A night multispectral low-light remote sensing image brightness anomaly detection method and device

Abstract

The application provides a night multispectral faint light remote sensing image brightness anomaly detection method and device, comprising: acquiring a to-be-detected image, the to-be-detected image being a night multispectral faint light remote sensing image; performing radiation calibration on the to-be-detected image to obtain a radiation brightness value corresponding to each pixel point in the to-be-detected image; sequentially judging whether the radiation brightness value corresponding to each pixel point is greater than a preset brightness anomaly threshold value; in the case that the radiation brightness value is greater than the preset brightness anomaly threshold value, taking the pixel point corresponding to the radiation brightness value greater than the brightness anomaly threshold value as a brightness anomaly initial point; and generating an anomaly region based on the brightness anomaly initial point by using a predefined region generation algorithm. The detection process of the application is fast and simple, does not need to rely on multiple different types of data, and has high detection reliability.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of image recognition, and in particular to a night multispectral low-light remote sensing image brightness anomaly detection method and device. BACKGROUND

[0002] Disaster monitoring and early warning is an important application field of remote sensing satellite data. Due to global climate change, geopolitical conflicts and other causes, various natural and man-made disasters have increased, and the risks caused by them have expanded. Timely monitoring and accurate early warning of surface anomalies have become a major demand to ensure the stable and healthy development of society and economy. Due to the diurnal rhythm of human activities and the lack of visible light, night anomalies are more hidden and sudden than daytime, and are more likely to cause untimely discovery and response of anomalies, thereby causing greater harm.

[0003] China has a vast territory, and low-light remote sensing images taken at night provide a direct data and information source for large-scale night surface anomaly detection. Currently available domestic and foreign main low-light remote sensing data sources include: (1) VIIRS sensor data on the Suomi-NPP satellite of the United States, containing one panchromatic band, with a spatial resolution greater than 500 meters and a temporal resolution of 12 hours; (2) low-light sensor data on the Luojia-1 01 satellite of China, containing one panchromatic band, with a spatial resolution of about 130 meters and a temporal resolution of about 15 days; (3) low-light sensor data on the Jilin-1 03B satellite of China, containing red, green and blue bands, with a spatial resolution of 0.92 meters and a temporal resolution of about 4.5 days; (4) low-light sensor data on the Sustainable Development Satellite 1 of China, containing panchromatic and red, green and blue bands, with a panchromatic band spatial resolution of 10 meters, red, green and blue band spatial resolutions of 40 meters, and a temporal resolution of about 11 days.

[0004] Early low-light sensors generally only contain one panchromatic band, i.e. only the brightness intensity can be detected, and lack color information such as red, green and blue. However, typical night disasters such as fires, armed conflicts and high-intensity light pollution are difficult to distinguish from normal conditions based on brightness intensity alone, especially in urban areas with strong night lights. For example, the normal night light intensity in many central business districts of large cities far exceeds the light intensity caused by fires and armed conflicts. Therefore, it is difficult to effectively monitor night disasters based on single panchromatic low-light images, which can easily cause a high false scene rate. Most related research on anomaly detection using panchromatic low-light images also introduces other data to distinguish between normal and abnormal conditions.

[0005] With the rapid development of remote sensing satellite platform and sensor technology in recent years, the newly launched micro-light sensor has gradually possessed the observation ability of multi-spectral bands such as red, green and blue. The multi-spectral micro-light remote sensing image itself has enough information to accurately distinguish various night-time abnormalities and normal conditions. For example, under the same full-color band brightness, the red light brightness produced by the flame is much higher than the green light brightness, which can reach more than ten times the proportion, and the blue light brightness is even weaker. It can be considered that the brightness of the full-color band is mainly caused by the red light produced by the flame, and for normal night lighting in the city, the brightness proportion of red light, green light and blue light is more balanced, and the proportion of red light is slightly higher.

[0006] With the increase in the number of bands, the spatial resolution of night micro-light images has also been significantly improved, from hundreds of meters to tens of meters or even meters. At the spatial resolution of hundreds of meters, the range of abnormal fire points is usually one or several pixels, and at the spatial resolution of tens of meters or even meters, the detailed range and regional boundary detection of abnormal fire points become a problem that needs to be concerned.

[0007] In summary, a method that can utilize the characteristics of different spectral bands and the relationship between spectral bands in multi-spectral micro-light images to accurately detect night disaster abnormalities and their detailed ranges is urgently needed in the current disaster monitoring field. SUMMARY

[0008] The present application provides a night multi-spectral micro-light remote sensing image brightness anomaly detection method and device to solve the above problems.

[0009] The present application provides a night multi-spectral micro-light remote sensing image brightness anomaly detection method, comprising:

[0010] Obtaining a to-be-detected image, which is a night multi-spectral micro-light remote sensing image;

[0011] Radiometric calibration is performed on the to-be-detected image to obtain a radiation brightness value corresponding to each pixel point in the to-be-detected image;

[0012] The radiation brightness value corresponding to each pixel point is sequentially judged to determine whether it is greater than a preset brightness anomaly threshold value;

[0013] In the case where the radiation brightness value is greater than the preset brightness anomaly threshold value, the pixel point corresponding to the radiation brightness value greater than the brightness anomaly threshold value is taken as a brightness anomaly initial point;

[0014] Based on the brightness anomaly initial point, a predefined region generation algorithm is used to generate an abnormal region.

[0015] According to the night multispectral faint light remote sensing image brightness anomaly detection method provided by the application, the type corresponding to the abnormal area is flame anomaly, and the radiation brightness value includes a red waveband radiation brightness value and a green waveband radiation brightness value.

[0016] Correspondingly, in the case that the radiation brightness value is greater than the preset brightness anomaly threshold value, the pixel point corresponding to the radiation brightness value greater than the brightness anomaly threshold value is taken as a brightness anomaly initial point, and the method comprises the following steps of:

[0017] In the case that the red waveband radiation brightness value of the pixel point is greater than a preset first flame anomaly threshold value and the ratio between the red waveband radiation brightness value and the green waveband radiation brightness value is greater than a preset second flame anomaly threshold value, the corresponding pixel point is taken as a flame anomaly initial point.

[0018] According to the night multispectral faint light remote sensing image brightness anomaly detection method provided by the application, the type corresponding to the abnormal area is strong light pollution, and the radiation brightness value includes a red waveband radiation brightness value, a green waveband radiation brightness value and a blue waveband radiation brightness value.

