Multispectral infrared imaging gas detection method based on microlens and filter array

Abstract

The application discloses a multispectral infrared imaging gas detection method based on a microlens and a filter array, which is suitable for multispectral infrared gas detection of industrial dangerous gas leakage and belongs to the technical fields of infrared imaging and gas detection. A multispectral infrared imaging system is formed based on a single uncooled infrared focal plane detector, a microlens array and a partition filter array, a plurality of spectral intervals correspond to optical absorption peaks of specific gases respectively; the single detector is capable of uniform response characteristics, which is convenient for calculating spectral normalized absorption rate curves of each waveband and comparing with a special gas spectral normalized absorption rate curve database to determine the type of the current gas. When applied to the field of gas detection, the multispectral infrared imaging gas detection method can solve the problem that the existing infrared imaging system cannot effectively distinguish the type of the gas, improve the gas detection capability of the infrared imaging system, and the method can be used for automatic monitoring of gas leakage and improve the automatic detection level of the multispectral infrared imaging gas detection system.

Description

TECHNICAL FIELD

[0001] The present application relates to a kind of multispectral infrared imaging gas detection method based on microlens and filter array, in particular to multispectral infrared imaging system technology, it is suitable for multispectral infrared gas detection of industrial dangerous gas leakage, belongs to infrared imaging and gas detection technical field. BACKGROUND

[0002] Industrial dangerous gas leakage detection technology is currently widely used in industrial field, and is usually based on the physical or chemical characteristics of the leakage gas for leakage gas detection. The traditional gas detection sensor uses a contact measurement method, but it needs to be deployed in a large number of potential leakage points, and cannot directly locate the leakage point. In recent years, the infrared imaging gas detection technology based on the infrared spectral absorption characteristics of the leakage gas designs a detector for detection in a specific infrared waveband of gas spectral absorption, which can directly display the leakage gas trace and is convenient for judging the leakage scale and locating the leakage point, and the use is more intuitive.

[0003] Currently, the main application is a gas detection device based on a single waveband infrared imaging system, which can observe the dangerous gas leakage tail that the visible light waveband imaging system cannot capture, but in actual industrial scenes, the leakage gas is usually not only one kind, and includes safe gases such as water vapor and carbon dioxide, which will interfere with the infrared image and cause the system to be unable to distinguish between dangerous gas and safe gas, posing a challenge to the existing single-spectrum infrared imaging gas detection technology.

[0004] The multispectral infrared imaging technology usually has several to dozens of infrared imaging spectrums, which can divide several independent wavebands in the 3-14 μm waveband range for imaging. For the characteristics that a certain gas has multiple infrared absorption peaks in the 3-14 μm waveband range, targeted imaging is performed. For example, methane gas usually has absorption in the 3.2-3.4 μm and 6-8 μm wavebands. By configuring a multispectral infrared imaging system to divide the 3-14 μm waveband range into multiple wavebands with equal width, the gas cloud trace of methane gas existing in the 3-4 μm and 6-7 μm wavebands can be used to determine that the current gas is methane.

[0005] The dangerous gas detection technology based on the multispectral infrared imaging technology can effectively distinguish various leakage gases for the user of the device and exclude the interference of common gases such as water vapor and carbon dioxide, effectively improving the detection capability of the current infrared imaging gas detection technology. In order to improve the level of automatic monitoring of dangerous gas, a scheme for automatically distinguishing the type of leakage gas based on the information difference existing in the multispectral infrared image needs to be designed based on the multispectral infrared imaging system, which is convenient for the implementation of the automatic alarm technology of the device in actual use scenarios and improves the automation level of the dangerous gas monitoring system. SUMMARY

[0006] The purpose of this invention is to provide a multispectral infrared imaging gas detection method based on microlenses and filter arrays, which is used for detecting gas cloud targets in hazardous gas leaks.

