Optimization method and system for SF6 gas detection under multi-temperature environment

Abstract

The present application relates to the technical field of gas detection, and relates to an optimization method and system for SF6 gas detection under multiple temperature environments, wherein the optimization method comprises: simultaneously acquiring a double-band infrared image of a to-be-detected scene, the double-band infrared image comprising a detection band infrared image and a reference band infrared image; preprocessing the double-band infrared image; registration processing the preprocessed double-band infrared image; performing difference operation on the registration-processed double-band infrared image to obtain a difference image; determining a dynamic segmentation threshold according to a current environmental temperature, and performing threshold segmentation on the difference image according to the dynamic segmentation threshold to identify an SF6 gas leakage area. The optimization method and system can solve the problem of performance decline of a fixed threshold caused by background radiation change under different environmental temperatures by determining a dynamic segmentation threshold according to a current environmental temperature, so that stable and reliable detection sensitivity can be maintained within a wide temperature range.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of gas detection, and particularly to an SF6 gas detection optimization method and system under multiple temperature environments. BACKGROUND

[0002] Sulfur hexafluoride (SF6) is widely used in gas insulated switchgear, circuit breakers, disconnectors and other high-voltage power equipment due to its excellent electrical insulation and arc extinguishing performance, significantly improving the safe and economic operation level of the power system and improving the frequent maintenance situation of traditional oil-filled electrical equipment. However, once the SF6 gas insulated equipment in operation leaks, not only will it cause the internal insulation performance to decline, threatening the safe and stable operation of the power equipment, but also its partial decomposition products are toxic and corrosive, which may endanger the health and safety of maintenance personnel in close proximity. Therefore, in order to achieve rapid and accurate detection and positioning of SF6 gas leakage, protect the operation of the power system and personnel safety, it is of great significance to develop efficient and reliable SF6 leakage detection technology.

[0003] Traditional SF6 power equipment leak detection methods mainly include vacuum leak detection method, soap bubble method and halogen effect leak detection method. The vacuum leak detection method and the soap bubble method are usually only suitable for leak detection after the equipment is shipped or overhauled, and have limitations such as poor precision, long time consumption, high implementation cost, and inability to realize live detection. The halogen effect leak detection method can detect SF6 gas regions of certain concentration, but it is difficult to accurately locate the leakage point and requires power-off operation, which affects the normal operation of the power system. In addition, the traditional method usually requires the test personnel to check the sealed parts in close proximity, which is time-consuming and inefficient.

[0004] Infrared imaging leak detection technology, as a non-contact, long-distance live detection method, has become a research hotspot. This technology is based on the characteristic infrared absorption peak of SF6 gas near 10.6 μm. By tuning the working wavelength of the infrared imaging system to the SF6 absorption spectrum through a narrow-band filtering device, the originally colorless and odorless SF6 gas is visualized in the infrared image in the form of smoke, thereby realizing visual identification and positioning of the gas. The infrared imaging method not only can capture the leakage dynamics in real time, but also can quantitatively analyze the gas concentration, and has the advantages of high detection efficiency and accurate positioning. Currently, gas leak detectors such as GF306 represented by the United States FLIR company have been widely used internationally, however, affected by technology monopoly, such equipment is expensive and limited in domestic introduction. Domestic related products are still in the early stage of research and development, and there is a gap in measurement range and stability compared with foreign products, which seriously restricts the independent safety of China's power industry chain.

[0005] Existing infrared imaging leak detection methods are mostly based on wide-band infrared imaging analysis, covering one absorption characteristic peak of SF6, and judging the leakage by comparing the differences between consecutive frames of images, which has limited detection accuracy. In addition, the traditional method is easily affected by background radiation fluctuations in complex temperature environments, resulting in a decrease in detection sensitivity. In order to improve the detection accuracy and adaptability, the infrared spectral characteristics of SF6 gas should be fully explored, and a new gas recognition method combining multi-band imaging and intelligent image processing should be explored.

[0006] In recent years, domestic and foreign scholars have carried out a number of studies on SF6 gas infrared detection. For example, by introducing a Michelson interferometer, qualitative and quantitative analysis of SF6 decomposition components is realized, but the system structure is complex and not suitable for on-site rapid detection; there are also studies on the development of small SF6 gas sensors based on non-dispersive infrared (NDIR) sensing technology, but there are problems such as light source aging, noise interference, etc., which affect the long-term stability of the system. In terms of image processing, background difference, frame difference, optical flow method and Gaussian mixture model motion target detection algorithms can enhance the visual saliency of gas leakage areas, but false positives may occur under conditions of camera shaking or complex background.

[0007] In summary, the existing SF6 gas infrared imaging detection technology still has the following shortcomings: first, it is mostly based on single-band or wide-band imaging, which does not fully utilize the multi-spectral characteristics of SF6 and is easily affected by other gases; second, it lacks adaptability to temperature changes, and fixed threshold segmentation methods have unstable detection performance under different environmental temperatures; third, the system cost is high and the core technology is subject to others, which restricts its popularization and application in the domestic power industry.

[0008] Therefore, the present application provides an optimization method and system for SF6 gas detection in multiple temperature environments. SUMMARY

[0009] To this end, the purpose of the present application is to overcome the problem of detection performance decline caused by background radiation changes under different environmental temperatures in the prior art with fixed threshold, so that stable and reliable detection sensitivity can be maintained within a wide temperature range (such as 23℃ to 60℃, 30℃ to 60℃).