[0019] Correspondingly, in the case that the radiation brightness value is greater than the preset brightness anomaly threshold value, the pixel point corresponding to the radiation brightness value greater than the brightness anomaly threshold value is taken as a brightness anomaly initial point, and the method comprises the following steps of:

[0020] In the case that the red waveband radiation brightness value of the pixel point is greater than a preset first strong light pollution threshold value, the green waveband radiation brightness value is greater than a preset second strong light pollution threshold value and the blue waveband radiation brightness value is greater than a preset third strong light pollution threshold value, the corresponding pixel point is taken as a strong light pollution anomaly initial point.

[0021] According to the night multispectral faint light remote sensing image brightness anomaly detection method provided by the application, the predefined region generation algorithm is an eight-neighbor region growing algorithm with an additional access condition.

[0022] Correspondingly, the method comprises the following steps of:

[0023] S1, an abnormal region set and an abnormal pixel point set are constructed, the abnormal region set is used for storing abnormal regions, and the abnormal pixel point set is used for storing all non-repeated pixel points in the abnormal regions;

[0024] S2, in the case that the brightness anomaly initial point and the pixel points in the abnormal pixel point set are all different, the brightness anomaly initial point is taken as a brightness anomaly pixel point, and eight pixel points adjacent to the brightness anomaly initial point are determined as to-be-inspected pixel points;

[0025] S3, in the case that the pixel point to be investigated meets the condition of joining the abnormal region, taking the pixel point to be investigated as a luminance abnormal pixel point;

[0026] S4, taking the pixel point not yet investigated among the eight pixel points adjacent to the luminance abnormal pixel point as a new pixel point to be investigated, judging whether the new pixel point to be investigated meets the condition of joining the abnormal region;

[0027] S5, repeatedly executing steps S3-S4 until there is no new pixel point to be investigated meeting the condition of joining the abnormal region, so as to obtain the abnormal region and store it into the abnormal region set, the pixel points corresponding to the abnormal region are all luminance abnormal pixel points, and all the luminance abnormal pixel points are stored into the abnormal pixel point set;

[0028] S6, repeatedly executing steps S2-S5 until all the luminance abnormal initial points are traversed, so as to obtain all the abnormal regions.

[0029] According to the night multispectral micro-light remote sensing image luminance abnormality detection method provided by the application, in the case that the type corresponding to the abnormal region is flame abnormality, the case that the pixel point to be investigated meets the condition of joining the abnormal region includes:

[0030] According to the red waveband radiation luminance value and the green waveband radiation luminance value of the pixel point to be investigated, judging whether the pixel point to be investigated meets the condition of joining the flame abnormal region;

[0031] The condition of the flame abnormal region is that the red waveband radiation luminance value of the pixel point to be investigated is greater than a preset third flame abnormal threshold value, or the red waveband radiation luminance value of the pixel point to be investigated is greater than a preset fourth flame abnormal threshold value and the ratio between the red waveband radiation luminance value and the green waveband radiation luminance value is greater than a preset fifth flame abnormal threshold value.

[0032] In the case that the pixel point to be investigated meets the condition of joining the flame abnormal region, taking the pixel point to be investigated as a flame abnormal pixel point.

[0033] According to the night multispectral micro-light remote sensing image luminance abnormality detection method provided by the application, in the case that the type corresponding to the abnormal region is strong light pollution, the case that the pixel point to be investigated meets the condition of joining the abnormal region includes:

[0034] According to the red waveband radiation luminance value, the green waveband radiation luminance value and the blue waveband radiation luminance value of the pixel point to be investigated, judging whether the pixel point to be investigated meets the condition of joining the strong light pollution abnormal region;

[0035] The condition of the strong light pollution abnormal area is that the red waveband radiation brightness value of the pixel point to be investigated is greater than a fourth preset strong light pollution threshold value, the green waveband radiation brightness value is greater than a fifth preset strong light pollution threshold value, and the blue waveband radiation brightness value is greater than a sixth preset strong light pollution threshold value.

[0036] In a case where the pixel point to be investigated meets the condition of joining the strong light pollution abnormal area, the pixel point to be investigated is taken as a strong light pollution abnormal pixel point.

[0037] According to the night multispectral micro-light remote sensing image brightness abnormality detection method provided by the application, the radiation calibration is performed on the image to be detected, and the radiation brightness value corresponding to each pixel point in the image to be detected is obtained.

[0038] The radiation brightness value is obtained according to the pixel digital value of the image to be detected, the gain coefficient and the bias coefficient corresponding to the micro-light sensor.

[0039] According to the night multispectral micro-light remote sensing image brightness abnormality detection method provided by the application, after the abnormal area is generated based on the brightness abnormality initial point by using the pre-defined area generation algorithm, the method further comprises:

[0040] The abnormal area is investigated according to the geographical environment corresponding to the abnormal area, and the abnormal area after investigation is obtained.

[0041] The application further provides a night multispectral micro-light remote sensing image brightness abnormality detection device, comprising:

[0042] An image acquisition module is configured to acquire an image to be detected, which is a night multispectral micro-light remote sensing image.

[0043] A radiation calibration module is configured to perform radiation calibration on the image to be detected, and obtain a radiation brightness value corresponding to each pixel point in the image to be detected.

[0044] An abnormality judgment module is configured to judge whether the radiation brightness value corresponding to each pixel point is greater than a preset brightness abnormality threshold value.

[0045] A brightness abnormality initial point acquisition module is configured to, in a case where the radiation brightness value is greater than the preset brightness abnormality threshold value, take the pixel point corresponding to the radiation brightness value greater than the brightness abnormality threshold value as a brightness abnormality initial point.

[0046] An abnormal area generation module is configured to generate an abnormal area based on the brightness abnormality initial point by using a pre-defined area generation algorithm.

[0047] The application further provides an electronic device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, and the processor implements the night multispectral low-light remote sensing image brightness anomaly detection method as described above when executing the program.

[0048] The night multispectral low-light remote sensing image brightness anomaly detection method and device provided by the application determines the brightness anomaly initial point through a preset brightness anomaly threshold, utilizes the relationship between the typical characteristics of the typical night anomaly in different spectral bands and the spectral bands, thereby accurately distinguishes the abnormal pixel points, and detects the abnormal region range through the region growing algorithm. BRIEF DESCRIPTION OF DRAWINGS

[0049] In order to more clearly illustrate the technical solutions in the application or the prior art, the following will briefly introduce the drawings needed to be used in the embodiments or the prior art description. Obviously, the drawings in the following description are some embodiments of the application, and for those skilled in the art, other drawings can also be obtained without creative labor on the basis of these drawings.

