CN115830806B - Natural disaster condition analysis system based on high-resolution remote sensing - Google Patents

Abstract

The invention discloses a natural disaster condition analysis system based on high-resolution remote sensing, which comprises a shooting point comparison unit, a comparison abnormal point extraction unit and an abnormal image definition processing unit, wherein the shooting point comparison unit is used for comparing the shooting points of the natural disaster condition; the normal image of the same regional point and the image information of the regional point which is currently shot are combined through the shooting point comparison unit, the characteristic points of the normal image and the regional point are compared, the abnormal regional point is determined through the abnormal point comparison extraction unit, then the image of the abnormal regional point is subjected to image noise reduction through the abnormal image sharpness processing unit, so that a sharpness image is obtained, the abnormal image is analyzed before the monitoring image is not subjected to noise reduction, at the moment, the normal image is not required to be subjected to noise reduction, and only the image of the abnormal regional point is required to be subjected to noise reduction, so that noise reduction cost is saved, and the abnormal image processing efficiency is improved.

Description

Natural disaster situation analysis system based on high-resolution remote sensing

Technical Field

The present invention relates to the technical field of natural disaster monitoring, and in particular to a natural disaster situation analysis system based on high-resolution remote sensing.

Background technique

Applying remote sensing technology to the field of disaster prevention and control can give full play to the advantages of remote sensing technology itself. Research based on high-resolution remote sensing image data will make disaster assessment more accurate and provide assistance for rapid post-disaster rescue, thereby minimizing losses.

The application number is 201711289373.5, and the name of the invention is an efficient automatic early warning system for natural disasters, which includes an information collection device, a communication device, an information processing device and an early warning device. The information collection device is a vehicle-mounted mobile measurement system for obtaining measurement information of the early warning area. The communication device is used to transmit the measurement information to the information processing device. The information processing device is used to process the measurement information and generate an early warning signal based on the processing result. The early warning device is used to receive the early warning signal and issue an alarm. The system needs to analyze and process each received image, regardless of whether the received image is a normal image when no natural disaster occurs. The number of images that need to be processed every day is too large.

Summary of the invention

The purpose of the present invention is to provide a natural disaster situation analysis system based on high-resolution remote sensing to solve the problems raised in the above background technology.

To achieve the above-mentioned purpose, a natural disaster situation analysis system based on high-resolution remote sensing is provided, including a high-resolution remote sensing monitoring unit, wherein the high-resolution remote sensing monitoring unit uses high-resolution remote sensing technology to perform real-time monitoring and processing on a monitoring area, wherein an output end of the high-resolution remote sensing monitoring unit is connected to a point-by-point regional shooting unit, wherein the point-by-point regional shooting unit is used to determine monitoring points at different positions in the monitoring area, and to perform point-by-point shooting on the determined monitoring points to generate regional point image information, wherein an output end of the point-by-point regional shooting unit is connected to a shooting point comparison unit, wherein an input end of the shooting point comparison unit is further connected to a shooting point normal feature extraction unit, wherein the shooting point normal feature extraction unit extracts information based on the determined monitoring points. The normal image when no natural disaster occurs is recorded, and the normal image is transmitted to the shooting point comparison unit. The shooting point comparison unit combines the normal image of the same regional point and the image information of the currently shot regional point, compares the feature points of the two, and obtains a comparison value. The output end of the shooting point comparison unit is connected to a comparison abnormal point extraction unit. The comparison abnormal point extraction unit presets a comparison threshold. The regional point image information whose comparison value exceeds the comparison threshold is marked as an abnormal regional point. The output end of the comparison abnormal point extraction unit is connected to an abnormal image clearing processing unit. The abnormal image clearing processing unit performs image denoising on the image of the abnormal regional point to obtain a cleared image.

As a further improvement of the present technical solution, the normal feature extraction unit of the shooting point includes a regional point geographic environment analysis module, which analyzes and processes the geographic environment of the monitoring point according to real-time monitoring information. The output end of the regional point geographic environment analysis module is connected to a significant feature point marking module, and the significant feature point marking module marks the significant feature points of the monitoring area point according to the results of the geographic environment analysis and processing of the monitoring point.

As a further improvement of the technical solution, the output end of the obvious feature point marking module is connected to a feature point replacement module, and the feature point replacement module replaces the feature point position in real time according to the law of environmental changes at the monitoring point.