[0010] To solve the above technical problems, on the one hand, the present application provides an optimization method for SF6 gas detection in multiple temperature environments, comprising the following steps:

[0011] Simultaneously acquiring a dual-band infrared image of the scene to be detected, the dual-band infrared image comprising a detection band infrared image and a reference band infrared image;

[0012] Pretreating the dual-band infrared image;

[0013] Registering the pretreated dual-band infrared image;

[0014] Differential operations are performed on the registered dual-band infrared image to obtain a differential image;

[0015] Based on the current ambient temperature, a dynamic segmentation threshold is determined, and the differential image is then segmented according to the dynamic segmentation threshold to identify SF6 gas leakage areas.

[0016] Preferably, the preprocessing includes at least one of dark current correction, infrared radiometric correction, and image noise reduction;

[0017] The dark current correction includes subtracting the dark current image from the dual-band infrared image to eliminate the noise of the infrared imaging system itself.

[0018] The infrared radiation correction includes: correcting the dual-band infrared image based on the two-point linear calibration method to eliminate the thermal radiation of the infrared imaging system itself;

[0019] The image denoising includes: performing block matching, three-dimensional transformation, threshold shrinkage, and inverse transformation processing on dual-band infrared images based on a three-dimensional filtering algorithm.

[0020] Preferably, the infrared radiation correction includes;

[0021] Preliminary radiometric correction of dual-band infrared images is performed using the response function and radiometric bias of the infrared imaging system.

[0022] Secondary radiometric correction was performed on the pre-corrected dual-band infrared image using the inverse function of Planck's formula.

[0023] The secondary radiation correction includes the following formula:

[0024] ;

[0025] In the formula, After secondary radiation correction Brightness temperature of infrared images in the spectral band; It is Planck's first radiation constant; is Planck's second radiation constant.

[0026] Preferably, the preliminary radiation correction includes the following formula:

[0027] ;

[0028] In the formula, For bands; for Raw spectral measurements of the band infrared image; This refers to the radiation bias caused by the thermal radiation of the infrared imaging system itself. Let this be the response function of the infrared imaging system; the preliminary radiation correction spectrum radiance of the waveband infrared image;

[0029] Before the preliminary radiation correction step, the infrared radiation correction further comprises:

[0030] obtaining original spectrum measurement values of the high-temperature blackbody and the low-temperature blackbody by two-point calibration of the infrared imaging system using the high-temperature blackbody and the low-temperature blackbody;

[0031] calculating theoretical spectrum radiance corresponding to the high-temperature blackbody and the low-temperature blackbody respectively based on Planck blackbody radiation law;

[0032] calculating a response function and a radiation bias of the infrared imaging system according to the original spectrum measurement values and the theoretical spectrum radiance;

[0033] wherein the temperature of the high-temperature blackbody is not lower than the highest temperature in a range of the environment temperature to be measured, and the temperature of the low-temperature blackbody is not higher than the lowest temperature in the range of the environment temperature to be measured.

[0034] Preferably, the registration processing comprises:

[0035] selecting an area in the preprocessed reference waveband infrared image as a matching template;

[0036] sliding the matching template on the preprocessed detection waveband infrared image, and calculating a normalized cross-correlation coefficient between the matching template and a current window area of the detection waveband infrared image at each sliding position;

[0037] in response to the normalized cross-correlation coefficient being maximum, determining a corresponding window area as a matching position of the matching template in the detection waveband infrared image;

[0038] calculating a spatial transformation parameter between the detection waveband infrared image and the reference waveband infrared image according to the matching template and the matching position, and performing geometric transformation on the detection waveband infrared image according to the spatial transformation parameter, so that the detection waveband infrared image is spatially aligned with the reference waveband infrared image.

[0039] Preferably, the normalized cross-correlation coefficient comprises the following formula:

[0040] ;

[0041] wherein, is a normalized cross-correlation coefficient at the position; is the matching template; is the matching template; a sub-image block with (x, y) as the top-left corner point and the same size as the matching template in the pre-processed detection waveband infrared image; is and a covariance of is a standard deviation of pixel values of is a standard deviation of pixel values of the matching template.

[0042] Preferably, the difference operation comprises:

[0043] performing a pixel-level difference operation on the registered reference waveband infrared image and the registered detection waveband infrared image.

[0044] Preferably, the threshold segmentation of the difference image according to the dynamic segmentation threshold comprises:

[0045]

[0046] wherein, is a binary result image obtained by threshold segmentation of the difference image, wherein 1 represents a gas leakage area and 0 represents a background; is the registered reference waveband infrared image; is the registered detection waveband infrared image; is a dynamic segmentation threshold;

[0047] The dynamic segmentation threshold is determined according to the current ambient temperature, comprising:

[0048]

[0049] wherein, is the current ambient temperature; is a temperature base parameter determined according to the highest temperature of the ambient temperature, and ; is a sensitivity coefficient determined according to the lowest temperature of the ambient temperature and the temperature base parameter, and ; is a temperature decay rate determined according to the lowest temperature, the highest temperature of the ambient temperature and the temperature base parameter, and .

[0050] Preferably, the current ambient temperature is 23℃ to 60℃.