[0050] Figure 1 Fig. 1 is one of the flowcharts of the night multispectral low-light remote sensing image brightness anomaly detection method provided by the embodiments of the application;

[0051] Figure 2 Fig. 2 is another of the flowcharts of the night multispectral low-light remote sensing image brightness anomaly detection method provided by the embodiments of the application;

[0052] Figure 3 Fig. 3 is a comparison schematic diagram before and after the flame anomaly detection provided by the embodiments of the application;

[0053] Figure 4 Fig. 4 is a flowchart of the eight-neighbor region growing algorithm provided by the embodiments of the application;

[0054] Figure 5 Fig. 5 is a flame anomaly region deduplication effect comparison diagram provided by the embodiments of the application;

[0055] Figure 6 Fig. 6 is a structural schematic diagram of the night multispectral low-light remote sensing image brightness anomaly detection device provided by the embodiments of the application;

[0056] Figure 7 Fig. 7 is a physical structure schematic diagram of an electronic device provided by the embodiments of the application. DETAILED DESCRIPTION

[0057] In order to make the objects, technical solutions and advantages of the present application clearer, the technical solutions in the present application will be clearly and completely described below in combination with the drawings in the present application. Obviously, the described embodiments are only a part 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 of ordinary skill in the art without creative work belong to the protection scope of the present application.

[0058] The present application considers the timeliness requirement of disaster monitoring, and supports the application mode of instant warning on satellite. The abnormality detection method should have the characteristics of rapidity, simplicity, less dependence on data types, high detection reliability, and the like, and can guarantee the detection effect under the condition of limited computing and communication resources of satellite. Alternatively, the abnormality detection method serves as a kind of abnormality pre-screening means, and provides guidance for subsequent fine abnormality diagnosis.

[0059] Figure 1 is one of flowcharts of the night multispectral low-light remote sensing image brightness abnormality detection method provided by the embodiments of the present application; Figure 2 is another flowchart of the night multispectral low-light remote sensing image brightness abnormality detection method provided by the embodiments of the present application; as shown in Figure 1 and 2, the night multispectral low-light remote sensing image brightness abnormality detection method comprises the following steps.

[0060] S101, acquiring a to-be-detected image, the to-be-detected image being a night multispectral low-light remote sensing image, i.e., a multi-band low-light remote sensing image imaged at night.

[0061] S102, performing radiation calibration on the to-be-detected image to obtain a radiation brightness value corresponding to each pixel point in the to-be-detected image.

[0062] In this step, the pixel value of the remote sensing image is converted into a radiation brightness value with physical meaning. The conversion process is radiation calibration. The unit of the calibrated radiation brightness value is W / m 2 / sr / μm, wherein W represents watt, m 2 represents square meter, sr represents steradian, and μm represents micrometer.

[0063] The specific radiation calibration method and calibration parameters can be determined according to the satellite sensor. The satellite operating mechanism has corresponding calibration parameters of the sensor. In addition, assuming that the brightness abnormality detection method is directly operated on the satellite, the satellite calibration capability is required. Alternatively, the threshold parameters involved in the present application can be converted from the radiation brightness physical value into the original data value (i.e., the pixel value) of the corresponding satellite for corresponding comparison.

[0064] It should be noted that the radiation brightness value here refers to the multi-band radiation brightness value of each pixel point, including the red-band radiation brightness value, the green-band radiation brightness value, the blue-band radiation brightness value, and the like.

[0065] S103, judging whether the radiation brightness value corresponding to each pixel point is greater than a preset brightness abnormal threshold value in sequence. That is, the brightness abnormal threshold value is set according to the typical characteristics corresponding to the typical night-time abnormalities such as flame, high-intensity light pollution, and the brightness abnormal threshold value is used to judge whether the pixel point in the image is the pixel point of the abnormal region.

[0066] S104, in the case where the radiation brightness value is greater than the preset brightness abnormal threshold value, the pixel point corresponding to the radiation brightness value greater than the brightness abnormal threshold value is taken as the brightness abnormal initial point. That is, a batch of pixel points capable of indicating the potential position of the abnormality occurrence are screened out as the brightness abnormal initial point through the brightness abnormal threshold value.

[0067] If the radiation brightness value is less than or equal to the preset brightness abnormal threshold value, it is determined that the pixel point corresponding to the radiation brightness value is the pixel point of the normal region.

[0068] S105, generating the abnormal region based on the brightness abnormal initial point by using a predefined region generation algorithm.

[0069] In this step, since the screened brightness abnormal initial point is the pixel point with the most significant characteristics of the typical night-time abnormality, the pixel points around it can be the abnormal pixel points with insignificant characteristics or the normal pixel points. Therefore, it is necessary to generate the abnormal region around the determined brightness abnormal initial point, and the specific generation process is realized by using the predefined region generation algorithm, so as to obtain the abnormal region and complete the abnormality detection.

[0070] Taking the flame abnormality as an example, the comparison of the abnormality detection results obtained by the micro-light image and the night-time multispectral micro-light remote sensing image brightness abnormality detection method provided by the present application is as shown in the figure. Figure 3 The abnormality detection result accurately distinguishes the city light from the flame abnormality, and the flame abnormality region is captured completely.

[0071] In addition, after obtaining the abnormal region, the abnormal region can be further investigated and confirmed, for example, by manual confirmation or by investigating the geographical position to which the abnormal region belongs.

[0072] The night-time multispectral micro-light remote sensing image brightness abnormality detection method provided by the embodiment of the present application determines the brightness abnormal initial point by using the preset brightness abnormal threshold value, utilizes the relationship between the typical characteristics of the typical night-time abnormality and the spectral bands, so as to accurately judge the abnormal pixel point, and detects the range of the abnormal region by using the region growing algorithm. The detection process of the night-time multispectral micro-light remote sensing image brightness abnormality detection method provided by the present application is fast and simple, does not need to rely on multiple different types of data, and has high detection reliability.

[0073] In some embodiments of the present application, the type corresponding to the abnormal area is flame abnormality, and the radiation brightness value includes a red waveband radiation brightness value and a green waveband radiation brightness value.

