Method for detecting and visualizing holographic sound field based on schlieren equipment
Through the combination of deep learning models and image-patterning equipment, high-precision detection and three-dimensional visualization of the holographic sound field are realized, solving the problems of low spatial resolution, difficulty in real-time visualization, background noise interference and high equipment costs in the existing technology, and improving the problem of high spatial resolution, improving the The overall level of holographic sound field detection and visualization.
Patent Information
- Application Number
- CN202410932959.2
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-07-12
- Publication Date
- 2024-11-05
AI Technical Summary
In the prior art, holographic sound field detection and visualization have problems such as low spatial resolution, difficulty in real-time visualization, background noise interference and high equipment costs, which limit the application and popularization of the technology.
The deep learning model is used to reversely solve the vibration phase value of the ultrasonic transducer array, and combine the real-time reception and processing of acoustic hologram data of the pattern and the high-precision detection and three-dimensional visualization of the holographic sound field. By adjusting the cutting angle of the pattern and the Parameters optimize ultrasonic reception, reduce background noise interference, and reduce equipment costs.
The spatial resolution of the holographic sound field is improved, real-time dynamic visualization is realized, background noise interference is reduced, equipment costs are reduced, operation process is simplified, and the overall level of holographic sound field detection and visualization is improved.
Smart Images

Figure CN118896677A_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the field of acoustic holography and machine learning, and more specifically to a method for detecting and visualizing a holographic sound field based on a schlieren device. Background Art
[0002] The Schlieren method, also known as the Schlieren method, is a classic optical display technology. Combining a high-speed camera with high time resolution with the Schlieren method, it becomes a high-speed Schlieren method. This method is used to display flow fields, shock wave fronts and propagation in transparent media, observe micro-matter ejection on the free surface under high pressure, wave system conditions on the interface, interface instability, and the development of weak shock waves such as spark discharge under high pressure. It is an optical testing technology with a wide range of uses.
[0003] At present, in the process of holographic sound field detection and visualization technology, binocular vision measurement technology is used to measure the spatial position of the sound field of the moving sound source, and the holographic sound field is reconstructed through microphone array and acoustic holography method; simulation tools, source path analysis software and other acoustic measurement tools are used to simulate and reconstruct the holographic sound field. For the detection and visualization of holographic sound fields by Schlieren equipment, conventional technologies have the following problems:
[0004] 1. Spatial resolution limitation: The spatial resolution limitation of the holographic sound field is determined by the sampling density of the sound field and the physical characteristics of the ultrasonic transducer array. When performing fine detection or requiring a high-resolution holographic sound field image, it may be impossible to accurately capture subtle changes in the sound field, especially in the case of high-frequency sound waves.
[0005] 2. Real-time visualization: Converting acoustic holographic data into visual images is a complex process, especially in application scenarios that require real-time processing. Complex algorithms are needed to fit the holographic sound field. Real-time processing and visualization require the system to have high-speed data acquisition, processing and image rendering capabilities.
[0006] 3. Background noise interference: In actual experimental environments, the presence of background noise may interfere with acoustic holography measurements and affect the accuracy of holographic sound field reconstruction. Although there are technologies that try to reduce the impact of noise through specific algorithms, it is still a challenge in high-noise environments.
[0007] 4. Equipment and cost issues: High-quality acoustic holography detection equipment is often expensive, which may limit the popularization and application of the technology. At the same time, the construction and maintenance of the equipment also require professional knowledge, which increases the complexity of the operation. Summary of the invention
[0008] In view of this, the present invention provides a method for detecting and visualizing a holographic sound field based on a schlieren device, aiming to solve the problems existing in the prior art such as low spatial resolution, difficulty in real-time visualization, background noise interference and high equipment cost, and can complete the detection and visualization of the holographic sound field.
[0009] In order to achieve the above object, the present invention adopts the following technical solution:
[0010] The present invention provides a method for detecting and visualizing a holographic sound field based on a schlieren device, comprising the following steps:
[0011] S1. Use the deep learning model to reversely solve the vibration phase value of the ultrasonic transducer array, control the phase distribution of the phase transducer array, and obtain the required acoustic hologram;
[0012] S2, receiving feedback data of the acoustic hologram in real time through a schlieren device, and realizing real-time dynamic display of the holographic sound field image to obtain a corresponding standing wave image;
[0013] S3. Perform three-dimensional modeling based on the standing wave image to achieve dynamic real-time display of stereoscopic visualization of the acoustic holographic image.
