direction of the camera so that the camera can shoot aiming at the speaker, so that the dependence on the camera shooting personnel is reduced, and the labor cost can be reduced.

However, the existing audio positioning technology generally has the problem of inaccurate positioning for some complex scenes, such as scenes with many people, so how to improve the accuracy of audio positioning becomes a problem that needs to be solved at present.

Disclosure of Invention

The research process of the applicant finds that: compared with the audio-based positioning, the accuracy of the position information obtained based on the face detection is higher, so that the position information obtained based on the face detection can be used for correcting the audio-based positioning information so as to improve the accuracy.

In order to achieve the above object, the present application provides the following technical solutions:

an audio positioning method based on face detection comprises the following steps:

acquiring position information of a sound source determined using an audio positioning algorithm;

acquiring position information of a face obtained by performing face detection on an image, wherein the acquisition range of the image comprises the sound source;

determining the position information of the sound source in the image according to the position information of the sound source;

determining target position information according to the position information of the sound source in the image and the position information of the human face, wherein the target position information is the position information of the human face matched with the position information of the sound source;

and obtaining corrected position information of the sound source according to the target position information and the position information of the sound source in the image.

Optionally, the obtaining the position information of the sound source determined by using the audio positioning algorithm includes:

acquiring a horizontal angle of a sound source determined by using a direction of arrival audio positioning algorithm;

determining the position information of the sound source in the image according to the position information of the sound source, wherein the determining comprises the following steps:

acquiring a mapping relation between an angle and an image coordinate based on sound acquisition equipment and image acquisition equipment;

and determining the corresponding abscissa of the horizontal angle of the sound source in the image as the position information of the sound source in the image by using the mapping relation.

Optionally, the determining the target location information according to the location information of the sound source in the image and the location information of the face includes:

and taking the abscissa in the position information of the human face, which has a difference value with the corresponding abscissa of the sound source in the image not greater than a preset threshold value, as target position information.

Optionally, the obtaining the corrected position information of the sound source according to the target position information and the position information of the sound source in the image includes:

and taking the sum of the target position information and a first numerical value as corrected position information of the sound source, wherein the first numerical value is determined according to the difference between the target position information and the corresponding abscissa of the sound source in the image.

Optionally, the position information of the sound source determined by using the audio localization algorithm includes:

determining cross-correlation values for a plurality of microphone pairs;

determining a peak value of the cross-correlation value and an angle range corresponding to the peak value;

determining an angle corresponding to the peak value from the angle range;

and determining the position information of the sound source according to the angle.

An audio positioning device based on face detection, comprising:

a first acquisition unit for acquiring position information of a sound source determined using an audio localization algorithm;

the second acquisition unit is used for acquiring the position information of a human face obtained by detecting the human face of an image, and the acquisition range of the image comprises the sound source;

a first determining unit, configured to determine position information of the sound source in the image according to the position information of the sound source;

a second determining unit, configured to determine target location information according to location information of the sound source in the image and location information of the face, where the target location information is location information of the face that matches the location information of the sound source;

and the third determining unit is used for obtaining the corrected position information of the sound source according to the target position information and the position information of the sound source in the image.

Optionally, the first obtaining unit is configured to obtain position information of a sound source determined by using an audio localization algorithm, and includes:

the first obtaining unit is specifically configured to obtain a horizontal angle of a sound source determined using a direction of arrival audio localization algorithm;

the first determining unit is configured to determine the position information of the sound source in the image according to the position information of the sound source, and includes:

the first determining unit is specifically configured to obtain a mapping relationship between an angle and an image coordinate based on the sound collecting device and the image collecting device; and determining the corresponding abscissa of the horizontal angle of the sound source in the image as the position information of the sound source in the image by using the mapping relation.

Optionally, the second determining unit is configured to determine target location information according to the location information of the sound source in the image and the location information of the face, and includes:

the second determining unit is specifically configured to use, as the target position information, an abscissa of the abscissa in the position information of the face, where a difference between the abscissa and the corresponding abscissa of the sound source in the image is not greater than a preset threshold.

Optionally, the third determining unit is configured to obtain corrected location information of the sound source according to the target location information and the location information of the sound source in the image, and includes:

the third determining unit is specifically configured to use a sum of the target position information and a first value as corrected position information of the sound source, where the first value is determined according to a difference between the target position information and an abscissa of the sound source corresponding to the image.