[0051] Preferably, the current ambient temperature is 30℃ to 60℃.

[0052] Preferably, the center wavelength of the detection waveband is 10500nm to 10630nm, and the center wavelength of the reference waveband is 8050nm to 8360nm. ​​

[0053] In another aspect, the present application provides an optimization system for SF6 gas detection in a multi-temperature environment, comprising:

[0054] A collection module is configured to simultaneously acquire a dual-band infrared image of a scene to be detected, wherein the dual-band infrared image comprises a detection band infrared image and a reference band infrared image;

[0055] A preprocessing module is configured to preprocess the dual-band infrared image;

[0056] A registration module is configured to register the dual-band infrared image after preprocessing;

[0057] A difference module is configured to perform a difference operation on the dual-band infrared image after registration to obtain a difference image;

[0058] An identification module is configured to determine a dynamic segmentation threshold according to a current ambient temperature, perform threshold segmentation on the difference image according to the dynamic segmentation threshold, and identify an SF6 gas leakage area.

[0059] The above technical solutions of the present application have the following beneficial effects compared with the prior art:

[0060] The optimization method and system for SF6 gas detection in a multi-temperature environment according to the present application can highlight the radiation difference caused by SF6 gas absorption by simultaneously acquiring infrared images of a detection band (SF6 absorption peak) and a reference band (background band) and performing a difference operation thereon, thereby effectively distinguishing gas leakage from slow changes in the background environment and reducing the false alarm rate. In addition, in order to solve the problem of performance degradation caused by changes in background radiation at different ambient temperatures due to a fixed threshold, the present application determines a dynamic segmentation threshold according to the current ambient temperature, so that stable and reliable detection sensitivity can be maintained within a wide temperature range (such as 23℃ to 60℃, 30℃ to 60℃). BRIEF DESCRIPTION OF DRAWINGS

[0061] In order to make the content of the present application more easily understood, the present application will be further described in detail below according to specific embodiments of the present application and in conjunction with the accompanying drawings.

[0062] Figure 1 is a flowchart of the optimization method for SF6 gas detection in a multi-temperature environment provided by an embodiment of the present application.

[0063] Figure 2 is a comparison chart of dark current correction effects provided by an embodiment of the present application.

[0064] Figure 3 is a comparison chart of image noise reduction effects provided by an embodiment of the present application.

[0065] Figure 4An infrared image registration processing schematic diagram provided by an embodiment of the present application.

[0066] Figure 5 An SF6 gas leakage area identification effect diagram (a binary result image) provided by an embodiment of the present application.

[0067] Figure 6 A detection result comparison diagram of SF6 gas under different standard blackbody background temperatures (30℃, 40℃, 50℃ and 60℃) provided by an embodiment of the present application.

[0068] Figure 7 A detection result and gas position label comparison diagram of SF6 gas under a 40℃ blackbody background provided by an embodiment of the present application.

[0069] Figure 8 A structure block diagram of an SF6 gas detection optimization system under a multi-temperature environment provided by an embodiment of the present application. DETAILED DESCRIPTION

[0070] The present application will be further described below in conjunction with the drawings and specific embodiments, so that those skilled in the art can better understand the present application and implement it, but the embodiments are not intended to limit the present application.

[0071] The words expressing position and direction described in the present application are all described by taking the drawings as examples, but changes can also be made according to needs, and the changes are all included in the protection scope of the present application.

[0072] In order to solve the problem of detection performance decline caused by background radiation change under different environmental temperatures due to fixed threshold, the present application introduces an SF6 gas detection optimization method and optimization system under a multi-temperature environment, which determines a dynamic segmentation threshold according to the current environmental temperature, so that stable and reliable detection sensitivity can be maintained within a wider temperature range (such as 23℃ to 60℃, 30℃ to 60℃).

[0073] Further, the optimization method and the optimization system of the present application can highlight the radiation difference caused by SF6 gas absorption by acquiring infrared images of the detection band (SF6 absorption peak) and the reference band (a band similar to the SF6 absorption peak in terms of environmental background radiation characteristics) at the same time and performing difference operation, thereby effectively distinguishing gas leakage from slow changes in the background environment and reducing the false alarm rate. Compared with the frame difference method which needs to analyze continuous multiple frames of images, the present application is based on instantaneous difference of dual-band images, does not need to rely on the motion characteristics of the gas cloud, and can realize real-time detection and alarm of single-frame images. In addition, by selecting the reference band (a band similar to the SF6 absorption peak in terms of environmental background radiation characteristics), the present application can effectively suppress the interference of water vapor and other gases with absorption characteristics in the infrared band, and improve the specificity of SF6 gas recognition.

[0074] Reference Figure 1 The optimization method for SF6 gas detection in a multi-temperature environment of the present application includes steps SS1 to SS5.

[0075] Step SS1: simultaneously acquiring dual-band infrared images of the scene to be detected.

[0076] In application, the dual-band infrared images of the scene to be detected are collected by an infrared imaging system. The dual-band infrared images include a detection band infrared image and a reference band infrared image. The center wavelength of the detection band is 10500 nm to 10630 nm, and the center wavelength of the reference band is 8050 nm to 8360 nm.