[0074] Correspondingly, the case where the radiation brightness value is greater than the preset brightness abnormality threshold value includes:

[0075] In the case where the red waveband radiation brightness value of the pixel point is greater than the preset first flame abnormality threshold value and the ratio between the red waveband radiation brightness value and the green waveband radiation brightness value is greater than the preset second flame abnormality threshold value, the corresponding pixel point is taken as a flame abnormality initial point.

[0076] In this embodiment, the determination of the flame abnormality initial point is specifically described by taking the flame abnormality such as combustion and explosion as an example.

[0077] Each pixel point in the image to be detected is traversed, and if the red waveband radiation brightness value R of a certain pixel point is greater than 0.0068 W / m 2 / sr / μm (i.e., the first flame abnormality threshold value) and the ratio between the red waveband radiation brightness value R and the green waveband radiation brightness value G is greater than 16 times (i.e., the second flame abnormality threshold value), i.e., R>16×G, the pixel point is considered to be a flame abnormality initial point. After all the pixel points are traversed, all the flame abnormality initial points are obtained.

[0078] It should be noted that the first flame abnormality threshold value 0.0068 W / m 2 / sr / μm and the second flame abnormality threshold value 16 can be adjusted according to more accurate analysis and calculation, and the present application does not make adjustment.

[0079] The night multispectral low-light remote sensing image brightness abnormality detection method provided by the embodiment of the present application determines the flame abnormality initial point according to the brightness abnormality threshold value by analyzing the flame abnormality, a typical night brightness abnormality, to determine the brightness abnormality threshold value corresponding to the typical night brightness abnormality, and the process is simple and efficient.

[0080] In some embodiments of the present application, the type corresponding to the abnormal area is strong light pollution, and the radiation brightness value includes a red waveband radiation brightness value R, a green waveband radiation brightness value G and a blue waveband radiation brightness value B.

[0081] Correspondingly, the case where the radiation brightness value is greater than the preset brightness abnormality threshold value includes:

[0082] In a case that the red waveband radiation luminance value of the pixel point is greater than a preset first strong light pollution threshold, the green waveband radiation luminance value is greater than a preset second strong light pollution threshold, and the blue waveband radiation luminance value is greater than a preset third strong light pollution threshold, the corresponding pixel point is regarded as a strong light pollution abnormal initial point.

[0083] In the embodiment, the determination process of the strong light pollution abnormal initial point is described by taking the typical abnormality of high-intensity light pollution as an example.

[0084] Specifically, each pixel point in the to-be-detected image is traversed, and if the red waveband radiation luminance value R of a certain pixel point is greater than 0.054 W / m 2 / sr / μm (i.e., the first strong light pollution threshold), the green waveband radiation luminance value G is greater than 0.020 W / m 2 / sr / μm (i.e., the third strong light pollution threshold), the pixel point is regarded as a strong light pollution abnormal initial point. After all the pixels are traversed, all the strong light pollution abnormal initial points are obtained.

[0085] It should be noted that the first strong light pollution threshold 0.054 W / m 2 / sr / μm, the second strong light pollution threshold 0.020 W / m 2 / sr / μm, and the third strong light pollution threshold 0.020 W / m2 / sr / μm can be adjusted according to more accurate analysis and calculation, and the present application does not make adjustment.

[0086] The night multispectral low-light remote sensing image luminance anomaly detection method provided by the embodiment of the present application determines the luminance anomaly threshold corresponding to the typical night luminance anomaly of strong light pollution anomaly by analyzing the typical night luminance anomaly, and then determines the strong light pollution abnormal initial point according to the luminance anomaly threshold. The process is simple and efficient.

[0087] In some embodiments of the present application, the pre-defined region generation algorithm is an eight-neighbor region growing algorithm with additional access conditions.

[0088] Figure 4 is a flowchart of the implementation of the eight-neighbor region growing algorithm provided by the embodiment of the present application; as shown in Figure 4 based on the luminance anomaly initial point, an abnormal region is generated by using a pre-defined region generation algorithm, which includes:

[0089] S1, an abnormal region set Z and an abnormal pixel point set P are constructed, the abnormal region set is used to store abnormal regions, and the abnormal pixel point set is used to store all non-repeated pixel points in the abnormal regions.

[0090] S2, in the case that the luminance abnormality initial point and the pixel points in the abnormal pixel point set are not the same, the luminance abnormality initial point is taken as a luminance abnormal pixel point, and eight pixel points adjacent to the luminance abnormality initial point are determined as to-be-inspected pixel points. At this time, it is explained that the luminance abnormality initial point has not yet had a corresponding abnormal area, and the abnormal area needs to be generated according to the luminance abnormality initial point.

[0091] If the luminance abnormality initial point is repeated with a pixel point in the abnormal pixel point set, it is explained that when the abnormal area is generated by other luminance abnormality initial points, the generated abnormal area has already contained the luminance abnormality initial point, and repeated abnormal area generation is not needed. Therefore, in the case of pixel point repetition, the current luminance abnormality initial point is skipped, and the next luminance abnormality initial point is judged.

[0092] In the case that the luminance abnormality initial point and the pixel points in the abnormal pixel point set are not the same, the luminance abnormality initial point is taken as a luminance abnormal pixel point, and eight pixel points adjacent to the luminance abnormality initial point are determined as to-be-inspected pixel points. At this time, it is explained that the luminance abnormality initial point has not yet had a corresponding abnormal area, and the abnormal area needs to be generated according to the luminance abnormality initial point.

[0093] S3, in the case that the to-be-inspected pixel points meet the condition of joining the abnormal area, the to-be-inspected pixel points are taken as luminance abnormal pixel points.

[0094] In this step, whether the determined eight to-be-inspected pixel points meet the condition of joining the abnormal area is further judged. The condition of joining the abnormal area needs to be determined according to the type of the abnormal area, which is described in detail below. If the to-be-inspected pixel points meet the condition of joining the abnormal area, the to-be-inspected pixel points are determined as luminance abnormal pixel points; if the to-be-inspected pixel points do not meet the condition of joining the abnormal area, the to-be-inspected pixel points are determined as normal pixel points. All the luminance abnormal pixel points are collected into the set S.

[0095] It should be noted that when judging whether the to-be-inspected pixel points meet the condition of joining the abnormal area, it is first determined that the to-be-inspected pixel points are not pixel points of other abnormal areas.

[0096] S4, the pixel points in the eight pixel points adjacent to the luminance abnormal pixel points and not yet inspected are taken as new to-be-inspected pixel points, and whether the new to-be-inspected pixel points meet the condition of joining the abnormal area is judged.