Optionally, the first obtaining unit is configured to obtain position information of a sound source determined by using an audio localization algorithm, and includes: the first obtaining unit is specifically configured to determine cross-correlation values of a plurality of microphone pairs; determining a peak value of the cross-correlation value and an angle range corresponding to the peak value; determining an angle corresponding to the peak value from the angle range; and determining the position information of the sound source according to the angle.

The method and the device obtain the position information of the sound source determined by using an audio positioning algorithm; meanwhile, the position information of the face obtained by detecting the face of the image is obtained, the position information of the sound source in the image is determined according to the position information of the sound source, the target position information is determined according to the position information of the sound source in the image and the position information of the face, and finally the corrected position information of the sound source is obtained according to the target position information and the position information of the sound source in the image. Therefore, according to the technical scheme provided by the application, the position information of the sound source determined by the audio positioning algorithm is combined with the position information of the face obtained by detecting the image face, so that the target position information of the sound source is obtained. Meanwhile, the obtained target position information is corrected according to the target position information and the position information of the sound source in the image, so that the corrected position information of the sound source is more accurate. In conclusion, the technical scheme can improve the accuracy of audio positioning.

Drawings

In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the drawings used in the description of the embodiments or the prior art will be briefly described below, it is obvious that the drawings in the following description are only some embodiments of the present application, and for those skilled in the art, other drawings can be obtained according to the drawings without creative efforts.

Fig. 1 is a flowchart of an audio positioning method based on face detection disclosed in an embodiment of the present application;

FIG. 2 is a flow chart of a method for determining location information of a sound source using an audio localization algorithm as disclosed in an embodiment of the present application;

fig. 3 is a schematic structural diagram of an audio positioning apparatus based on face detection according to an embodiment of the present application;

fig. 4 is a schematic structural diagram of an electronic device disclosed in an embodiment of the present application.

Detailed Description

The technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the drawings in the embodiments of the present application, and it is obvious that the described embodiments are only a part of the embodiments of the present application, and not all of the embodiments. All other embodiments, which can be derived by a person skilled in the art from the embodiments given herein without making any creative effort, shall fall within the protection scope of the present application.

Fig. 1 is a schematic diagram illustrating an audio positioning method based on face detection according to an embodiment of the present application, which includes the following steps:

s101, obtaining the position information of the sound source determined by using an audio positioning algorithm.

One way of determining the position information of a sound source using an audio localization algorithm is to: and determining the horizontal angle of the sound source by adopting a direction of arrival audio positioning algorithm. The direction of arrival audio localization algorithm is: according to the audio signal of the sound source collected by a sound collecting device (such as a microphone array), the audio signal is processed to obtain the horizontal angle of the sound source. It should be noted that, the horizontal angle of the sound source refers to an angle between the sound source and a horizontal line, and a specific position of the horizontal line may be set according to actual conditions. In which detailed processes of processing audio signals to obtain the horizontal angle of a sound source can refer to the prior art.

The microphone array is optional, a ring six-microphone array can be adopted, other matrix microphones can be adopted, and the specific microphone array can be determined by combining with actual requirements.

Compared with the existing audio positioning algorithm based on beam forming and the audio positioning algorithm based on high-resolution spectrum estimation, the direction-of-arrival audio positioning algorithm adopted by the embodiment is simple in calculation and small in calculation amount, is more beneficial to real-time processing of audio, and is suitable for non-repetitive audio signals of a single sound source.

In this embodiment, another way of determining the position information of the sound source by using the audio localization algorithm may also refer to the flow shown in fig. 2.

S102, obtaining position information of a human face obtained by human face detection of the image.

For example, a MTCNN algorithm (Multi-task convolution neural network) is used to perform face detection on an image, thereby determining position information of all faces included in the image. In this embodiment, the image is obtained by shooting an area including the sound source by the shooting device, that is, the acquisition range of the image includes the sound source.

The process of determining the location information of the face by using the MTCNN algorithm to perform face detection on the image will be described below with reference to an MTCNN algorithm model. Of course, other existing face detection algorithms may also be used for face detection, which is not limited herein.

And S103, determining the position information of the sound source in the image according to the position information of the sound source.