[0077] In actual application, the infrared imaging system adopts a parallel dual-optical-path structure to ensure that the two optical channels can simultaneously image the same scene to be detected. The infrared imaging system includes a binocular baseline support, two infrared cameras, two identical infrared front lenses (lens groups), two narrow-band interference filters (the film system material can be germanium and zinc sulfide, and the substrate material can be silicon), and two uncooled infrared detector cores. Specifically, the binocular baseline support is a rigid support structure that fixes the two infrared cameras side by side to form a stable optical baseline; an identical infrared lens group is installed in front of each camera as a front lens of the imaging system; in each optical path, a narrow-band interference filter is integrated behind the lens group and in front of the detector, and the filter is installed close to the detector window. The two uncooled infrared detector cores are photosensitive elements that are connected to the filters inside the cameras and arranged in parallel to receive infrared signals from their respective optical paths. Further, the two optical channels are a detection channel and a reference channel.

[0078] In actual implementation, the center wavelength of the detection channel is 10500 nm to 10630 nm, the bandwidth is 517 nm, and the peak transmittance is greater than or equal to 53.1%; the center wavelength of the reference channel is 8050 nm to 8360 nm, the bandwidth is 602 nm, and the peak transmittance is greater than or equal to 60.2%.

[0079] Step SS2: pre-processing the dual-band infrared image.

[0080] In application, the pre-processing of the present application at least includes one of dark current correction, infrared radiation correction and image noise reduction. Preferably, the pre-processing of the present application includes dark current correction, infrared radiation correction and image noise reduction. The dark current correction can eliminate the fixed pattern noise caused by the dark current noise of the infrared imaging system (such as a detector) itself, improve the uniformity and contrast of the image, and lay a foundation for subsequent difference calculation. The infrared radiation correction can realize the quantification of data by converting the gray value of the image into a physical quantity (such as spectral radiance or brightness temperature) reflecting the real radiation characteristics of the target, so that the image data collected at different times and by different devices has comparability, and the influence of the thermal radiation of the infrared imaging system (such as a detector) itself is eliminated. The image noise reduction can effectively suppress random noise and stripe noise while better preserving the edge and detail information of the image through three-dimensional block matching filtering, significantly improving the signal-to-noise ratio of the difference image, and making it easier to detect weak gas signals.

[0081] In some embodiments, the dark current is also called the dark level, that is, the current when the environment is completely dark, which is used to define the signal level corresponding to the image data of 0. The dark current correction of the present application includes: subtracting the dark current image from the dual-band infrared image to eliminate the noise of the infrared imaging system itself. Specifically, an ice block is placed in front of the camera to approximate the problem to absolute zero, thereby obtaining the dark current pixel value of the camera, and then the dark current correction is completed by subtracting the dark current pixel value from the pixel value of the infrared image collected in real time. Figure 2 b is the original infrared image without dark current correction, which is affected by the camera noise and the reflection of the filter, the background noise of the camera is too large, and the gray distribution of the original infrared image is obviously uneven. By shooting the dark current image (i.e. Figure 2 a) the overall uneven distribution is obtained, and then the original infrared image is corrected to obtain Figure 2 c is the infrared image after dark current correction, which makes the background unevenness significantly improved, and the contrast of the detected target more significant.

[0082] In some embodiments, the infrared radiation correction of the present application performs two radiation corrections, such as a preliminary radiation correction and a secondary radiation correction. Further, the infrared radiation correction of the present application includes correcting the dual-band infrared image based on a two-point linear calibration method to eliminate the thermal radiation of the infrared imaging system itself. Specifically, the infrared radiation correction of step SS2 includes steps SS21 to SS25.

[0083] Step SS21: Two-point calibration of the infrared imaging system using a high-temperature blackbody and a low-temperature blackbody to obtain raw spectral measurements of the high-temperature blackbody and the low-temperature blackbody.

[0084] In application, the temperature of the high-temperature blackbody is not lower than the highest temperature in the range of the environment temperature to be measured, and the temperature of the low-temperature blackbody is not higher than the lowest temperature in the range of the environment temperature to be measured.

[0085] In practical application, the range of the environment temperature to be measured is 23℃ to 60℃. Preferably, the range of the environment temperature to be measured is 30℃ to 60℃.

[0086] Step SS22: Calculating the theoretical spectral radiance corresponding to the high-temperature blackbody and the low-temperature blackbody, respectively, based on the Planck blackbody radiation law.

[0087] Step SS23: Calculating the response function and the radiation bias of the infrared imaging system according to the raw spectral measurements and the theoretical spectral radiance.

[0088] In application, the response function and the radiation bias of the infrared imaging system include the following formula:

[0089] ;

[0090] ;

[0091] In the formula, is the response function of the infrared imaging system; is the raw spectral measurement of the high-temperature blackbody; is the raw spectral measurement of the low-temperature blackbody; is the theoretical spectral radiance of the high-temperature blackbody; is the theoretical spectral radiance of the low-temperature blackbody; is the radiation bias caused by the thermal radiation of the infrared imaging system itself.

[0092] Step SS24: Preliminary radiation correction of the dual-band infrared image using the response function and the radiation bias.

[0093] In application, step SS24 includes the following formula:

[0094] ;

[0095] wherein, is a waveband; is a raw spectral measurement value of the waveband infrared image; is a radiation bias caused by self-thermal radiation of the infrared imaging system; is a response function of the infrared imaging system; is a preliminary radiation corrected spectral radiance of the waveband infrared image.