[0097] In this step, one luminance abnormal pixel point is obtained from the set S, and one pixel point not yet investigated among the eight pixel points adjacent to the luminance abnormal pixel point is obtained as a new pixel point to be investigated, and it is further judged whether the new pixel point to be investigated meets the condition of joining the abnormal region.

[0098] It should be noted that in the process of searching for the eight points adjacent to the luminance abnormal pixel point, if a point beyond the image range is encountered, i.e., the horizontal coordinate is less than 1 or greater than Width, and the vertical coordinate is less than 1 or greater than Height, the point is skipped and other adjacent points within the image range are searched.

[0099] S5, the steps S3-S4 are repeatedly executed until no new pixel point to be investigated meets the condition of joining the abnormal region, so as to obtain an abnormal region and store it into the abnormal region set, the pixel points corresponding to the abnormal region are all luminance abnormal pixel points, and all the luminance abnormal pixel points are stored into the abnormal pixel point set.

[0100] In this step, if the new pixel point to be investigated meets the condition of joining the abnormal region, the new pixel point to be investigated is stored into the set S as a new luminance abnormal pixel point, if the new pixel point to be investigated does not meet the condition of joining the abnormal region, the pixel point to be investigated is a normal pixel point, and the cycle is repeated until there is no new luminance abnormal pixel point in the set S, i.e., no new pixel point to be investigated meets the condition of joining the abnormal region, so as to obtain one abnormal region, store it into the abnormal region set Z, and store all the pixel points corresponding to the abnormal region into the abnormal pixel point set P, so as to prevent the next luminance abnormal initial point from generating a repeated region.

[0101] S6, the steps S2-S5 are repeatedly executed until all the luminance abnormal initial points are traversed, so as to obtain all the abnormal regions.

[0102] In this step, the next luminance abnormal initial point is obtained, and the steps S2-S5 are repeatedly executed until all the luminance abnormal initial points have corresponding abnormal regions.

[0103] The night multispectral low-light remote sensing image luminance abnormality detection method provided by the embodiment of the application has high detection reliability, because the abnormal region is generated by using the eight-neighbor region growing algorithm with additional access conditions according to the determined luminance abnormal initial point.

[0104] In some embodiments of the application, in the case that the type corresponding to the abnormal region is a flame abnormality, in the case that the pixel point to be investigated meets the condition of joining the abnormal region, the pixel point to be investigated is a luminance abnormal pixel point, which includes:

[0105] According to the red waveband radiant luminance value and the green waveband radiant luminance value of the pixel point to be investigated, it is judged whether the pixel point to be investigated meets the condition of joining the flame abnormal area.

[0106] The condition of the flame abnormal area is that the red waveband radiant luminance value of the pixel point to be investigated is greater than a preset third flame abnormal threshold value, or the red waveband radiant luminance value of the pixel point to be investigated is greater than a preset fourth flame abnormal threshold value and the ratio between the red waveband radiant luminance value and the green waveband radiant luminance value is greater than a preset fifth flame abnormal threshold value.

[0107] In the case that the pixel point to be investigated meets the condition of joining the flame abnormal area, the pixel point to be investigated is taken as a flame abnormal pixel point.

[0108] For example, if the red waveband radiant luminance value R of the pixel point to be investigated is greater than 0.0136 W / m 2 / sr / μm (i.e. the third flame abnormal threshold value), it is determined that the pixel point to be investigated is a flame abnormal pixel point. Alternatively, if the red waveband radiant luminance value R of the pixel point to be investigated is greater than 0.00068 W / m 2 / sr / μm (i.e. the fourth flame abnormal threshold value) and the ratio between the red waveband radiant luminance value R and the green waveband radiant luminance value G is greater than 8 (i.e. the fifth flame abnormal threshold value), i.e. R>8×G, it is determined that the pixel point to be investigated is a flame abnormal pixel point. The third flame abnormal threshold value 0.0136 W / m 2 / sr / μm, the fourth flame abnormal threshold value 0.00068 W / m 2 / sr / μm and the fifth flame abnormal threshold value 8 are obtained according to analysis and calculation, and may fluctuate in practice, which is not limited in the present application.

[0109] The night multi-spectral low-light remote sensing image luminance abnormality detection method provided by the embodiments of the present application is aimed at the typical night luminance abnormality of flame abnormality, limits the corresponding abnormal area access condition, and thus expands from the determined luminance abnormality initial point to the entire abnormal area, so that the detection result is more reliable.

[0110] In some embodiments of the present application, in the case that the type corresponding to the abnormal area is strong light pollution, the case that the pixel point to be investigated meets the condition of joining the abnormal area includes:

[0111] According to the red waveband radiant luminance value, the green waveband radiant luminance value and the blue waveband radiant luminance value of the pixel point to be investigated, it is judged whether the pixel point to be investigated meets the condition of joining the strong light pollution abnormal area.

[0112] The condition of the strong light pollution abnormal area is that the red waveband radiation brightness value of the pixel point to be investigated is greater than a fourth preset strong light pollution threshold value, the green waveband radiation brightness value is greater than a fifth preset strong light pollution threshold value, and the blue waveband radiation brightness value is greater than a sixth preset strong light pollution threshold value.

[0113] In a case where the pixel point to be investigated meets the condition of joining the strong light pollution abnormal area, the pixel point to be investigated is taken as a strong light pollution abnormal pixel point.

[0114] For example, if the red waveband radiation brightness value R of the pixel point to be investigated is greater than 0.054 W / m 2 / sr / μm (i.e., the fourth strong light pollution threshold value), the green waveband radiation brightness value G is greater than 0.020 W / m 2 / sr / μm (i.e., the fifth strong light pollution threshold value), and the blue waveband radiation brightness value B is greater than 0.020 W / m 2 / sr / μm (i.e., the sixth strong light pollution threshold value), it is determined that the pixel point to be investigated is a strong light pollution abnormal pixel point. The fourth strong light pollution threshold value 0.054 W / m 2 / sr / μm, the fifth strong light pollution threshold value 0.020 W / m 2 / sr / μm, and the sixth strong light pollution threshold value 0.020 W / m2 / sr / μm fluctuate according to actual conditions, and the application does not limit this.

[0115] The night multi-spectral low-light remote sensing image brightness abnormality detection method provided by the embodiment of the application is aimed at the typical night brightness abnormality of strong light pollution abnormality, limits the conditions met by the corresponding abnormal area, and thus expands from the determined brightness abnormality initial point to the surface, so that the detection result is more reliable.