As a further improvement of the technical solution, the shooting point comparison unit adopts an image analysis algorithm, and its algorithm formula is as follows:

U = [μ ₁ , μ ₂ , … , μ _m ];

Where μ is the image mean of each image, i is the label of different images, n is the number of pixels, _pi is the pixel value, U is the mean set of images taken in normal state at each monitoring point, _μ1 to _μm are the mean images taken in normal state at each monitoring point, f(μ) is the mean contrast function, is the comparison value between the mean value of the currently captured image and the mean value of the image in the normal state, /> is the comparison threshold, when the comparison value/> When it is less than the contrast threshold, the mean contrast function f(μ) outputs 0, indicating that the current image is abnormal. When it is not less than the contrast threshold, the mean contrast function f(μ) outputs 1, indicating that the currently captured image is in a normal state.

As a further improvement of the present technical solution, the abnormal image clarity processing unit includes a block point ambiguity determination module, which is used to divide the abnormal images, divide each abnormal image into the same number of blocks according to the same separation method, and determine the degree of blur of each block point. The output end of the block point ambiguity determination module is connected to a feature point determination module, which is used to determine the block position of the feature point of each abnormal image. The output end of the feature point determination module is connected to a feature block point clarity module, which clarifies the blocks with feature points according to the block position of the features of each abnormal image.

As a further improvement of the technical solution, the output end of the abnormal image clearing processing unit is connected to a data storage unit, and the data storage unit is used to allocate a database to classify and store different types of data.

As a further improvement of the present technical solution, the output end of the high-resolution remote sensing monitoring unit is connected to a shooting area position determination unit, the output of the shooting area position determination unit is connected to the input end of the point area shooting unit, and the shooting area position determination unit is used to determine the geographical location of the monitoring point for each shooting.

As a further improvement of the present technical solution, the output end of the shooting area position determination unit is connected to the input end of the shooting point normal feature extraction unit, and the output end of the shooting point normal feature extraction unit is connected to the input end of the data storage unit. The shooting area position determination unit provides the shooting point normal feature extraction unit with the geographic location of the monitoring point, so that it determines the image information of the geographic location of the monitoring point under normal conditions.

As a further improvement of the present technical solution, the input end of the point area shooting unit is connected to a shooting interval determination unit, the shooting interval determination unit is used to determine the interval time of each shooting, and the output end of the shooting interval determination unit is connected to the input end of the data storage unit.

Compared with the prior art, the present invention has the following beneficial effects:

The present invention combines the normal image of the same area point and the image information of the currently shot area point through a shooting point comparison unit, compares the feature points of the two, determines the abnormal area point through a comparison abnormal point extraction unit, and then performs image denoising on the image of the abnormal area point through an abnormal image clearing processing unit to obtain a cleared image, thereby analyzing the abnormal image before the monitoring image is denoised. At this time, there is no need to perform denoising on the normal image, and only needs to perform denoising on the image of the abnormal area point, thereby saving denoising costs and improving abnormal image processing efficiency.

The present invention uses a block point fuzziness determination module to divide abnormal images, divides each abnormal image into the same number of blocks according to the same segmentation method, and determines the fuzziness of each block point. At the same time, the feature point determination module determines the block position of the feature of each abnormal image, and the feature block point clarity module performs clarity processing on the blocks with abnormal feature points. There is no need to perform comprehensive noise reduction on each abnormal image, and only needs to perform clarity processing on the blocks with abnormal feature points, thereby improving processing efficiency and reducing the processing flow of abnormal images.

The present invention determines the geographical location of the monitoring point each time it is photographed through a shooting area position determination unit to form a shooting memory. When monitoring to the determined geographical location of the monitoring point, the corresponding geographical location image is photographed for subsequent accurate comparison, thereby reducing the generation of comparison errors.

BRIEF DESCRIPTION OF THE DRAWINGS

Fig. 1 is an overall flow chart of the present invention;

FIG2 is a flow chart of a normal feature extraction unit for a shooting point according to the present invention;

FIG3 is a flow chart of an abnormal image clearing processing unit of the present invention.

The meaning of each number in the figure is:

10. High-resolution remote sensing monitoring unit;

20. Point area shooting unit;

30. Shooting point comparison unit;

40. Normal feature extraction unit of shooting point; 410. Regional point geographical environment analysis module; 420. Obvious feature point marking module; 430. Feature point replacement module;

50. Comparison and abnormal point extraction unit;

60. Abnormal image clearing processing unit; 610. Block point fuzziness determination module; 620. Feature point determination module; 630. Feature block point clearing module;