[0096] Step SS25: performing secondary radiation correction on the preliminary corrected dual-waveband infrared image by using an inverse function of the Planck formula.

[0097] The application is, by the inverse function of the Planck formula, converting the spectral radiance into a brightness temperature. The brightness temperature image is less affected by the environmental temperature fluctuation, and provides a more stable and reliable image data source for subsequent dynamic threshold segmentation.

[0098] In actual application, step SS25 includes the following formula:

[0099] ;

[0100] wherein, is a brightness temperature of the secondary radiation corrected waveband infrared image; is the first radiation constant of Planck, and the value is generally 1.191 x 10 －12 ; is the second radiation constant of Planck, and the value is generally 1.439 x 10 4 .

[0101] In some other embodiments, the blackbody radiation standard values of 0 μm to 14 μm at 60℃, 120℃ and 200℃ are simulated respectively, and the simulated blackbody radiation standard values are compared with the raw data actually measured by the infrared imaging system, and there is still a large difference. By using the two-point correction method, the radiation calibration correction parameters of each waveband are calculated. In the 10630 nm waveband, due to the reflection of the filter and other reasons, the curve fluctuates too much. The data of this waveband is removed by using the rejection operation, and the missing data is filled by using the bilinear interpolation of the adjacent two waveband data, so that the radiation correction result is obtained, and at this time, the blackbody radiation curve region at each temperature has been basically stable.

[0102] In some embodiments, the image denoising includes: based on a three-dimensional filtering algorithm (Block-Matching 3D, BM3D), performing block matching, three-dimensional transformation, threshold shrinkage and inverse transformation processing on the dual-waveband infrared image. Further, the image denoising can further include block aggregation processing and merging processing on the dual-waveband infrared image.

[0103] In practice, block matching includes: first, divide the infrared image into overlapping small blocks (each small block is usually non-overlapping, with a size of MxN pixels). Then, for each small block, search for blocks with similar content in the entire infrared image to establish a block matching group.

[0104] Three-dimensional transformation includes: superimposing the blocks in each block matching group into a three-dimensional block in a certain superimposed manner (for example, integrating the information of similar blocks together to improve the noise reduction effect). Then, perform a transformation operation on each three-dimensional block, usually using a two-dimensional discrete cosine transform (DCT), to convert the signal from the spatial domain to the frequency domain, which helps to analyze and reduce noise.

[0105] Threshold shrinkage includes: performing a threshold operation on the frequency domain representation of each three-dimensional block to set low-amplitude frequency domain coefficients to zero, which helps to reduce the impact of noise.

[0106] Inverse transformation processing includes: performing an inverse transformation on the three-dimensional block after threshold processing to restore the signal from the frequency domain to the spatial domain.

[0107] Block aggregation processing includes: restoring all denoised three-dimensional blocks to image blocks and superimposing them according to their positions in the image to eliminate artifacts introduced by block matching.

[0108] Merge processing includes: merging the results of all block aggregation into a final denoised image, for example Figure 3 b. Infrared image after denoising processing relative to the unprocessed image Figure 3 a, denoised image Figure 3 The vertical line phenomenon in b is significantly improved, the image background is more uniform, and the texture is clearer.

[0109] Step SS3: registration processing of the preprocessed dual-band infrared image.

[0110] In practice, the registration processing of the present application is not sensitive to changes in illumination (radiation intensity), and even if there is a difference in overall grayscale between the two-band infrared images, it can accurately find the spatial correspondence, ensuring that the subsequent difference operation is performed between pixels at the same spatial position, avoiding false edges and noise introduced by image misalignment.

[0111] In practice, step SS3 includes steps SS31 to SS35.

[0112] Step SS31: select an area in the preprocessed reference band infrared image (for example Figure 4 left view) as a matching template (for example Figure 4 template).

[0113] In application, a region with significant features is selected as the matching template in the preprocessed reference band infrared image. The region with significant features refers to a region with local contrast higher than a preset threshold in the infrared image. In actual application, a typical local contrast range that can produce stable and accurate registration results can be summarized by analyzing existing image data, so as to determine an empirical fixed threshold as the preset threshold.

[0114] Step SS32: sliding the matching template on the preprocessed detection band infrared image (for example, the right view of FIG. 2) and calculating the normalized cross-correlation coefficient between the matching template and the current window region of the detection band infrared image at each sliding position. Figure 4

[0115] In application, the normalized cross-correlation coefficient includes the following formula:

[0116]

[0117] In the formula, Cxy(x,y) is the normalized cross-correlation coefficient at the position (x,y); is the matching template; is the preprocessed detection band infrared image with the same size as the matching template and with (x,y) as the upper left corner point; is the covariance of and is the standard deviation of the pixel value of is the standard deviation of the pixel value of is the standard deviation of the pixel value of the matching template.

[0118] Step SS34: referring to the formula (1), the corresponding window region is determined as the matching position of the matching template in the detection band infrared image in response to the maximum normalized cross-correlation coefficient. Figure 4

[0119] In application, the sliding position that makes the normalized cross-correlation coefficient maximum is determined as the best matching position of the matching template in the preprocessed detection band infrared image.

[0120] Step SS35: calculating the spatial transformation parameter between the detection band infrared image and the reference band infrared image according to the matching template and the matching position, and performing geometric transformation on the detection band infrared image according to the spatial transformation parameter, so that the detection band infrared image is spatially aligned with the reference band infrared image.

