Method and apparatus for creating multispectral infrared gas imaging datasets

Abstract

The application relates to the technical field of infrared gas imaging, and particularly provides a method and device for manufacturing a multispectral infrared gas imaging dataset, which comprises the following steps: constructing an infrared radiation transmission model of a gas; constructing a propagation and diffusion geometric model of the gas, wherein the propagation and diffusion geometric model comprises a gas type, a gas concentration and a gas profile; and manufacturing a multispectral infrared gas imaging dataset based on the propagation and diffusion geometric model and the infrared radiation transmission model. Thus, the multispectral infrared gas imaging dataset is obtained according to the infrared radiation transmission model and the propagation and diffusion geometric model of the gas, the manufacturing of the infrared gas imaging dataset is completed under the premise that no hazardous gas is discharged, safety is ensured, the multispectral infrared gas imaging dataset can be manufactured based on the gas type and concentration information, and the requirement of a dataset for deep learning training can be met.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of infrared gas imaging, in particular to a method for manufacturing a multispectral infrared gas imaging dataset and a device for manufacturing a multispectral infrared gas imaging dataset. BACKGROUND

[0002] Since most gases have no response in the visible band and weak response in the infrared band, the traditional algorithm has a high false detection rate when facing the detection task of such a complex background weak signal, and a deep learning method needs to be used. However, the deep learning method relies on a complete and rich infrared gas image database, and there is currently no complete dangerous and hazardous gas image database. Due to the hazards and safety problems of dangerous and hazardous gases, it is impossible to obtain a dangerous and hazardous gas image database by discharging dangerous and hazardous gases and taking pictures multiple times.

[0003] Therefore, there is an urgent need for a new way to obtain an infrared gas imaging dataset suitable for a spectral system of dangerous and hazardous gases. SUMMARY

[0004] To solve one of the above technical problems, the present application proposes the following technical solutions.

[0005] The first aspect of the present application proposes a method for manufacturing a multispectral infrared gas imaging dataset, comprising the following steps: constructing an infrared radiation transmission model of a gas; constructing a propagation and diffusion geometric model of the gas, wherein the propagation and diffusion geometric model comprises a gas type, a gas concentration, and a gas profile; and manufacturing a multispectral infrared gas imaging dataset based on the propagation and diffusion geometric model and the infrared radiation transmission model.

[0006] In addition, the method for manufacturing a multispectral infrared gas imaging dataset according to the above embodiments of the present application can have the following additional technical features.

[0007] According to one embodiment of the present application, the expression of the infrared radiation transmission model is:

[0008] DN=kτ srf τ opt (τ1τ2L B -τ2L bgas )(e -αCL -1)

[0009] wherein DN represents a signal change value caused by gas absorption, k represents a radiation calibration coefficient, τ srf represents the response rate of the detector to different spectra, τ opt represents the transmittance of the optical system, τ1 and τ2 represent the atmospheric transmittance between the observation background and the gas and between the gas and the detection system, respectively, and L Brepresents thermal radiation of the observed background, L bgas represents the radiance of the gas equivalent to blackbody radiation, a is the absorption coefficient of the gas, C is the concentration of the gas, and L is the optical path of the gas.

[0010] According to one embodiment of the present application, the propagation and diffusion geometric model is constructed by CFD simulation software.

[0011] According to one embodiment of the present application, the multispectral infrared gas imaging dataset is produced based on the propagation and diffusion geometric model and the infrared radiation transmission model, comprising: acquiring an infrared background image of the observed background by an infrared camera; combining the propagation and diffusion geometric model and the infrared radiation transmission model to obtain a simulated gas image; and superimposing the simulated gas image and the infrared background image to obtain a multispectral infrared gas image dataset.

[0012] According to one embodiment of the present application, after acquiring the infrared background image of the observed background by the infrared camera, the method further comprises: preprocessing the infrared background image.

[0013] The second aspect embodiment of the present application proposes a device for producing an infrared gas imaging dataset, comprising: a first construction module for constructing an infrared radiation transmission model of a gas; a second construction module for a propagation and diffusion geometric model of the gas, wherein the propagation and diffusion geometric model contains the type of the gas, the concentration of the gas, and the profile of the gas; and a production module for producing a multispectral infrared gas imaging dataset based on the propagation and diffusion geometric model and the infrared radiation transmission model.