[0116] In some embodiments of the application, the radiation calibration of the image to be detected includes:

[0117] The radiation brightness value corresponding to each pixel point in the image to be detected is obtained according to the pixel digital value of the image to be detected, the gain coefficient of the low-light sensor, and the bias coefficient.

[0118] Specifically, for the radiation calibration of the low-light image, different sensors have different methods, and the specific calibration formula is: L=DN×Gain+Bias, wherein L represents the radiation brightness of the pixel, that is, the result of the radiation calibration, the unit of L is W / m 2 / sr / μm, DN represents the pixel digital value of the image, Gain is the gain coefficient, and Bias is the bias coefficient.

[0119] The gain coefficient Gain and the bias coefficient Bias are published with the satellite sensor data specification, and will be continuously updated subsequently.

[0120] In addition to the above-mentioned common calibration method, a more refined calibration method can also be used, and the present application does not limit the specific calibration method, as long as the calibration result value represents the radiation brightness of the specified unit.

[0121] In some embodiments of the present application, after the abnormal region is generated based on the initial point of brightness anomaly using a predefined region generation algorithm, the method further comprises:

[0122] According to the geographical environment corresponding to the abnormal region, the abnormal region is investigated to obtain the investigated abnormal region. Specifically, according to the geographical information in the image, the pixel position of the abnormal region is associated with the geographical position. If it is a flame abnormal region and located in an oil and gas production area, the abnormality can be excluded. If it is a strong light pollution anomaly and located in a city area, the abnormality can also be excluded, or the abnormal region is investigated according to the geographical position of other typical night-time anomalies.

[0123] The night-time multispectral low-light remote sensing image brightness anomaly detection method provided by the embodiments of the present application can improve the detection accuracy by investigating the generated abnormal region.

[0124] It should be noted again that the threshold values mentioned above are generated by a large number of normal and abnormal samples, and the size of the determined threshold value has been determined to minimize the false alarm rate and the false alarm rate of the entire abnormal region detection result.

[0125] In addition, one feature of the night-time flame anomaly that is prone to false alarms is that due to the imaging mechanism of the sensor when shooting high-temperature and high-brightness objects, ghosting of the flame is prone to occur in the image. The multi-band characteristics of the flame ghosting are mainly that the ratio of the red band to the green band is much smaller than that of the real flame. Therefore, the flame ghosting generally appears yellow-green in the night-time low-light image. The threshold values in the present application consider the exclusion of flame ghosting, so that the abnormal detection result can avoid such interference, as shown in Figure 5 The flame ghosting appearing in the image to be detected is excluded from the detection result, and has no effect on the detection result of the flame abnormal region, and has good flame ghosting exclusion effect.

[0126] The night-time multispectral low-light remote sensing image brightness anomaly detection device provided by the present application will be described below. The night-time multispectral low-light remote sensing image brightness anomaly detection device described below can be correspondingly referred to the night-time multispectral low-light remote sensing image brightness anomaly detection method described above.

[0127] Figure 6 The night-time multispectral low-light remote sensing image brightness anomaly detection device provided by the embodiments of the present application is shown in the structural schematic diagram, as shown in Figure 6As shown, the night multispectral low-light remote sensing image brightness anomaly detection device includes an image acquisition module 601, a radiation calibration module 602, an anomaly judgment module 603, a brightness anomaly initial point acquisition module 604, and an anomaly region generation module 605.

[0128] The image acquisition module 601 is configured to acquire a to-be-detected image, which is a night multispectral low-light remote sensing image, i.e., a multi-band low-light remote sensing image imaged at night.

[0129] The radiation calibration module 602 is configured to perform radiation calibration on the to-be-detected image to obtain a radiation brightness value corresponding to each pixel point in the to-be-detected image.

[0130] In this module, the pixel value of the remote sensing image is converted into a radiation brightness value with physical meaning. The conversion process is radiation calibration. The unit of the calibrated radiation brightness value is W / m 2 / sr / μm, where W represents watt, m 2 represents square meter, sr represents steradian, and μm represents micrometer.

[0131] The specific radiation calibration method and calibration parameters can be determined according to the satellite sensor. The satellite operating mechanism has corresponding calibration parameters of the sensor. In addition, assuming that the brightness anomaly detection method is directly operated on the satellite, the satellite calibration capability is required. The threshold parameters involved in the present application can also be converted from the radiation brightness physical value to the original data value (i.e., the pixel value) of the corresponding satellite for corresponding comparison.

[0132] It should be noted that the radiation brightness value here refers to the multi-band radiation brightness value of each pixel point, including the red band radiation brightness value, the green band radiation brightness value, the blue band radiation brightness value, etc.

[0133] The anomaly judgment module 603 is configured to judge whether the radiation brightness value corresponding to each pixel point is greater than a preset brightness anomaly threshold value in sequence. That is, the brightness anomaly threshold value is set according to the typical characteristics corresponding to the flame, high-intensity light pollution, and other typical night anomalies. The brightness anomaly threshold value is used to judge whether the pixel point in the image is a pixel point of an anomaly region.

[0134] The brightness anomaly initial point acquisition module 604 is configured to, in a case where the radiation brightness value is greater than the preset brightness anomaly threshold value, take the pixel point corresponding to the radiation brightness value greater than the brightness anomaly threshold value as a brightness anomaly initial point. That is, a batch of pixel points capable of indicating the potential position of the anomaly occurrence are screened out as the brightness anomaly initial point through the brightness anomaly threshold value.

[0135] If the radiation brightness value is less than or equal to the preset brightness anomaly threshold value, it is determined that the pixel point corresponding to the radiation brightness value is a pixel point of a normal region.

[0136] The abnormal region generation module 605 is configured to generate an abnormal region based on the luminance abnormal initial point by using a predefined region generation algorithm.

[0137] In the module, since the screened luminance abnormal initial point is a pixel point with the most significant feature of typical night abnormality, and the pixel points around the pixel point can be abnormal pixel points with insignificant features or normal pixel points, it is necessary to generate an abnormal region around the determined luminance abnormal initial point, and the specific generation process is realized by using a predefined region generation algorithm, so as to obtain the abnormal region and complete the abnormality detection.

[0138] In addition, after the abnormal region is obtained, the abnormal region can be further investigated and confirmed, for example, by manual confirmation or by investigation based on the geographical position of the abnormal region.