[0121] ​​​​​​​In application, the geometric transformation is a translation transformation; and the spatial transformation parameter is a translation vector (Δx, Δy). Wherein, Δx is a translation vector in X-axis direction, and Δy is a translation vector in Y-axis direction.

[0122] Step SS4: performing difference operation on the dual-band infrared image after registration processing to obtain a difference image.

[0123] In application, pixel-level difference operation is performed on the registered reference-band infrared image and the registered detection-band infrared image. Through pixel-level difference operation, the present application amplifies the tiny difference between the detection band and the reference band due to SF6 gas absorption, so that the invisible gas becomes a visible signal with higher background contrast in the difference image.

[0124] Step SS5: determining a dynamic segmentation threshold according to the current ambient temperature, and performing threshold segmentation on the difference image according to the dynamic segmentation threshold to identify an SF6 gas leakage area.

[0125] In application, the threshold segmentation on the difference image according to the dynamic segmentation threshold comprises the following formula:

[0126] ;

[0127] In the formula, is a binary result image obtained by threshold segmentation on the difference image, wherein 1 represents a gas leakage area (reference Figure 5 , and the red part is an SF6 gas area to be detected), and 0 represents a background; is the registered reference-band infrared image; is the registered detection-band infrared image; is the dynamic segmentation threshold.

[0128] In actual application, the dynamic segmentation threshold comprises the following formula:

[0129] ;

[0130] In the formula, is the current ambient temperature, which can be 23-60℃, preferably 30-60℃ in the present application; is a temperature base parameter, which is determined according to the highest temperature of the ambient temperature, and ; is a sensitivity coefficient, which is determined according to the lowest temperature of the ambient temperature and the temperature base parameter, and ; is a temperature decay rate, which is determined according to the lowest temperature, the highest temperature of the ambient temperature and the temperature base parameter, and .

[0131] In some other embodiments, the method for determining the sensitivity coefficient, temperature decay rate and temperature base parameter of the present application comprises steps SS511 to SS517.

[0132] Step SS511: A series of temperature points (e.g. 30℃, 35℃, 40℃, 45℃, 50℃, 55℃, 60℃) are set at fixed intervals (e.g. 5℃ or 10℃) within the target temperature range (e.g. 23℃ to 60℃, 30℃ to 60℃), and SF6 gas leakage detection experiments are performed at multiple temperature points.

[0133] Step SS512: At each temperature point, an infrared imaging system is used to collect SF6 gas leakage videos for a period of time, and artificial labeled gas position labels are recorded synchronously.

[0134] Step SS513: The data collected in step SS512 is preprocessed and registered in sequence using the method described in steps SS2 to SS4, to obtain difference images.

[0135] Step SS514: The difference images are segmented by traversing candidate thresholds (e.g. from 0 to the maximum gray value of the difference images, with a step of 1).

[0136] Step SS515: For each candidate threshold, the segmentation result is compared with the gas true value recorded in the gas position label in step SS52, and a performance indicator (generally calculated using the F1-Score method) is calculated.

[0137] Step SS516: A threshold performance relationship curve is drawn according to the candidate threshold and its corresponding performance indicator, and the optimal segmentation threshold for each temperature point is determined according to the threshold performance relationship curve. When applied, the optimal segmentation threshold corresponds to the maximum F1-Score in the threshold performance relationship curve.

[0138] Step SS517: Using a nonlinear least squares method, multiple temperature points and their corresponding optimal segmentation thresholds are curve fitted with the formula of the dynamic segmentation threshold, to finally determine the specific values of the sensitivity coefficient, temperature decay rate and temperature base parameter that minimize the curve fitting error.

[0139] In some other embodiments, the temperature base parameter of the present application comprises: wherein, is the temperature base parameter, is the optimal segmentation threshold measured by experiment at the highest ambient temperature. Further, the sensitivity coefficient of the present application comprises: and wherein, is the sensitivity coefficient, is the optimal segmentation threshold measured by experiment at the lowest ambient temperature. Further, the temperature decay rate of the present application includes: wherein, is the temperature decay rate. It should be noted that the ambient temperature range of the present application should cover at least the ambient temperature range of the actual application.

[0140] It is worth noting that the temperature decay rate determines how fast the dynamic segmentation threshold decreases as the current ambient temperature rises: the greater the temperature decay rate, the faster the dynamic segmentation threshold decreases, and the more sensitive it is to weak gas signals at high temperatures; the smaller the temperature decay rate, the slower the dynamic segmentation threshold decreases, and the more gentle the sensitivity changes at different temperatures. If the temperature decay rate is too large, it means that the dynamic segmentation threshold changes too drastically with temperature. At temperatures close to the highest ambient temperature, the dynamic segmentation threshold may become too low, causing background noise to be misjudged as gas leakage. In order to prevent excessive sensitivity at high temperatures and lose stability, the present application uses the total change range of the segmentation threshold within the entire temperature range and the normalization factor to constrain the upper limit of the temperature decay rate, ensuring that the balance between sensitivity and robustness can be achieved within a wide temperature range of 23°C to 60°C.