One way to determine the position information of the sound source in the image based on the position information of the sound source is: based on the sound collection device and the image collection device, a mapping relation between the angle of the sound source and the horizontal direction and the image coordinate is obtained, and the horizontal coordinate corresponding to the horizontal angle of the sound source in the image is determined as the position information of the sound source in the image by using the mapping relation.

Based on the sound acquisition equipment and the image acquisition equipment, the implementation mode of acquiring the mapping relation between the angle and the image coordinate is as follows: after a horizontal angle of a sound source is obtained by utilizing sound collection equipment and a direction-of-arrival audio positioning algorithm, coordinate information of the sound source in an actual space is obtained through calculation according to the horizontal angle, and a mapping relation between angles (such as a horizontal angle, a vertical angle and the like) of the sound source in all directions and image coordinates is obtained according to a conversion relation between an actual space coordinate system (namely a world coordinate system) and a camera coordinate system of image collection equipment. A more detailed calculation method for obtaining the mapping relationship between the angle and the image coordinate based on the sound collection device and the image collection device can be found in the prior art.

In this embodiment, since the abscissa of the X-axis direction in the image coordinates corresponds to the horizontal angle of the sound source in the real space, the horizontal angle of the sound source can be mapped to the corresponding abscissa in the image according to the mapping relationship between the angle and the image coordinates, and the abscissa of the sound source in the image can be used as the position information of the sound source in the image.

And S104, determining target position information according to the position information of the sound source in the image and the position information of the human face.

Since the acquisition range of the image includes the sound source, the image contains the position information of the face matching the sound source. In this embodiment, one way to determine the position information of the face matched with the sound source in the image is as follows: and taking the abscissa in the position information of the human face, which has a difference value with the corresponding abscissa of the sound source in the image not greater than a preset threshold value, as the target position information. That is, the target position information is position information of a face that matches the position information of the sound source. The preset threshold value can be determined according to actual requirements and parameters of the image acquisition equipment.

After the position information of the sound source is mapped to the image, the distance between the abscissa of the obtained sound source in the image and the abscissa of the position information of the face corresponding to the sound source is not large, so that the target position information of the sound source determined by the preset threshold has high accuracy.

In this embodiment, optionally, the image capturing device may adopt a panoramic camera.

And S105, obtaining corrected position information of the sound source according to the target position information and the position information of the sound source in the image.

Since the target position information obtained in S104 is obtained based on the abscissa of the face position information whose difference from the corresponding abscissa of the sound source in the image is not greater than the preset threshold, it is highly probable that the corresponding abscissa of the sound source in the image and the abscissa of the target position information are only similar abscissas, not identical abscissas, and therefore, in this embodiment, further, based on the target position information and the position information of the sound source in the image, the corrected position information of the sound source is obtained, so that the corrected position information is closer to the actual position of the sound source.

One way to obtain the corrected position information of the sound source according to the target position information and the position information of the sound source in the image may be: and taking the sum of the target position information and a first value as corrected position information of the sound source, wherein the first value is determined according to the difference between the target position information and the corresponding abscissa of the sound source in the image. For example, the first value may be one-half of the difference between the target position information and the abscissa of the sound source corresponding in the image. It should be noted that the difference may be an absolute value.

According to the method provided by the embodiment of the application, the position information of the sound source determined by the audio positioning algorithm is combined with the position information of the face obtained by detecting the image and the face, so that the target position information of the sound source is obtained. Meanwhile, the obtained target position information is corrected according to the target position information and the position information of the sound source in the image, so that the corrected position information of the sound source is more accurate. In conclusion, the technical scheme can improve the accuracy of audio positioning.

In addition, the technical method provided by the embodiment can improve the audio positioning accuracy, does not add extra cost to the scenes of face detection and audio sound source positioning, and has wider applicability particularly under the condition that the scenes of the audio detection and the face detection are more and more common.

On the other hand, after the corrected position information of the sound source is obtained by applying the method provided by the embodiment, the position information can be used to execute a face-related algorithm, for example, face tracking of the speaker, because the corrected position information of the sound source is more accurate, the execution accuracy of the face algorithm can be improved by feedback, and the tracked face and the speaker can be ensured to be the same person.

Fig. 2 is an implementation manner of determining position information of a sound source by using an audio localization algorithm according to an embodiment of the present application, which may include the following steps:

s201, cross-correlation values of a plurality of microphone pairs are determined.