[0139] The night multispectral low-light remote sensing image luminance abnormality detection device provided by the embodiment of the application determines a luminance abnormal initial point by using a preset luminance abnormal threshold, uses the relationship between the typical features of typical night abnormality and the spectral bands, accurately distinguishes abnormal pixel points, and detects the range of the abnormal region by using a region growing algorithm. The night multispectral low-light remote sensing image luminance abnormality detection method provided by the application has the advantages of fast and simple detection process, does not need to rely on multiple different types of data, and has high detection reliability.

[0140] Figure 7 An electronic device provided by the embodiment of the application is shown in a physical structure schematic diagram as shown in the figure, Figure 7 The electronic device can include a processor 710, a communications interface 720, a memory 730 and a communications bus 740. The processor 710, the communications interface 720 and the memory 730 can communicate with each other through the communications bus 740. The processor 710 can call the logic instructions in the memory 730 to execute the night multispectral low-light remote sensing image luminance abnormality detection method. The night multispectral low-light remote sensing image luminance abnormality detection method includes: acquiring a to-be-detected image, the to-be-detected image being a night multispectral low-light remote sensing image; performing radiation calibration on the to-be-detected image to obtain a radiation luminance value corresponding to each pixel point in the to-be-detected image; sequentially judging whether the radiation luminance value corresponding to each pixel point is greater than a preset luminance abnormal threshold; in the case where the radiation luminance value is greater than the preset luminance abnormal threshold, regarding the pixel point corresponding to the radiation luminance value greater than the luminance abnormal threshold as a luminance abnormal initial point; and generating an abnormal region based on the luminance abnormal initial point by using a predefined region generation algorithm.

[0141] In addition, the logic instructions in the memory 730 described above can be implemented in the form of software functional units and sold or used as independent products, and can be stored in a computer readable storage medium. Based on such understanding, the technical solutions of the present application essentially or the parts that contribute to the prior art or parts of the technical solutions can be embodied in the form of a software product. The computer software product is stored in a storage medium, and includes several instructions for causing a computer device (which can be a personal computer, a server, or a network device, etc.) to execute all or part of the steps of the methods of the embodiments of the present application. The aforementioned storage medium includes: a U disk, a mobile hard disk, a read-only memory (ROM, Read-Only Memory), a random access memory (RAM, Random Access Memory), a magnetic disk or an optical disk, and various media that can store program codes.

[0142] In another aspect, the present application also provides a non-transitory computer readable storage medium having stored thereon a computer program, which, when executed by a processor, implements a method for detecting brightness anomaly of night multispectral low-light remote sensing image. The method for detecting brightness anomaly of night multispectral low-light remote sensing image comprises: acquiring a to-be-detected image, the to-be-detected image being a night multispectral low-light remote sensing image; performing radiation calibration on the to-be-detected image to obtain a radiation brightness value corresponding to each pixel point in the to-be-detected image; sequentially judging whether the radiation brightness value corresponding to each pixel point is greater than a preset brightness anomaly threshold value; in the case where the radiation brightness value is greater than the preset brightness anomaly threshold value, taking the pixel point corresponding to the radiation brightness value greater than the brightness anomaly threshold value as a brightness anomaly initial point; and generating an anomaly region based on the brightness anomaly initial point by using a pre-defined region generation algorithm.

[0143] The device embodiments described above are only schematic, wherein the units illustrated as separate components can or can not be physically separate, and the components illustrated as units can or can not be physical units, i.e., can be located in one place, or can be distributed on a plurality of network units. Part or all of the modules can be selected according to actual needs to achieve the purpose of the embodiments. Those skilled in the art can understand and implement without creative labor.

[0144] Those skilled in the art can clearly understand the implementation of the various embodiments by means of software and necessary general hardware platforms through the description of the above embodiments, and of course, the implementation can also be through hardware. Based on such understanding, the above technical solutions can be embodied in the form of a software product, and the computer software product can be stored in a computer readable storage medium, such as a ROM / RAM, a magnetic disk, an optical disk, etc., and includes a plurality of instructions for causing a computer device (which can be a personal computer, a server, or a network device, etc.) to execute the method of each embodiment or some parts of the embodiment.

[0145] Finally, it should be noted that: the above examples are only used to illustrate the technical solutions of the present application, and not to limit them; although the present application has been described in detail with reference to the foregoing examples, those skilled in the art should understand that: it can still modify the technical solutions recorded in the foregoing examples, or make equivalent replacement for some technical features thereof; and these modifications or replacements do not make the essence of the corresponding technical solutions deviate from the spirit and scope of the technical solutions of the embodiments of the present application.

Claims

1. A method for detecting brightness anomalies in nighttime multispectral low-light remote sensing images, characterized in that, include: Acquire the image to be detected, which is a nighttime multispectral low-light remote sensing image; Radiometric calibration is performed on the image to be detected to obtain the radiance value corresponding to each pixel in the image to be detected. Sequentially determine whether the radiance value corresponding to each pixel is greater than the preset brightness anomaly threshold; If the radiance value is greater than a preset brightness anomaly threshold, the pixel corresponding to the radiance value that is greater than the brightness anomaly threshold is taken as the initial point of brightness anomaly. Based on the initial point of brightness anomaly, an abnormal region is generated using a predefined region generation algorithm; The predefined region generation algorithm is an eight-neighborhood region growth algorithm with additional admission conditions. Accordingly, the step of generating an abnormal region based on the initial point of brightness anomaly using a predefined region generation algorithm includes: S1, construct an abnormal region set and an abnormal pixel set. The abnormal region set is used to store abnormal regions, and the abnormal pixel set is used to store all non-repeating pixels in the abnormal regions. S2, if the initial point of brightness anomaly is different from all the pixels in the set of abnormal pixels, the initial point of brightness anomaly is taken as the pixel of brightness anomaly, and the eight pixels adjacent to the initial point of brightness anomaly are determined as pixels to be examined. S3, if the pixel to be examined meets the conditions for adding an abnormal region, the pixel to be examined is regarded as a brightness abnormal pixel. S4, take the eight pixels adjacent to the abnormal brightness pixel that have not yet been examined as new pixels to be examined, and determine whether the new pixels to be examined meet the conditions for being added to the abnormal region. S5. Repeat steps S3-S4 until no new pixels to be examined meet the conditions for adding to the abnormal region, so as to obtain the abnormal region and store it in the abnormal region set. The pixels corresponding to the abnormal region are all brightness abnormal pixels. Store all brightness abnormal pixels in the abnormal pixel set. S6. Repeat steps S2-S5 until all brightness anomaly initial points have been traversed to obtain all anomaly areas.