[0141] In a specific embodiment, Figure 6 is the detection result of SF6 gas under a 30°C ( Figure 6 a), 40°C ( Figure 6 b), 50°C ( Figure 6 c), and 60°C ( Figure 6 d) standard blackbody background. Experiments show that as the background temperature rises, the gas detection effect gradually improves, which is because the higher the background temperature, the more significant the image contrast formed after the gas absorbs infrared radiation, thereby improving the detection performance. Although the detection effect is better at high temperatures, the optimization method of the present application still has good robustness at a temperature of 30°C, verifying that the optimization method of the present application can be used for SF6 gas monitoring in room temperature environments. Further, SF6 gas is released under a 40°C blackbody background, and the optimization method of the present application is used for real-time monitoring, and the imaging results of the left and right fields of view are shown in Figure 7 a (reference band infrared image) and Figure 7 b (detection band infrared image). To quantitatively evaluate the detection performance, one frame of image is manually labeled with a gas region to generate a gas position label (such as Figure 7 c). By comparing the gas detection result (such as Figure 7 d) with the gas position label, a confusion matrix is obtained (TN is 326167, FN is 362, FP is 972, and TP is 179), and quantitative evaluation indicators are calculated accordingly: accuracy is 0.9959, precision is 0.1555, recall is 0.3309, and sensitivity is 0.9970.

[0142] In another aspect, referring to Figure 8 The application discloses a system for optimizing SF6 gas detection in a multi-temperature environment, which comprises a collection module, a preprocessing module, a registration module, a difference module and an identification module.

[0143] The collection module is configured to simultaneously acquire a dual-band infrared image of a scene to be detected, wherein the dual-band infrared image comprises a detection band infrared image and a reference band infrared image; the preprocessing module is configured to preprocess the dual-band infrared image; the registration module is configured to perform registration processing on the preprocessed dual-band infrared image; the difference module is configured to perform difference operation on the registration-processed dual-band infrared image to obtain a difference image; and the identification module is configured to determine a dynamic segmentation threshold according to a current environmental temperature, perform threshold segmentation on the difference image according to the dynamic segmentation threshold, and identify an SF6 gas leakage area.

[0144] The specific functions of the above functional modules are implemented by referring to the specific steps of the method for optimizing SF6 gas detection in a multi-temperature environment.

[0145] Those skilled in the art will understand that the embodiments of the present application can be provided as a method, a system or a computer program product. Therefore, the present application can be in the form of an entirely hardware embodiment, an entirely software embodiment or an embodiment combining software and hardware aspects. Moreover, the present application can be in the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0146] The present application is described with reference to flowcharts and / or block diagrams of the methods, devices (systems) and computer program products according to the embodiments of the present application. It should be understood that each flow and / or block in the flowcharts and / or block diagrams, and the combination of the flows and / or blocks in the flowcharts and / or block diagrams can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor or other programmable data processing device to produce a machine, so that the instructions executed by the processor of the computer or other programmable data processing device produce a device that implements the functions specified in the flowcharts and / or block diagrams. Figure 1 The functions specified in one flow or multiple flows and / or blocks Figure 1 The functions specified in one flow or multiple flows and / or blocks

[0147] These computer program instructions can also be stored in a computer-readable memory capable of guiding a computer or other programmable data processing device to work in a specific manner, so that the instructions stored in the computer-readable memory produce a product comprising instruction devices, which implement the functions specified in the flowcharts and / or block diagrams. Figure 1one or more processes and / or blocks Figure 1 the function specified in the one or more blocks.

[0148] These computer program instructions can also be loaded into computer or other programmable data processing devices, so that a series of operation steps are performed on the computer or other programmable data processing devices to generate computer-implemented processes, so that the instructions executed on the computer or other programmable data processing devices provide processes for implementing the flow Figure 1 one or more processes and / or blocks Figure 1 Figure 1 the function specified in the one or more blocks.

[0149] Obviously, the above embodiments are only examples for clearly illustrating, not limiting the embodiments. For those skilled in the art, on the basis of the above description, other different forms of changes or variations can also be made. Here, it is not necessary and also impossible to enumerate all the embodiments. The obvious changes or variations derived therefrom are still within the protection scope of the present application.

Claims

1. An optimized method for SF6 gas detection under multiple temperature conditions, characterized in that... This includes the following steps: Simultaneously acquire dual-band infrared images of the scene under test, the dual-band infrared images including a detection band infrared image and a reference band infrared image; Preprocess the dual-band infrared image; Dual-band infrared image after registration and preprocessing; Differential operations are performed on the registered dual-band infrared image to obtain a differential image; Based on the current ambient temperature, a dynamic segmentation threshold is determined, and the differential image is segmented according to the dynamic segmentation threshold to identify SF6 gas leakage areas. The step of thresholding the difference image based on the dynamic segmentation threshold includes: ； In the formula, This is the binary image obtained after thresholding the difference image, where 1 represents the gas leak area and 0 represents the background. Infrared image of the reference band after registration; Infrared images of the registered detection band; The dynamic segmentation threshold; The step of determining the dynamic segmentation threshold based on the current ambient temperature includes: ； In the formula, The current ambient temperature; This is a temperature-based parameter, determined based on the highest ambient temperature, and ; The sensitivity coefficient is determined based on the lowest ambient temperature and the temperature baseline parameter. ; The temperature decay rate is determined based on the lowest and highest ambient temperatures, as well as the base temperature parameter. .