[0014] The technical scheme of the embodiment of the present application obtains the infrared gas imaging dataset according to the infrared radiation transmission model and the propagation and diffusion geometric model of the gas, which not only completes the production of the multispectral infrared gas imaging dataset without emitting hazardous gas, ensuring safety, but also produces a rich and colorful multispectral infrared gas imaging dataset based on the type and concentration information of the gas, meeting the needs of the dataset for deep learning training. BRIEF DESCRIPTION OF DRAWINGS

[0015] Figure 1 The flowchart of the method for producing the multispectral infrared gas imaging dataset of the embodiment of the present application.

[0016] Figure 2 The schematic diagram of the infrared radiation transmission process of the gas of the embodiment of the present application.

[0017] Figure 3 The schematic diagram of the production of the multispectral infrared gas imaging data of one example of the present application.

[0018] Figure 4A structural block diagram of a device for making a multispectral infrared gas imaging dataset according to an embodiment of the present application. DETAILED DESCRIPTION

[0019] 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. Obviously, the described embodiments are only some of the embodiments of the present application, rather than all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative efforts fall within the scope of the present application.

[0020] Currently, the main technology is to convert a visible RGB image into an infrared image. For an infrared gas leakage scene, a gas leakage trace cannot be detected by visible light. Secondly, other superimposition methods lack theoretical analysis and only simply superimpose a gray gas profile, lack a concentration label, and are not applicable to a multispectral system.

[0021] In order to obtain a multispectral infrared gas imaging dataset with a gas profile and a concentration for deep learning training, the present application simulates the propagation path and diffusion effect of a hazardous gas in various scenes through a gas radiation transmission model and other algorithms, completes the establishment of a hazardous gas image database under the premise of not discharging the hazardous gas, lays a foundation for the training of a subsequent hazardous gas monitoring network and algorithm research, and can be applied to a multispectral system.

[0022] Figure 1 A flowchart of a method for making a multispectral infrared gas imaging dataset according to an embodiment of the present application.

[0023] As shown in Figure 1 , the method for making a multispectral infrared gas imaging dataset includes the following steps S1 to S3.

[0024] S1, an infrared radiation transmission model of a gas is constructed.

[0025] Specifically, when infrared radiation passes through a gas and interacts with molecules to cause vibration or rotation, energy is transferred to the molecules, resulting in the absorption of infrared radiation of a specific wavelength of the gas. This absorption phenomenon can be visualized through an absorption spectrum, similar to a fingerprint of the gas.

[0026] During the entire gas detection process, the infrared radiation transmission process of the gas is as shown in Figure 2 The radiation energy reaching the detection system includes thermal radiation L B of an observation background, thermal radiation L gas of the gas itself, and atmospheric path radiation L pThe pupil radiation is attenuated by the transmittance of the optical system lens, the transmittance of the filter and the response attenuation of the detector window, and finally converted into a DN value (signal value, which refers to a value reflecting the amount of received radiation, and the larger the value, the stronger the energy) by the detector.

[0027] According to Figure 2 The expression of the infrared radiation transmission model of the gas, i.e., the expression of the DN value, is constructed according to the infrared radiation transmission process shown in the figure.

[0028] S2, a propagation and diffusion geometry model of the gas is constructed, wherein the propagation and diffusion geometry model contains the type of the gas, the concentration of the gas and the profile of the gas.

[0029] The propagation and diffusion geometry model is constructed by using a CFD simulation software.

[0030] Specifically, the geometric gas modeling uses a CFD simulation software to simulate the propagation and diffusion of the leaked gas. Common CFD simulation software includes ANSYS Fluent, Open FOAM and COMSOL Multiphysics, etc. Among them, OpenFOAM is an open source CFD software with high customizability and flexibility, which is suitable for various fluid dynamics simulations. Users can write their own solvers and physical models according to their needs. OpenFOAM supports multiple grid types and is suitable for complex geometries and flow conditions, and is commonly used in academic research and industrial applications. OpenFOAM is used to simulate the diffusion of the leaked gas and generate a diffusion image, and a two-dimensional propagation and diffusion geometry model containing the type, concentration and profile of the gas is obtained.

[0031] S3, a multispectral infrared gas imaging dataset is obtained based on the propagation and diffusion geometry model and the infrared radiation transmission model.

[0032] Specifically, the gas geometry profile and concentration information of the propagation and diffusion geometry model are combined with the infrared radiation transmission model, and a multispectral infrared gas imaging dataset applied to a multispectral system is obtained according to the combination result, which contains gas type and gas concentration labels.

[0033] The current infrared gas imaging dataset lacks theoretical calculation basis and only contains profile information without concentration and type information of each point. The embodiment of the present application not only simulates the gas diffusion trend, but also simulates the gas concentration according to the gas radiation transmission model, and gives the profile, concentration and type information of the gas under the multispectral image.