2. The method for detecting brightness anomalies in nighttime multispectral low-light remote sensing images according to claim 1, characterized in that, The abnormal area corresponds to a flame abnormality, and the radiance value includes red band radiance value and green band radiance value. Accordingly, when the radiance value is greater than a preset brightness anomaly threshold, the pixel corresponding to the radiance value greater than the brightness anomaly threshold is taken as the initial point of brightness anomaly, including: If the red band radiance value of a pixel is greater than the preset first flame anomaly threshold and the ratio between the red band radiance value and the green band radiance value is greater than the preset second flame anomaly threshold, the corresponding pixel will be used as the initial point of the flame anomaly.

3. The method for detecting brightness anomalies in nighttime multispectral low-light remote sensing images according to claim 1, characterized in that, The abnormal area corresponds to the type of strong light pollution, and the radiance value includes the red band radiance value, the green band radiance value and the blue band radiance value. Accordingly, when the radiance value is greater than a preset brightness anomaly threshold, the pixel corresponding to the radiance value greater than the brightness anomaly threshold is taken as the initial point of brightness anomaly, including: If the red band radiance value of a pixel is greater than the preset first strong light pollution threshold, the green band radiance value is greater than the preset second strong light pollution threshold, and the blue band radiance value is greater than the preset third strong light pollution threshold, the corresponding pixel will be designated as the initial point of strong light pollution anomaly.

4. The method for detecting brightness anomalies in nighttime multispectral low-light remote sensing images according to claim 1, characterized in that, When the type corresponding to the abnormal region is flame abnormality, the step of treating the pixel to be examined as a brightness abnormality pixel when the pixel to be examined meets the conditions for being added to the abnormal region includes: Based on the red band radiance value and the green band radiance value of the pixel to be examined, determine whether the pixel to be examined meets the conditions for being added to the flame anomaly region. The conditions for the flame anomaly region are: the red band radiance value of the pixel to be examined is greater than a preset third flame anomaly threshold; or, the red band radiance value of the pixel to be examined is greater than a preset fourth flame anomaly threshold and the ratio between the red band radiance value and the green band radiance value is greater than a preset fifth flame anomaly threshold. If the pixel to be examined meets the conditions for being included in the flame anomaly region, the pixel to be examined is designated as a flame anomaly pixel.

5. The method for detecting brightness anomalies in nighttime multispectral low-light remote sensing images according to claim 1, characterized in that, When the type corresponding to the abnormal region is strong light pollution, the step of designating the pixel to be examined as a brightness abnormal pixel when the pixel to be examined meets the conditions for being included in the abnormal region includes: Based on the red band radiance value, green band radiance value and blue band radiance value of the pixel to be examined, determine whether the pixel to be examined meets the conditions for being added to the strong light pollution abnormal region. The conditions for the strong light pollution abnormal region are that the red band radiance value of the pixel to be examined is greater than the preset fourth strong light pollution threshold, the green band radiance value is greater than the preset fifth strong light pollution threshold, and the blue band radiance value is greater than the preset sixth strong light pollution threshold. If the pixel to be examined meets the conditions for being included in the strong light pollution abnormal region, the pixel to be examined is designated as a strong light pollution abnormal pixel.

6. The method for detecting brightness anomalies in nighttime multispectral low-light remote sensing images according to claim 1, characterized in that, The step of radiometric calibration of the image to be detected to obtain the radiance value corresponding to each pixel in the image to be detected includes: The radiance value is obtained based on the pixel digital value of the image to be detected, the gain coefficient of the low-light sensor, and the bias coefficient.

7. The method for detecting brightness anomalies in nighttime multispectral low-light remote sensing images according to claim 1, characterized in that, After generating the abnormal region using a predefined region generation algorithm based on the initial point of brightness anomaly, the method further includes: The abnormal areas are investigated based on their corresponding geographical environment to obtain the investigated abnormal areas.

8. A device for detecting abnormal brightness in nighttime multispectral low-light remote sensing images, characterized in that, include: The image acquisition module is used to acquire the image to be detected, which is a nighttime multispectral low-light remote sensing image. The radiometric calibration module is used to perform radiometric calibration on the image to be detected to obtain the radiance value corresponding to each pixel in the image to be detected. The anomaly detection module is used to sequentially determine whether the radiance value corresponding to each pixel is greater than the preset brightness anomaly threshold. The brightness anomaly initial point acquisition module is used to take the pixel point corresponding to the brightness anomaly value that is greater than the preset brightness anomaly threshold as the brightness anomaly initial point when the radiation brightness value is greater than the preset brightness anomaly threshold. An abnormal region generation module is used to generate an abnormal region based on the initial point of brightness anomaly using a predefined region generation algorithm. The predefined region generation algorithm is an eight-neighborhood region growth algorithm with additional admission conditions. Accordingly, the step of generating an abnormal region based on the initial point of brightness anomaly using a predefined region generation algorithm includes: S1, construct an abnormal region set and an abnormal pixel set. The abnormal region set is used to store abnormal regions, and the abnormal pixel set is used to store all non-repeating pixels in the abnormal regions. S2, if the initial point of brightness anomaly is different from all the pixels in the set of abnormal pixels, the initial point of brightness anomaly is taken as the pixel of brightness anomaly, and the eight pixels adjacent to the initial point of brightness anomaly are determined as pixels to be examined. S3, if the pixel to be examined meets the conditions for adding an abnormal region, the pixel to be examined is regarded as a brightness abnormal pixel. S4, take the eight pixels adjacent to the abnormal brightness pixel that have not yet been examined as new pixels to be examined, and determine whether the new pixels to be examined meet the conditions for being added to the abnormal region. S5. Repeat steps S3-S4 until no new pixels to be examined meet the conditions for adding to the abnormal region, so as to obtain the abnormal region and store it in the abnormal region set. The pixels corresponding to the abnormal region are all brightness abnormal pixels. Store all brightness abnormal pixels in the abnormal pixel set. S6. Repeat steps S2-S5 until all brightness anomaly initial points have been traversed to obtain all anomaly areas.

9. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the processor executes the program, it implements the nighttime multispectral low-light remote sensing image brightness anomaly detection method as described in any one of claims 1 to 7.

Images