[0007] The objective of this invention is achieved through the following technical solution:

[0008] The multispectral infrared imaging gas detection method based on microlenses and filter arrays disclosed in this invention includes the following steps:

[0009] S1. Construct a multispectral infrared imaging system based on a microlens array and partitioned filters. The multispectral infrared imaging system includes an uncooled infrared focal plane detector, a microlens array corresponding to the focal plane array, and partitioned filters. The microlens array and multiple partitioned filters are in one-to-one correspondence. Scene light beams pass through multiple partitioned filters to form multiple beams of different spectra. Each beam of light is focused onto the corresponding area of ​​the uncooled infrared focal plane detector after passing through the corresponding microlens and is acquired by the uncooled infrared focal plane detector. That is, all spectra are imaged on the same detector. In subsequent steps, when analyzing the spectral absorption characteristics of gas cloud targets in each band, the imaging characteristics of the multispectral images are consistent, and the spectral characteristics of gas cloud targets can be accurately analyzed based on the image grayscale. Each image output by the uncooled infrared focal plane detector is composed of multiple sub-images.

[0010] S2. The multispectral infrared imaging system built using S1 is used to acquire images of hazardous gas leaks. An image of the hazardous area is acquired, and multiple sub-images in the image are registered using multispectral imaging based on the features of the scene. Then, the multiple sub-images are preprocessed to eliminate stripe noise and Gaussian noise in the image. Finally, the multiple sub-images are cropped to a uniform size and then stitched together to obtain a complete image Y containing each channel.

[0011] S3. Statistical modeling of the complete image Y obtained in S2 is performed using the GMM detection method to obtain the static background image X. Then, the current frame Y of image Y is used... t With image X, the current frame X t The difference is used to obtain the grayscale image of the cloud target in each channel in the current frame G. t Based on the current frame G of the grayscale image of the cloud target. t Select the cloud region, and then process each pixel G within the cloud region. t Calculate the average value of (i,j) to obtain the average gray level P of gas cloud absorption in the current spectral channel. k Then, the average gray level of the gas cloud absorption in each spectral channel is normalized to obtain the spectral absorbance curve Q for each band.

[0012] S4. Locate the potentially leaked gas A from publicly available or self-collected gas standard spectral databases, and then obtain the high-resolution spectral curve T of gas A.o ; according to the spectral range corresponding to each sub-image in S2, the curve T o The spectral absorption rate corresponding to the spectral range is averaged, that is, the average gas absorption rate of each sub-image corresponding to the waveband is obtained; the average gas absorption rate corresponding to each sub-image is normalized to obtain a normalized curve T1;

[0013] S5, the method for obtaining N normalized curves T i (i=1, 2,..., N), that is, N normalized curves T i (i=1, 2,..., N) corresponding to N kinds of gases that may exist in S2 are obtained.

[0014] S6, the curve T i is matched with the curve Q obtained in S3 one by one to find the kind of leaked gas, that is, the multi-spectral infrared imaging gas detection is realized based on the microlens array and the partition filter.

[0015] The specific mode is that the curve T i is compared with the curve Q, the curve T i with the highest correlation coefficient is taken, and the gas corresponding to the curve T i is the leaked gas, that is, the multi-spectral infrared imaging gas detection is realized based on the microlens array and the partition filter.

[0016] Beneficial effects:

[0017] 1. The multi-spectral infrared imaging gas detection method based on the microlens array and the partition filter disclosed in the application is based on a single non-cooled infrared focal plane detector and a microlens array and a partition filter array to form a multi-spectral infrared imaging system, utilizes the unified ability of the single detector response characteristics, is convenient for calculating the spectral normalized absorption rate curve of each waveband and comparing with the special gas spectral normalized absorption rate curve database, and judges the kind of current gas. When applied to the field of gas detection, the problem that the existing infrared imaging system cannot effectively distinguish the kind of gas is solved, the gas detection capability of the infrared imaging system is improved, and the proposed method can perform automatic monitoring of gas leakage, and the automatic detection level of the multi-spectral infrared imaging gas detection system is improved.