[0034] Therefore, the multispectral infrared gas imaging dataset is obtained according to the infrared radiation transmission model and the propagation and diffusion geometric model of the gas, the multispectral infrared gas imaging dataset is completed without emission of hazardous gas, safety is ensured, and the multispectral infrared gas imaging dataset can be made based on the gas type and concentration information, so that the demand of the dataset for deep learning training can be met.

[0035] In one embodiment, the expression of the infrared radiation transmission model is as follows:

[0036] DN=kτ srf τ opt (τ1τ2L B -τ2L bgas )(e -αCL -1)

[0037] Wherein, DN represents the signal change value caused by gas absorption (the difference between the detector signal value when there is no gas and the detector signal value when there is gas), k represents the radiation calibration coefficient, τ srf represents the response rate of the detector to different spectra, τ opt represents the transmittance of the optical system, τ1 and τ2 respectively represent the atmospheric transmittance between the observation background and the gas and between the gas and the detection system, L B represents the thermal radiation of the observation background, L bgas represents the radiance of the gas equivalent to the blackbody radiation, α is the gas absorption coefficient, C is the gas concentration, and L is the gas path optical path.

[0038] Specifically, referring to Figure 2 When there is gas in the observation background, the radiance L I after the detector is represented as:

[0039] L I =τ srf τ opt (τ1τ2τ gas L B +τ2L gas +L p )

[0040] Wherein, τ srf represents the response rate of the detector to different spectra, τ opt represents the transmittance of the optical system, mainly including the transmittance of the optical lens and the transmittance of the optical filter. τ1 and τ2 respectively represent the atmospheric transmittance between the observation background and the gas and between the gas and the detection system. The thermal radiation L gas of the gas itself can be regarded as blackbody radiation, which can be represented as (1-τ gas )L bgas . Similarly, when there is no gas in the observation background, the radiance LI2 may be expressed as:

[0041] L I2 = τ srf τ opt (τ1τ2L B + L p )

[0042] Therefore, the change in radiance ΔL caused by the gas absorption in the observation background can be expressed as:

[0043] ΔL = L I - L I2 = τ srf τ opt (τ1τ2L B - τ2L bgas )(τ gas - 1)

[0044] Secondly, according to the Beer-Lambert law, the change in radiance ΔL is related to the gas concentration, τ gas may be expressed as:

[0045] τ gas = e -αCL

[0046] wherein α is the gas absorption coefficient, C is the gas concentration, and L is the gas path length. When the background radiation temperature T B , the gas temperature T gas and the gas path length L are determined, the DN value caused by the gas absorption can be expressed as:

[0047] DN = k τ srf τ opt (τ1τ2L B - τ2L bgas )(e -αCL - 1).

[0048] In one embodiment, the step S3 can include: acquiring, by the infrared camera, an infrared background image of the observation background; combining the propagation and diffusion geometry model and the infrared radiation transmission model to obtain a simulated gas image; and superimposing the simulated gas image and the infrared background image to obtain a multispectral infrared gas image dataset.

[0049] Further, after acquiring, by the infrared camera, the infrared background image of the observation background, the method can further include: pre-processing the infrared background image.

[0050] Specifically, as Figure 3As shown, first, an infrared background image is obtained by actually shooting an application scene through an infrared camera, and then the geometric profile of the image simulated by the software in step S2 and the corresponding concentration information C are combined with the infrared radiation transmission model obtained in step S1, the concentration information C is substituted into the expression of the transmission model to calculate a simulated gas image containing gray information of the image. Finally, the multispectral infrared image dataset is obtained by superimposing the preprocessed infrared background image, containing gas profile, gas type and gas concentration label.

[0051] Wherein, the preprocessing of multispectral image is generally non-uniformity correction, denoising, adjusting dynamic range, etc. The specific implementation can be determined according to actual needs. The purpose of preprocessing is to improve the image quality as a high-quality background map.

[0052] In summary, the multispectral infrared gas imaging dataset production method of the embodiment of the present application is based on a gas radiation transmission full-link simulation model, simulates multispectral gas leakage images in a specific environment, and solves the problem that traditional methods are difficult to obtain real gas image data. At the same time, by actually shooting the complex background environment through the multispectral camera, and combining the real environmental light and weather data, the authenticity of the dangerous and hazardous gas data is enhanced. Finally, the simulated gas image and the actually shot background image are superimposed to create a rich multispectral image database, which not only ensures safety, but also effectively meets the data needs of the deep learning model.