[0014] Furthermore, the step S2 also includes: optimizing the reception of ultrasonic waves that meet the experimental requirements by adjusting the cutting angle and parameters of the Schlieren device.
[0015] Furthermore, the method further comprises:
[0016] S4. Using the corresponding standing wave image obtained, verify the accuracy of the deep learning model.
[0017] Furthermore, the training process of the deep learning model in step S1 includes:
[0018] The ASM method is used to synthesize phase images as training data;
[0019] The network is trained using the deep learning method of the U-Net architecture, using the processed phase image and its target image as training pairs for training;
[0020] After the training is completed, a deep learning model is obtained to reconstruct the target acoustic hologram.
[0021] Furthermore, the synthesizing phase map using the ASM method as training data includes:
[0022] The sound wave propagation process is regarded as a linear spatial filter, and the phase map is generated using the transfer function;
[0023] The transfer function is:
[0024]
[0025] Among them, P0(f x ,f y ,0) is the angular spectrum at z = 0, P(f x ,f y ,z) is the angular spectrum at z = z; λ represents the wavelength of the sound wave;
[0026] When the propagation distance is greater than the preset wavelength, the evanescent wave has decayed to 0, and the transfer function is expressed as:
[0027]
[0028] The frequency on the x-axis Frequency on the y-axis cosα represents the cosine value of the angle of sound wave propagation in the x-axis direction; cosβ represents the cosine value of the angle of sound wave propagation in the y-axis direction; when the frequency satisfies When H(f x ,f y ) is 1, there is a phase shift, and the sampling restriction condition is:
[0029]
[0030] φ represents the phase, which is defined as the argument of the transfer function H; Δf x represents the sampling interval in the x direction; Δf y represents the sampling interval in the y direction;
[0031] In the above, the complex amplitude distribution of the plane sound field is linked to the linear superposition of plane waves in different propagation directions, and the phase distribution diagram of the acoustic hologram generated by the angular spectrum method is obtained.
[0032] It can be seen from the above technical solutions that, compared with the prior art, the present invention discloses a method for detecting and visualizing a holographic sound field based on a Schlieren device, which has the following advantages:
[0033] 1. Improve spatial resolution: Through Schlieren equipment and technical means, the holographic sound field data will be received by the Schlieren instrument and transmitted to the experimental equipment, so that it can more accurately capture subtle changes in the holographic sound field.
[0034] 2. Realize real-time visualization: The data of holographic sound field feedback is received in real time through the Schlieren instrument and transmitted to the experimental equipment in real time, so that the real-time dynamics of holographic generation can be seen.
[0035] 3. Reduce background noise interference: By adjusting the cutting angle of the schlieren instrument, the device can receive ultrasonic waves that meet the experiment and reduce the interference of background noise on acoustic holography in the actual environment.
[0036] 4. Reduce equipment costs: Through Schlieren equipment and technical means, the cost of high-quality acoustic holographic detection equipment can be reduced, and the equipment construction and maintenance process can be simplified, making it easier to popularize and apply.
[0037] 5. Create an acoustic holographic generation system: Use deep learning methods to reversely solve the vibration phase value of the ultrasonic transducer phase array, control the phase distribution of the phase transducer array, and obtain the required acoustic hologram. Then use the Schlieren device to collect data for detection, verify the accuracy of the deep learning reverse solution through three-dimensional modeling, and visualize the acoustic holographic image in real time. BRIEF DESCRIPTION OF THE DRAWINGS
[0038] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on the provided drawings without paying creative work.
[0039] Figure 1 Schematic diagram of the method for detecting and visualizing a holographic sound field based on a Schlieren device provided by the present invention.
[0040] Figure 2 Schematic diagram of the U-Net architecture provided by the present invention. DETAILED DESCRIPTION
[0041] 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 ordinary technicians in this field without creative work are within the scope of protection of the present invention.