70. Data storage unit;

80. Shooting area position determination unit;

90. Shooting interval time determination unit.

Detailed ways

The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

Example 1

Please refer to Figures 1 to 3. This embodiment provides a natural disaster situation analysis system based on high-resolution remote sensing, including a high-resolution remote sensing monitoring unit 10. The high-resolution remote sensing monitoring unit 10 uses high-resolution remote sensing technology to perform real-time monitoring and processing on the monitoring area. The output end of the high-resolution remote sensing monitoring unit 10 is connected to a point-by-point regional shooting unit 20. The point-by-point regional shooting unit 20 is used to determine monitoring points at different positions in the monitoring area, and to perform point-by-point shooting of the determined monitoring points to generate regional point image information. The output end of the point-by-point regional shooting unit 20 is connected to a shooting point comparison unit 30. The input end of the shooting point comparison unit 30 is also connected to a shooting point normal feature extraction unit 31. Element 40, the normal feature extraction unit 40 of the shooting point records the normal image of the monitoring point when no natural disaster occurs according to the determined monitoring point, and transmits the normal image to the shooting point comparison unit 30. The shooting point comparison unit 30 combines the normal image of the same area point and the image information of the area point currently shot, compares the feature points of the two, and obtains the comparison value. The comparison value is to compare the images of the same position shot at two different time points. This is also formulated according to the difference of the images. For example, in desert areas, there are fewer feature points, and only sand dunes or a small amount of green plants can be used as feature points. For mountains, the feature values are relatively more, which needs to be formulated manually. The output end of the shooting point comparison unit 30 is connected to the comparison abnormal point extraction unit 50. The comparison abnormal point extraction unit 50 presets a comparison threshold. The image information of the area point whose comparison value exceeds the comparison threshold is marked as an abnormal area point. The output end of the comparison abnormal point extraction unit 50 is connected to the abnormal image clearing processing unit 60. The abnormal image clearing processing unit 60 performs image noise reduction on the image of the abnormal area point to obtain a clear image.

During specific use, the high-resolution remote sensing monitoring unit 10 uses high-resolution remote sensing technology to perform real-time monitoring and processing on the monitoring area. The high-resolution remote sensing technology uses a high-resolution remote sensing image semantic segmentation network (GBFNet) based on a global-boundary fusion network. By introducing image context, the semantic consistency between discriminant features is enhanced. At the same time, a boundary network is used to introduce boundary information to increase the feature distance between adjacent objects and alleviate the indistinguishability between classes. The point-by-point regional shooting unit 20 is used to determine monitoring points at different positions in the monitoring area. When monitoring the determined monitoring points, the point-by-point regional shooting unit 20 performs point-by-point shooting on the determined monitoring points to generate regional point image information, and transmits the regional point image information to the shooting point comparison unit 30. The shooting point normal feature extraction unit 40 records the normal image of the determined monitoring point when no natural disaster occurs, and transmits the normal image to the shooting point comparison unit 30. The shooting point comparison unit 30 combines the normal image of the same regional point and the currently shot regional point image information, compares the feature points of the two, and obtains a comparison value. The comparison abnormal point extraction unit 50 presets a comparison threshold, and the regional point image information whose comparison value exceeds the comparison threshold is marked as an abnormal regional point. The abnormal image clearing processing unit 60 performs image denoising on the image of the abnormal regional point to obtain a clear image.

The present invention combines the normal image of the same area point and the image information of the currently shot area point through the shooting point comparison unit 30, compares the feature points of the two, determines the abnormal area point through the comparison abnormal point extraction unit 50, and then performs image denoising on the image of the abnormal area point through the abnormal image clearing processing unit 60 to obtain a clear image, thereby analyzing the abnormal image before the monitoring image is denoised. At this time, there is no need to perform denoising on the normal image, only the image of the abnormal area point needs to be denoised, thereby saving denoising costs and improving abnormal image processing efficiency.

In addition, the shooting point normal feature extraction unit 40 includes a regional point geographic environment analysis module 410, which analyzes and processes the geographic environment of the monitoring point based on real-time monitoring information. The output end of the regional point geographic environment analysis module 410 is connected to a significant feature point marking module 420, which marks the significant feature points of the monitoring area point based on the monitoring point geographic environment analysis processing results. During specific use, during normal regional point monitoring and shooting, due to differences in the geographical environment of the monitoring points and uneven geographical features, if the monitoring points are photographed randomly, it is difficult to determine the comparison value during subsequent comparison. At this time, the regional point geographical environment analysis module 410 analyzes and processes the geographical environment of the monitoring point based on real-time monitoring information, and transmits the analysis and processing results to the obvious feature point marking module 420. The obvious feature point marking module 420 marks the obvious feature points of the monitoring regional points based on the analysis and processing results of the geographical environment of the monitoring points. For example, during the monitoring and prediction of natural disasters such as debris flows, the monitoring points are generally cliffs or faults. At this time, the regional point geographical environment analysis module 410 analyzes the area of the monitoring point as a cliff or fault. After the position is determined, the feature points of the cliff or fault are observed, such as rocks protruding from the edge of the cliff or fault. In the subsequent comparison process, the changes in the feature points of the two images can be judged to determine in advance whether the image of the currently monitored regional point is abnormal, thereby improving the comparison efficiency and reducing the comparison error.