[0053] Corresponding to the multispectral infrared gas imaging dataset production method of the above embodiment, the present application also provides a multispectral infrared gas imaging dataset production device.

[0054] Figure 4 The structure block diagram of the infrared gas imaging dataset production device of the embodiment of the present application is shown.

[0055] As shown in Figure 4 The multispectral infrared gas imaging dataset production device includes a first construction module 10, a second construction module 20 and a production module 30.

[0056] Wherein, the first construction module 10 is used to construct the infrared radiation transmission model of the gas; the second construction module 20 is used to construct the propagation and diffusion geometric model of the gas, wherein the propagation and diffusion geometric model contains the gas type, the gas concentration and the gas profile; the production module 30 is used to produce the multispectral infrared gas imaging dataset based on the propagation and diffusion geometric model and the infrared radiation transmission model.

[0057] It should be noted that the specific implementation of the multispectral infrared gas imaging dataset production device can refer to the specific implementation of the multispectral infrared gas imaging dataset production method described above, and to avoid redundancy, it will not be described in detail here.

[0058] The multi-spectral infrared gas imaging dataset production device of the embodiment of the present application obtains the multi-spectral infrared gas imaging dataset according to the infrared radiation transmission model and the propagation and diffusion geometric model of the gas, completes the production of the multi-spectral infrared gas imaging dataset under the premise of not emitting dangerous and hazardous gases, guarantees the safety, and can produce colorful multi-spectral infrared gas imaging datasets based on the gas type and concentration information, thereby meeting the requirement of the dataset of the deep learning training.

[0059] In the description of the present specification, the description of the terms "one embodiment", "some embodiments", "an example", "a specific example", or "some examples" and the like means that the specific features, structures, materials or characteristics described in connection with the embodiment or example are contained in at least one embodiment or example of the present application. In the present specification, the illustrative description of the above terms does not necessarily refer to the same embodiment or example. Although the embodiments of the present application have been shown and described above, it is understood that the above embodiments are exemplary and should not be construed as limiting the present application, and those skilled in the art can make changes, modifications, replacements and variations to the above embodiments within the scope of the present application.

Claims

1. A method of producing a multispectral infrared gas imaging data set, comprising: The method comprises the following steps: constructing an infrared radiation transmission model of the gas; constructing a propagation and diffusion geometry model of the gas, wherein the propagation and diffusion geometry model comprises a gas species, a gas concentration, and a gas profile; obtaining a multispectral infrared gas imaging dataset based on the propagation and diffusion geometry model and the infrared radiation transmission model, an expression of the infrared radiation transmission model is: ； where DN represents the signal change value due to gas absorption, k represents a radiation calibration coefficient, represents the response rate of the detector to different spectra, represents the transmittance of the optical system, , respectively represent the atmospheric transmittances between the observation background and the gas, and between the gas and the detection system, represents the thermal radiation of the observation background, represents the radiance of the gas equivalent to the blackbody radiation, is the gas absorption coefficient, C is the gas concentration, and L is the optical path of the gas.

2. The method of claim 1, wherein the method further comprises: the propagation and diffusion geometry model is constructed by using a CFD simulation software.

3. The method of claim 1, wherein the method further comprises: The method for obtaining a multispectral infrared gas imaging dataset based on the propagation and diffusion geometry model and the infrared radiation transmission model comprises: capturing an infrared background image of an observation background by using an infrared camera; combining the propagation and diffusion geometry model and the infrared radiation transmission model to obtain a simulated gas image; superimposing the simulated gas image and the infrared background image to obtain an infrared gas image dataset.

4. The method of claim 1, wherein, After capturing the infrared background image of the observation background by using the infrared camera, the method further comprises: preprocessing the infrared background image.

5. An apparatus for producing a multispectral infrared gas imaging data set, characterized by The method comprises: a first construction module configured to construct an infrared radiation transmission model of the gas; a second construction module configured to construct a propagation and diffusion geometry model of the gas, wherein the propagation and diffusion geometry model comprises a gas species, a gas concentration, and a gas profile; a production module configured to obtain a multispectral infrared gas imaging dataset based on the propagation and diffusion geometry model and the infrared radiation transmission model, an expression of the infrared radiation transmission model is: ； where DN represents a signal change value due to gas absorption, k represents a radiation calibration coefficient, represents a response rate of the detector to different spectra, represents a transmittance of the optical system, , respectively represent atmospheric transmittances between the observation background and the gas, and between the gas and the detection system, represents thermal radiation of the observation background, represents a radiance of the gas equivalent to blackbody radiation, is a gas absorption coefficient, C is a gas concentration, and L is a gas path optical length.

Images