[0042] Reference Figure 1 As shown, an embodiment of the present invention discloses a method for detecting and visualizing a holographic sound field based on a schlieren device, comprising the following steps:
[0043] S1. Use the deep learning model to reversely solve the vibration phase value of the ultrasonic transducer array, control the phase distribution of the phase transducer array, and obtain the required acoustic hologram;
[0044] S2, receiving the feedback data of the acoustic hologram in real time through the schlieren device, and realizing the real-time dynamic display of the holographic sound field image, and obtaining the corresponding standing wave image; by using two schlieren devices, receiving the feedback data of the acoustic hologram from two different angles respectively, the two schlieren devices simultaneously receive the feedback data of the acoustic hologram reflected from the ultrasonic transducer array;
[0045] The data obtained by the two schlieren instruments are fused to ensure that the images received from different angles can complement each other, thus forming a complete three-dimensional holographic image. The received feedback data is processed by the system algorithm to generate an image of the holographic sound field. The processed image is displayed in real time and dynamically through the display device, and the observer can optimize the display effect by adjusting the angle and parameters. During the real-time display of the image, the corresponding standing wave image is obtained by analyzing the standing wave characteristics in the image, providing basic data for subsequent three-dimensional modeling.
[0046] S3. Perform three-dimensional modeling based on the standing wave image to achieve dynamic real-time display of stereoscopic visualization of the acoustic holographic image.
[0047] S4. Using the corresponding standing wave image obtained, verify the accuracy of the deep learning model.
[0048] The present invention uses a deep learning model to reversely solve the vibration phase value of the ultrasonic transducer array: using the deep learning model, by reversely solving the vibration phase value of the ultrasonic transducer array, the phase distribution of the phase transducer array is controlled to generate the required acoustic hologram. The feedback data of the acoustic hologram is received in real time through the Schlieren device: the feedback data of the generated acoustic hologram is received in real time using the Schlieren device. These data are used to realize the real-time dynamic display of the holographic sound field image, so as to obtain the corresponding standing wave image. Three-dimensional stereo modeling is performed to realize dynamic real-time display: according to the received standing wave image, three-dimensional stereo modeling is performed to realize stereoscopic visualization of the acoustic holographic image, and these images are displayed dynamically and in real time. In addition, the obtained standing wave image is used to verify the accuracy of the deep learning model to ensure the accuracy of the generated acoustic holographic image.
[0049] Specifically, Figure 1 Two Schlieren devices are shown, capturing images from two angles. This configuration facilitates 3D imaging because the different angles of the two Schlieren devices capture different perspectives of the sound field, thereby constructing a more comprehensive 3D image. The Schlieren device further improves the accuracy and quality of the image by adjusting the cut-in angle and parameters to optimize the reception of ultrasonic data that meets the experimental needs.
[0050] Through this method, by combining deep learning with Schlieren technology, high-precision detection and three-dimensional visualization of the holographic sound field are achieved, overcoming some limitations of the prior art, such as low spatial resolution, difficulty in real-time visualization, and background noise interference. The present invention reversely solves the vibration phase value of the ultrasonic transducer array through deep learning, combined with the real-time feedback of the Schlieren device, to achieve three-dimensional visualization of the holographic sound field.
[0051] In one embodiment, step S2 further includes: optimizing the reception of ultrasonic waves that meet experimental requirements by adjusting the cutting angle and parameters of the schlieren device.
[0052] The training process of the deep learning model in step S1 includes:
[0053] 1) Using the ASM method to synthesize phase images as training data;
[0054] 2) Train the network through the deep learning method of U-Net architecture, and use the processed phase image and its target image to form a training pair for training;
[0055] 3) After the training is completed, a deep learning model is obtained to reconstruct the target acoustic hologram.
[0056] In order to solve the problem of insufficient real training data for acoustic holograms, a set of phases on the hologram plane is used for data processing, and a sufficient amount of training data is synthesized using the ASM method. The training pairs composed of the data-processed phase map and its target image are put into the proposed U-Net DL method for training. After the training, the trained network can quickly reconstruct the target acoustic hologram.