Furthermore, the output end of the obvious feature point marking module 420 is connected to a feature point replacement module 430, which replaces the feature point position in real time according to the law of environmental changes at the monitoring point. In specific use, since the monitoring point is not only affected by natural disasters, but also by human influence, for example, when monitoring fires in areas with dense trees such as forests, the feature points of the monitoring area are generally selected from taller trees as feature points, but when the trees are cut down by humans, the feature points disappear, and the feature points are reselected by the feature point replacement module 430, and then the previously determined feature points are replaced.

Furthermore, the shooting point comparison unit 30 adopts an image analysis algorithm, and its algorithm formula is as follows:

U = [μ ₁ , μ ₂ , … , μ _m ];

Where μ is the image mean of each image, i is the label of different images, n is the number of pixels, _pi is the pixel value, U is the mean set of images taken in normal state at each monitoring point, _μ1 to _μm are the mean images taken in normal state at each monitoring point, f(μ) is the mean contrast function, is the comparison value between the mean value of the currently captured image and the mean value of the image in the normal state, /> is the comparison threshold, when the comparison value/> When it is less than the contrast threshold, the mean contrast function f(μ) outputs 0, indicating that the current image is abnormal. When it is not less than the contrast threshold, the mean contrast function f(μ) outputs 1, indicating that the currently captured image is in a normal state.

Specifically, the abnormal image clearing processing unit 60 includes a block point blurriness determination module 610, which is used to divide the abnormal image, divide each abnormal image into the same number of blocks according to the same separation method, and determine the blurriness of each block point. The output end of the block point blurriness determination module 610 is connected to a feature point determination module 620, which is used to determine the block position of the feature point of each abnormal image. The output end of the feature point determination module 620 is connected to a feature block point clearing module 630, which clarifies the block with feature points according to the block position of the feature of each abnormal image. During specific use, the block point blur determination module 610 is used to divide the abnormal images, and each abnormal image is divided into the same number of blocks according to the same separation method, and the blur degree of each block point is determined. At the same time, the feature point determination module 620 is used to determine the block position of the feature of each abnormal image, and the feature block point clearing module 630 is used to clarify the blocks with feature points. There is no need to perform comprehensive noise reduction on each abnormal image, and only the blocks with feature points need to be clarified, thereby improving processing efficiency and reducing the processing flow of abnormal images.

In addition, the output end of the abnormal image clearing processing unit 60 is connected to a data storage unit 70, and the data storage unit 70 is used to allocate a database and classify and store different types of data. In specific use, when the feature point determination module 620 determines the block position of the feature point of each abnormal image, generates feature point block information, and transmits the feature point block information to the data storage unit 70, the data storage unit 70 sets a feature point block database, and classifies and stores the feature point block information of each abnormal image. When performing secondary clearing processing later, only the corresponding abnormal image information needs to be input, and the corresponding feature point block information can be directly called from the data storage unit 70, without re-confirming the block position of the feature point of the abnormal image, thereby further improving the efficiency of the clearing processing of the abnormal image.

Furthermore, the output end of the high-resolution remote sensing monitoring unit 10 is connected to a shooting area position determination unit 80, and the shooting area position determination unit 80 output is connected to the input end of the point-by-point area shooting unit 20, and the shooting area position determination unit 80 is used to determine the geographical location of the monitoring point for each shooting. In specific use, in order to ensure the accuracy of comparison, during the monitoring point shooting process, it is necessary to determine the geographical location of the monitoring point for each shooting through the shooting area position determination unit 80 to form a shooting memory. When monitoring to the determined monitoring point geographical location, the corresponding geographical location image is shot for accurate comparison later, thereby reducing the generation of comparison errors.

Furthermore, the output end of the shooting area position determination unit 80 is connected to the input end of the shooting point normal feature extraction unit 40, and the output end of the shooting point normal feature extraction unit 40 is connected to the input end of the data storage unit 70. The shooting area position determination unit 80 provides the shooting point normal feature extraction unit 40 with the geographic location of the monitoring point, so that it determines the image information of the geographic location of the monitoring point under normal conditions, and transmits the geographic location of the monitoring point and the corresponding image information of the geographic location of the monitoring point under normal conditions to the data storage unit 70 for storage.