In this embodiment, a ring-shaped six-microphone array is adopted, and the six-microphone array is divided into three pairs of microphones, each pair of microphones includes two microphones. Each pair of microphones receives two paths of audio signals, wherein the two paths of audio signals are audio signals sent by the same sound source, and the microphones process the received audio signals to obtain the cross-correlation numerical values of the audio signals.

S202, determining the peak value of the cross-correlation value and the angle range corresponding to the peak value.

In this implementation, Voice Activity Detection (VAD) is performed on audio signals, specifically, logarithmic frame energy estimation is performed on each frame of audio signal to obtain logarithmic frame energy distribution range estimation of audio, a VAD decision threshold is calculated according to an OTSU algorithm (maximum inter-class variance algorithm), and finally, the VAD decision threshold is adopted to filter the audio to obtain a VAD decision result.

According to VAD judgment results, effective cross-correlation values of the audio can be accumulated, peak detection of coarse search is carried out on the cross-correlation values, and therefore the peak value of the cross-correlation values and the angle range corresponding to the peak value are determined.

And S203, determining an angle corresponding to the peak value from the angle range.

After the angle range corresponding to the peak is obtained, for example, an angle corresponding to the peak is further searched for in the angle range corresponding to the peak by 64-fold interpolation.

And S204, determining the position information of the sound source according to the angle.

And estimating the direction of arrival according to the angle corresponding to the obtained peak value of each pair of microphone pairs. And obtaining three directions of arrival for the three pairs of microphones, performing data fusion to obtain more accurate directions of arrival, and determining the position information of the sound source according to the directions of arrival.

A more detailed implementation of the individual process steps in fig. 2 can be found in the prior art.

According to the method provided by the embodiment, the peak value of the cross-correlation value and the angle range corresponding to the peak value are determined firstly, and the angle corresponding to the peak value is determined from the angle range, so that not only can the amount be calculated, but also the determined angle can be more accurate.

It should be noted that, in this embodiment, before the cross-correlation value of the audio signal is calculated, the audio signal may be preprocessed, that is, the input original audio speech is high-pass filtered to remove the direct-current component, so as to improve the resolvability of the peak position in the cross-correlation estimation.

It should be noted that the audio localization algorithm may also use other audio localization algorithms, such as a beamforming method and a method based on high-resolution spectrum estimation.

The MTCNN algorithm is a multitask convolutional neural network algorithm, and the face region detection and the face key point detection are combined together, so that the face detection and the face alignment can be directly realized. The MTCNN algorithm model may be divided into three parts, a first part: P-Net (Proposal Network, proposed Network), second subsection: R-Net (optimized Network), and a third part O-Net (Output Network). The MTCNN algorithm performs image face recognition according to the three parts of the model as follows:

step one, P-Net firstly generates a candidate window and a frame regression vector, corrects the generated candidate window by using a frame regression method, and combines overlapped candidate frames by using NMS (Non-Maximum Suppression). The P-Net finally outputs a plurality of candidate windows in which human faces can exist. Wherein P-Net is a full-convolution Network similar to RPN (regional recommendation Network) in fast-RCNN.

And step two, inputting the candidate window output by the P-Net into the R-Net, wherein the R-Net filters a large number of candidate frames with poor effect, and finally performing frame regression and NMS (network management system) merging on the selected candidate frames to further improve the candidate window.

And step three, outputting the final face frame and the face feature point position by using O-Net. The O-Net has similar effect to the R-Net, and is different from the R-Net only in that one convolution layer is added, and the structure of the convolution layer can identify the facial region through more supervision and regress the facial feature points of the human, and finally five facial feature points are output.

The MTCNN algorithm is used to perform face detection on an image, and the prior art can be referred to obtain detailed information of position information of a face.

The method provided by the embodiment is based on the MTCNN algorithm, the image is subjected to face detection to obtain the position information of the face, and the MTCNN algorithm adopts the multi-task convolutional neural network algorithm, and the convolutional neural network has the characteristics of high accuracy and high efficiency in image classification, so that the MTCNN algorithm can be used for more accurately determining the position information of the face.