[0018] 2. The multi-spectral infrared imaging gas detection method based on the microlens array and the partition filter disclosed in the application is multi-spectral waveband design for common gases in the current petroleum and chemical industry, is based on the spectral characteristics of the leaked gas in the infrared multi-spectral channel, extracts the gray scale information of the gas cloud target based on the foreground detection method, constructs the spectral absorption rate curve of the gas cloud target in each channel, calculates the curve T i and the curve Q Pearson correlation coefficient, takes the curve T i with the highest correlation coefficient, and the curve Ti The corresponding gas is the leakage gas, thereby improving the accuracy and efficiency of identifying common leakage gas types. The application can effectively distinguish gas types during actual detection, exclude the interference of safe gases such as carbon dioxide and water vapor, judge the gas type of the current gas cloud target, and improve the detection capability and practicability of the infrared imaging dangerous gas detection system. BRIEF DESCRIPTION OF DRAWINGS

[0019] Figure 1 A flowchart of a multispectral infrared imaging gas detection method based on a microlens array and a partitioned filter according to an embodiment of the application is shown in the figure.

[0020] Figure 2 An imaging schematic diagram of a multispectral infrared imaging gas detection system based on a microlens array and a partitioned filter according to an embodiment of the application for industrial scene gas leakage detection is shown in the figure.

[0021] Figure 3 A multispectral image of difluoroethane leakage collected according to an embodiment of the application is shown in the figure. DETAILED DESCRIPTION

[0022] To better illustrate the purposes and advantages of the application, the content of the application is further described below in combination with the figures and examples.

[0023] Embodiment 1

[0024] The embodiment discloses a multispectral infrared imaging gas detection method based on a microlens array and a partitioned filter, which aims to detect the leakage of a pipeline in a chemical plant. The gases that may be released by the pipelines with leakage risks in the chemical plant include methane, ethylene, hydrogen sulfide, hydrogen chloride, ammonia, etc. Assuming that ethylene gas leakage occurs at present, the detection target is to automatically identify the gas plume in the multispectral image as ethylene based on the proposed method. The implementation steps are as shown in the figure. Figure 1

[0025] As shown in the figure, the multispectral infrared imaging gas detection method based on a microlens and a filter array disclosed in the embodiment has the following specific implementation steps: Figure 1

[0026] S1, build a multispectral infrared imaging system based on a microlens array and a partitioned filter, mainly including a non-cooled infrared focal plane detector, a microlens array corresponding to the focal plane array, and a partitioned filter, and the imaging light path is as shown in the figure. Figure 2 ​​As shown. The microlens array and the multiple partition filters are one-to-one corresponding; the scene light beam passes through the multiple partition filters to form multiple beams of different spectra; each beam of light is focused to the corresponding area of the uncooled infrared focal plane detector after passing through the corresponding microlens, and is collected by the uncooled infrared focal plane detector, i.e. each spectrum is imaged on the same detector, and the imaging characteristics of the multispectral image are consistent when analyzing the spectral absorption characteristics of the gas cloud target in the subsequent step, so that the spectral characteristics of the gas cloud target can be accurately analyzed according to the image gray scale; each image output by the uncooled infrared focal plane detector after collection is composed of multiple sub-images.

[0027] The microlens array includes 3x2 lens groups, wherein a single lens group is composed of two infrared lenses. The spectral bands of the 3x2 filters are 3-5 μm (channel 1), 6-8 μm (channel 2), 7-9 μm (channel 3), 9-11 μm (channel 4), 10-12 μm (channel 5), and 12-14 μm (channel 6).

[0028] S2, using the multispectral infrared imaging system obtained in S1 to collect images of gas leakage in a dangerous area; collecting an image of a dangerous area, as shown in FIG. 2. Figure 3 According to the features in the scene, the six sub-images in the image are registered; then the six sub-images are preprocessed to eliminate the stripe noise and Gaussian noise in the image; then the six sub-images are cropped to a uniform size and spliced to obtain a complete image Y containing each channel, as shown in FIG. 3. Figure 3

[0029] S3, using the GMM detection method to statistically model the complete image Y obtained in S2 to obtain a static background image X, and then using the current frame Y t of the image Y t to subtract the current frame X t of the image X to obtain the current frame G t of the gas cloud target gray image in each channel; based on the current frame G t of the gas cloud target gray image, selecting a gas cloud area, and averaging each pixel G k (i,j) in the gas cloud area to obtain the average gas cloud absorption gray P o of the current spectral channel; and then gray normalizing the average gas cloud absorption gray of each spectral channel to obtain the spectral absorption rate curve Q.