[0057] The above step 1) uses the ASM method to synthesize the phase map as training data, including:
[0058] The sound wave propagation process is regarded as a linear spatial filter, and the phase map is generated using the transfer function;
[0059] The transfer function is:
[0060]
[0061] Among them, P0(f x ,f y ,0) is the angular spectrum at z = 0, which is the phase distribution of the sound wave at the starting point; P(f x ,f y ,z) is the angular spectrum at z = z, which is the phase change of the sound wave during propagation; f x Represents the spatial frequency in the x direction, which is the frequency component of the sound wave propagating in the x direction. y It represents the spatial frequency in the y direction, which is the frequency component of the sound wave propagating in the y direction. z represents the distance the sound wave propagates, which is the distance from the initial position to the current observation point.
[0062] λ represents the wavelength of sound waves, which is the inherent length scale of sound waves when they propagate in a medium. x ,f y ) represents the transfer function, which is used to describe the propagation characteristics of sound waves in space.
[0063] Describes the phase change that a sound wave experiences during propagation, where j is an imaginary unit and 2π is a phase change in one cycle; Indicates the magnitude of the phase change.
[0064] When the propagation distance is greater than the preset wavelength, the evanescent wave has decayed to 0, and the transfer function is expressed as:
[0065]
[0066] The frequency on the x-axis Frequency on the y-axis cosα represents the cosine value of the angle of sound wave propagation in the x-axis direction; cosβ represents the cosine value of the angle of sound wave propagation in the y-axis direction; when the frequency satisfies When H(f x ,f y ) is 1, there is a phase shift, and the sampling restriction condition is:
[0067]
[0068] φ represents the phase, which is defined as the argument (arg) of the transfer function H; Δf x represents the sampling interval in the x direction; Δf y represents the sampling interval in the y direction; The phase of f x The partial derivative of f indicates that the phase changes with f x The rate of change. The phase of f y The partial derivative of f indicates that the phase changes with f y The rate of change.
[0069] The inequality in the formula represents the restriction on the rate of phase change to ensure that the phase x and Δf y The acoustic hologram can be correctly reconstructed under the condition of avoiding aliasing in the x and y directions.
[0070] Through the above sampling constraints, it can be ensured that the phase distribution of the acoustic hologram can be accurately recorded and reconstructed at a given sampling interval, avoiding reconstruction errors caused by excessively high phase change rates.
[0071] In the above, the complex amplitude distribution of the plane sound field is linked to the linear superposition of plane waves in different propagation directions, and the phase distribution diagram of the acoustic hologram generated by the angular spectrum method is obtained.
[0072] The above process uses the ultrasonic transducer phase distribution solved by deep learning inversely combined with the real-time data reception and processing of the Schlieren device to efficiently realize the three-dimensional visualization of the holographic sound field. The system optimizes the resolution and accuracy of the image through multi-angle Schlieren data fusion, thereby realizing high-precision detection and real-time dynamic display of the holographic sound field in the experiment.
[0073] In step 2) above, U-Net’s DL method training
[0074] In the study of acoustic hologram reconstruction, the semantic segmentation U-Net network is an important deep learning architecture. In order to maximize its performance and build a better performing architecture, a series of optimizations were performed on the U-Net network. The architecture was modified and extended to enable it to process very few training images and produce more accurate segmentations. For positioning, the high-resolution features of the contraction path are combined with the upsampled output. Then, consecutive convolutional layers can learn to combine more accurate outputs based on this information.
[0075] The network architecture diagram is as follows Figure 2 As shown in the figure. The U-Net structure consists of three main parts: the contraction path on the left (which takes the target image as its input), the expansion path on the right (which outputs the predicted acoustic hologram), and the bottleneck layer in the middle (which bridges the features between the contraction path and the expansion path). The contraction path follows the typical architecture of the convolutional network. It consists of two parts, the encoder and the decoder, where the encoder is used to extract features and the decoder is used to restore the high-resolution acoustic hologram. The U-Net network has a total of 5 layers, which downsample and upsample the input target image 4 times respectively.
[0076] In the feature extraction part, it is a contraction network that reduces the image size through 4 downsampling. In this continuous downsampling process, the features extract shallow information. The specific process is that the image goes through two 3x3 convolutions, each followed by a rectified linear unit (RELU) and a 2x2 maximum pooling operation with a step size of 2 for downsampling. In each downsampling step, the number of feature channels is doubled. There are four splicing operations in U-Net, also called Skip connect, the purpose is to fuse feature information and merge deep and shallow information. When splicing, it should be noted that not only the image size should be consistent, but also the dimension (channels) of the features should be the same before splicing. Each step in the expansion path includes upsampling of the feature map, that is, the image size becomes larger and deep information is extracted. It is followed by a 2x2 convolution (up convolution) that halves the number of feature channels, a cascade with the corresponding cropped feature map from the contraction path, and two 3x3 convolutions, each followed by a RELU.