Corresponding to the audio positioning method based on face detection provided by the embodiment of the present application, referring to fig. 3, a schematic structural diagram of an audio positioning device based on face detection is shown, which includes:

a first acquisition unit 301 for acquiring position information of a sound source determined using an audio localization algorithm;

a second obtaining unit 302, configured to obtain position information of a face obtained by performing face detection on an image, where an acquisition range of the image includes the sound source;

a first determining unit 303, configured to determine position information of the sound source in the image according to the position information of the sound source;

a second determining unit 304, configured to determine target location information according to location information of the sound source in the image and location information of the face, where the target location information is location information of the face matched with the location information of the sound source;

a third determining unit 305, configured to obtain corrected position information of the sound source according to the target position information and the position information of the sound source in the image.

The specific implementation manner of the first obtaining unit 301 obtaining the position information of the sound source determined by using the audio positioning algorithm is as follows: acquiring a horizontal angle of a sound source determined using a direction of arrival audio localization algorithm,

the first determining unit 303 determines, according to the position information of the sound source, the position information of the sound source in the image in a specific implementation manner as follows: acquiring a mapping relation between an angle and an image coordinate based on sound acquisition equipment and image acquisition equipment; and determining the corresponding abscissa of the horizontal angle of the sound source in the image as the position information of the sound source in the image by using the mapping relation.

The second determining unit 304 determines the target position information according to the position information of the sound source in the image and the position information of the face in a specific implementation manner as follows: and taking the abscissa in the position information of the human face, which has a difference value with the corresponding abscissa of the sound source in the image not greater than a preset threshold value, as the target position information.

The third determining unit 305 obtains the corrected position information of the sound source according to the target position information and the position information of the sound source in the image by: and taking the sum of the target position information and a first numerical value as corrected position information of the sound source, wherein the first numerical value is determined according to the difference between the target position information and the corresponding abscissa of the sound source in the image.

The specific implementation manner of the first obtaining unit 301 obtaining the position information of the sound source determined by using the audio positioning algorithm may also be: determining cross-correlation values of the plurality of microphone pairs, determining a peak value of the cross-correlation values and an angle range corresponding to the peak value, determining an angle corresponding to the peak value from the angle range, and determining position information of the sound source according to the angle.

According to the device provided by the embodiment of the application, the position information of the sound source determined by the audio positioning algorithm is combined with the position information of the face obtained by detecting the image face, so that the target position information of the sound source is obtained. Meanwhile, in order to further improve the positioning accuracy, the obtained target position information is corrected according to the target position information and the position information of the sound source in the image, so that the corrected position information of the sound source is more accurate. In conclusion, the technical scheme can improve the accuracy of audio positioning.

An embodiment of the present invention further provides an electronic device, a schematic structural diagram of which is shown in fig. 4, and the electronic device specifically includes: a processor 401 and a memory 402, the memory 402 being for storing programs; the processor 401 is configured to run a program to implement the audio positioning method based on face detection in the embodiment of the present invention.

The embodiment of the present invention further provides a computer-readable storage medium, where instructions are stored in the computer-readable storage medium, and when the instructions are run on a computer, the computer is enabled to execute the audio positioning method based on face detection in the embodiment of the present invention.

The functions described in the method of the embodiment of the present application, if implemented in the form of software functional units and sold or used as independent products, may be stored in a storage medium readable by a computing device. Based on such understanding, part of the contribution to the prior art of the embodiments of the present application or part of the technical solution may be embodied in the form of a software product stored in a storage medium and including several instructions for causing a computing device (which may be a personal computer, a server, a mobile computing device or a network device) to execute all or part of the steps of the method described in the embodiments of the present application. And the aforementioned storage medium includes: a U-disk, a removable hard disk, a Read-Only Memory (ROM), a Random Access Memory (RAM), a magnetic disk or an optical disk, and other various media capable of storing program codes.

The embodiments are described in a progressive manner, each embodiment focuses on differences from other embodiments, and the same or similar parts among the embodiments are referred to each other.

The previous description of the disclosed embodiments is provided to enable any person skilled in the art to make or use the present application. Various modifications to these embodiments will be readily apparent to those skilled in the art, and the generic principles defined herein may be applied to other embodiments without departing from the spirit or scope of the application. Thus, the present application is not intended to be limited to the embodiments shown herein but is to be accorded the widest scope consistent with the principles and novel features disclosed herein.