2. The optimized method for SF6 gas detection under multiple temperature environments according to claim 1, characterized in that, The preprocessing includes at least one of dark current correction, infrared radiometric correction, and image noise reduction. The dark current correction includes subtracting the dark current image from the dual-band infrared image to eliminate the noise of the infrared imaging system itself. The infrared radiation correction includes: correcting the dual-band infrared image based on the two-point linear calibration method to eliminate the thermal radiation of the infrared imaging system itself; The image denoising includes: performing block matching, three-dimensional transformation, threshold shrinkage, and inverse transformation processing on dual-band infrared images based on a three-dimensional filtering algorithm.

3. The optimized method for SF6 gas detection under multiple temperature environments according to claim 2, characterized in that, The infrared radiation correction includes; Preliminary radiometric correction of dual-band infrared images is performed using the response function and radiometric bias of the infrared imaging system. Secondary radiometric correction was performed on the pre-corrected dual-band infrared image using the inverse function of Planck's formula. The secondary radiation correction includes the following formula: ； In the formula, After secondary radiation correction Brightness temperature of infrared images in the spectral band; It is Planck's first radiation constant; is Planck's second radiation constant.

4. The optimized method for SF6 gas detection under multiple temperature environments according to claim 3, characterized in that, The preliminary radiation correction includes the following formula: ； In the formula, For bands; for Raw spectral measurements of the band infrared image; This refers to the radiation bias caused by the thermal radiation of the infrared imaging system itself. Let this be the response function of the infrared imaging system; After preliminary radiation correction Spectral radiance of infrared images in the spectral band; Prior to the initial radiation correction step, the infrared radiation correction also includes: Two-point calibration of the infrared imaging system was performed using a high-temperature blackbody and a low-temperature blackbody to obtain the original spectral measurements of the high-temperature blackbody and the low-temperature blackbody. Based on Planck's blackbody radiation law, the theoretical spectral radiance of the high-temperature blackbody and the low-temperature blackbody are calculated respectively. Based on the original spectral measurements and the theoretical spectral radiance, the response function and radiative bias of the infrared imaging system are calculated. Wherein, the temperature of the high-temperature blackbody is not lower than the highest temperature in the range of the measured environment, and the temperature of the low-temperature blackbody is not higher than the lowest temperature in the range of the measured environment.

5. The optimized method for SF6 gas detection under multiple temperature environments according to claim 1, characterized in that, The registration process includes: A region in the preprocessed reference band infrared image is selected as the matching template. The matching template is slid across the preprocessed infrared image of the detection band, and at each sliding position, the normalized cross-correlation coefficient between the matching template and the current window region of the infrared image of the detection band is calculated. In response to the maximum normalized cross-correlation coefficient, the corresponding window region is determined as the matching position of the matching template in the infrared image of the detection band. Based on the matching template and the matching position, the spatial transformation parameters between the detection band infrared image and the reference band infrared image are calculated, and the detection band infrared image is geometrically transformed according to the spatial transformation parameters to make the detection band infrared image and the reference band infrared image spatially aligned.

6. The optimized method for SF6 gas detection under multiple temperature environments according to claim 5, characterized in that, The normalized cross-correlation coefficient includes the following formula: ； In the formula, for Normalized cross-correlation coefficient at position; To match the template; In the preprocessed infrared image of the detection band, a sub-image block with (x,y) as the upper left corner and the same size as the matching template is selected. for and covariance; for The standard deviation of pixel values; The standard deviation of pixel values ​​for matching the template.

7. The optimized method for SF6 gas detection under multiple temperature environments according to claim 1, characterized in that, The difference operation includes: The registered reference band infrared image and the registered detector band infrared image are subjected to pixel-level difference calculation.

8. The optimized method for SF6 gas detection under multiple temperature environments according to claim 1, characterized in that, The current ambient temperature is between 23°C and 60°C. And / or, the center wavelength of the detection band is from 10500nm to 10630nm, and the center wavelength of the reference band is from 8050nm to 8360nm.

9. An optimized system for SF6 gas detection under multiple temperature environments, characterized in that, include: The acquisition module is used to simultaneously acquire dual-band infrared images of the scene under test, the dual-band infrared images including a detection band infrared image and a reference band infrared image; A preprocessing module is used to preprocess the dual-band infrared image; The registration module is used to register the pre-processed dual-band infrared image; The differential module is used to perform differential operations on the registered dual-band infrared image to obtain a differential image. The identification module is used to determine a dynamic segmentation threshold based on the current ambient temperature, and to perform threshold segmentation on the differential image based on the dynamic segmentation threshold in order to identify the SF6 gas leakage area. The step of thresholding the difference image based on the dynamic segmentation threshold includes: ； In the formula, This is the binary image obtained after thresholding the difference image, where 1 represents the gas leak area and 0 represents the background. Infrared image of the reference band after registration; Infrared images of the registered detection band; The dynamic segmentation threshold; The step of determining the dynamic segmentation threshold based on the current ambient temperature includes: ； In the formula, The current ambient temperature; This is a temperature-based parameter, determined based on the highest ambient temperature, and ; The sensitivity coefficient is determined based on the lowest ambient temperature and the temperature baseline parameter. ; The temperature decay rate is determined based on the lowest and highest ambient temperatures, as well as the base temperature parameter. .

Images