[0030] S4, obtaining five curves corresponding to the five gases that may exist in the leakage in S2; finding the five gases of methane, ethylene, hydrogen sulfide, hydrogen chloride and ammonia from the public National Institute of Standards and Technology (NIST) gas standard spectrum database, and then obtaining the high-resolution spectral curve T of each gas.o ; according to the spectral range corresponding to each sub-image in S2, the curve T o The spectral absorption rate corresponding to the spectral range is averaged, that is, the average gas absorption rate of each sub-image corresponding to the waveband is obtained; the average gas absorption rate corresponding to each sub-image is normalized to obtain a normalized curve T1.

[0031] S5, five normalized curves T i (i = 1, 2, ··· 5) are obtained by using the method of S4.

[0032] S6, five curves T i obtained by S5 are matched with the curve Q obtained by S3 to find the leaked gas species; the specific method is to calculate the Pearson correlation coefficient of the curve T i and the curve Q, and take the curve T i with the highest correlation coefficient, and the gas corresponding to the curve T i is the leaked gas.

[0033] The above specific description further details the purpose, technical scheme and beneficial effects of the application. It should be understood that the above description is only a specific embodiment of the application and is not used to limit the protection scope of the application. Any modification, equivalent replacement, improvement, etc. made within the spirit and principles of the application shall be included in the protection scope of the application.

Claims

1. A method of multispectral infrared imaging gas detection based on microlens and filter array, characterized in that: Comprise the following steps, S1, build a multispectral infrared imaging system based on microlens array and partition filter, the multispectral infrared imaging system includes uncooled infrared focal plane detector, microlens array corresponding to focal plane array and partition filter;Microlens array and multiple partition filters one to one;Scene beam passes through multiple partition filters, forms multiple light beams of different spectra;Each light is focused to the corresponding area of uncooled infrared focal plane detector after passing through the corresponding microlens, is collected by uncooled infrared focal plane detector, that is, each spectrum is imaged on the same detector, and the imaging characteristics of multispectral image are consistent when analyzing the spectral absorption characteristics of gas cloud target in subsequent steps, and the spectral characteristics of gas cloud target are accurately analyzed according to image gray scale;Each image output by uncooled infrared focal plane detector after collection is composed of multiple sub-images; S2, the multispectral infrared imaging system built by S1 is used to collect images of dangerous gas leakage;Collect an image of a dangerous area, and perform multispectral image registration on multiple sub-images in the image according to the characteristics in the scene;Then, the multiple sub-images are preprocessed to eliminate stripe noise and Gaussian noise in the image;Then, the multiple sub-images are cropped to a uniform size and spliced to obtain a complete image Y containing each channel. S3, using GMM detection method to S2 get the complete image Y statistical modeling, get static background image X, then use the current frame Y t of image Y t Difference to obtain each channel gas cloud target gray image current frame G t ; Based on the current frame G of the gas cloud target gray image t , selecting the gas cloud area, and averaging each pixel G t (i,j) in the gas cloud area to obtain the average gas cloud absorption gray P k of the current spectral channel; and then normalizing the average gas cloud absorption gray of each spectral channel to obtain the spectral absorption rate curve Q of each waveband; S4, finding the possible leaked gas A from the public or self-collected gas standard spectrum database, and then obtaining the high-resolution spectrum curve T of the gas A o ; according to the spectrum range corresponding to each sub-image in S2, averaging the spectral absorption rate of the spectrum range corresponding to the curve T o , to obtain the average gas absorption rate of each sub-image corresponding to the wave band; normalizing the average gas absorption rate corresponding to each sub-image to obtain the normalized curve T1; S5, N normalization curves T are obtained by using the method of S4 i , i = 1, 2...N, that is, N normalization curves T corresponding to N kinds of gases that S2 may leak are obtained i , i = 1, 2...N; S6, the curve T obtained from S5 i With the curve Q obtained from S3, the leaked gas species is found, that is, the multispectral infrared imaging gas detection is realized based on the microlens array and the partition filter.

2. The method for multispectral infrared imaging gas detection based on microlens and filter array according to claim 1, characterized in that: In S6, the curve T i is compared with the curve Q, and the curve T i with the highest correlation coefficient is taken i The corresponding gas is the leakage gas, that is, the multi-spectral infrared imaging gas detection is realized based on the microlens array and the partition filter.

Images