[0077] Cropping is necessary because boundary pixels are lost in each convolution. In the final layer, a 2x2 convolution is used to map each 64-component feature vector to the desired number of classes. The network has a total of 26 convolutional layers.
[0078] Through the detailed steps of the above embodiment, high-precision holographic sound field detection and real-time stereoscopic visualization display based on the Schlieren device can be realized, effectively solving the problems of low spatial resolution, difficulty in real-time visualization, background noise interference and high equipment cost in the prior art, and improving the overall level of holographic sound field detection and visualization.
[0079] In this specification, each embodiment is described in a progressive manner, and each embodiment focuses on the differences from other embodiments. The same or similar parts between the embodiments can be referred to each other. For the device disclosed in the embodiment, since it corresponds to the method disclosed in the embodiment, the description is relatively simple, and the relevant parts can be referred to the method part.
[0080] The above description of the disclosed embodiments enables one skilled in the art to implement or use the present invention. Various modifications to these embodiments will be apparent to one skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the present invention. Therefore, the present invention will not be limited to the embodiments shown herein, but rather to the widest scope consistent with the principles and novel features disclosed herein.
Claims
1. A method for detecting and visualizing a holographic sound field based on a Schlieren device, characterized in that: The following steps are involved: S1. Use the deep learning model to reversely solve the vibration phase value of the ultrasonic transducer array, control the phase distribution of the phase transducer array, and obtain the required acoustic hologram; S2, receiving feedback data of the acoustic hologram in real time through a schlieren device, and realizing real-time dynamic display of the holographic sound field image to obtain a corresponding standing wave image; S3. Perform three-dimensional modeling based on the standing wave image to achieve dynamic real-time display of stereoscopic visualization of the acoustic holographic image.
2. The method for detecting and visualizing a holographic sound field based on a Schlieren device according to claim 1, characterized in that: The step S2 also includes: optimizing the reception of ultrasonic waves that meet experimental requirements by adjusting the cutting angle and parameters of the Schlieren device.
3. The method for detecting and visualizing a holographic sound field based on a Schlieren device according to claim 1, characterized in that: The method further includes: S4. Using the corresponding standing wave image obtained, verify the accuracy of the deep learning model.
4. The method for detecting and visualizing a holographic sound field based on a Schlieren device according to claim 1, characterized in that: The training process of the deep learning model in step S1 includes: The ASM method is used to synthesize phase images as training data; The network is trained using the deep learning method of the U-Net architecture, using the processed phase image and its target image as training pairs for training; After the training is completed, a deep learning model is obtained to reconstruct the target acoustic hologram.
5. The method for detecting and visualizing a holographic sound field based on a Schlieren device according to claim 4, characterized in that: The method of synthesizing a phase map using the ASM method as training data includes: The sound wave propagation process is regarded as a linear spatial filter, and the phase map is generated using the transfer function; The transfer function is: Among them, P0(f x ,f y ,0) is the angular spectrum at z = 0, P(f x ,f y ,z) is the angular spectrum at z = z; λ represents the wavelength of the sound wave; When the propagation distance is greater than the preset wavelength, the evanescent wave has decayed to 0, and the transfer function is expressed as: The frequency on the x-axis Frequency on the y-axis cosα represents the cosine value of the angle of sound wave propagation in the x-axis direction; cosβ represents the cosine value of the angle of sound wave propagation in the y-axis direction; when the frequency satisfies When H(f x ,f y ) is 1, there is a phase shift, and the sampling restriction condition is: φ represents the phase, which is defined as the argument of the transfer function H; Δf x represents the sampling interval in the x direction; Δf y represents the sampling interval in the y direction; In the above, the complex amplitude distribution of the plane sound field is linked to the linear superposition of plane waves in different propagation directions, and the phase distribution diagram of the acoustic hologram generated by the angular spectrum method is obtained.
Citation Information
Cited By
Visualization device for measuring sound field in low and medium frequency standing wave tube
CN120668246A