In addition, the input end of the point-area shooting unit 20 is connected to a shooting interval determination unit 90, which is used to determine the interval of each shooting, and the output end of the shooting interval determination unit 90 is connected to the input end of the data storage unit 70. When in use, the shooting interval determination unit 90 determines the interval of each shooting, generates shooting interval information, and transmits the shooting interval information to the point-area shooting unit 20. The point-area shooting unit 20 performs interval shooting of the monitoring area according to the shooting interval information, and transmits the shooting interval information to the data storage unit 70 for storage.

The above shows and describes the basic principles, main features and advantages of the present invention. Those skilled in the art should understand that the present invention is not limited by the above embodiments. The above embodiments and descriptions are only preferred examples of the present invention and are not intended to limit the present invention. Without departing from the spirit and scope of the present invention, the present invention may have various changes and improvements, which fall within the scope of the present invention. The scope of protection of the present invention is defined by the attached claims and their equivalents.

Claims (9)

1. The utility model provides a natural disaster condition analysis system based on high-resolution remote sensing, includes high-resolution remote sensing monitor cell (10), high-resolution remote sensing monitor cell (10) adopt high-resolution remote sensing technique to carry out real-time monitoring to the monitored area and handle, high-resolution remote sensing monitor cell (10) output is connected with minute point regional shooting unit (20), minute point regional shooting unit (20) are used for confirming the monitored point in the different positions of monitored area to carry out minute point shooting to the monitored point of confirming, generate regional point image information, its characterized in that: the method comprises the steps that an imaging point comparison unit (30) is connected to the output end of a point area imaging unit (20), an imaging point normal feature extraction unit (40) is further connected to the input end of the imaging point comparison unit (30), normal images of the imaging point normal feature extraction unit (40) when natural disasters do not occur are recorded according to determined monitoring points, the normal images are transmitted to the imaging point comparison unit (30), the imaging point comparison unit (30) combines normal images of the same area points and image information of the area points imaged currently, feature points of the imaging point comparison unit (30) are compared to obtain comparison values, a comparison abnormal point extraction unit (50) is connected to the output end of the imaging point comparison unit (30), the comparison abnormal point extraction unit (50) is preset with a comparison threshold, the image information of the area points with the comparison values exceeding the comparison threshold is marked as abnormal area points, the output end of the comparison abnormal point extraction unit (50) is connected with an abnormal image sharpening processing unit (60), and the abnormal image sharpening processing unit (60) carries out image noise reduction on the images of the abnormal area points to obtain sharpening images.

2. The natural disaster situation analysis system based on high-resolution remote sensing as set forth in claim 1, wherein: the shooting point normal feature extraction unit (40) comprises a regional point geographical environment analysis module (410), the regional point geographical environment analysis module (410) analyzes and processes the geographical environment of the monitored point according to real-time monitoring information, the output end of the regional point geographical environment analysis module (410) is connected with an obvious feature point marking module (420), and the obvious feature point marking module (420) marks obvious feature points of the monitored regional point according to the analysis and processing result of the geographical environment of the monitored point.

3. The natural disaster situation analysis system based on high-resolution remote sensing as set forth in claim 2, wherein: the output end of the obvious characteristic point marking module (420) is connected with a characteristic point replacing module (430), and the characteristic point replacing module (430) replaces the position of the characteristic point in real time according to the environmental change rule of the monitoring point.

4. The natural disaster situation analysis system based on high-resolution remote sensing as set forth in claim 1, wherein: the shooting point comparison unit (30) adopts an image analysis algorithm, and the algorithm formula is as follows:

U＝[μ₁,μ₂,…,μ_m]；

Wherein mu is the image mean value of each image, i is the label of different images, n is the number of pixels, p _i is the pixel value, U is the image mean value set shot in the normal state of each monitoring point, mu ₁ to mu _m are the image mean values shot in the normal state of each monitoring point, f (mu) is the mean value contrast function, Is the contrast value of the average value of the currently shot image and the average value of the image in the normal state of the image,/>For comparison threshold, when the comparison value/>When the average value comparison function f (mu) is smaller than the comparison threshold value, the output of the average value comparison function f (mu) is 0, which indicates that the currently shot image is abnormal, and when the comparison value/>And when the average value comparison function f (mu) is not smaller than the comparison threshold value, outputting the average value comparison function f (mu) to be 1, and indicating that the currently shot image is in a normal state.

5. The natural disaster situation analysis system based on high-resolution remote sensing as set forth in claim 1, wherein: the abnormal image sharpening processing unit (60) comprises a block point ambiguity determining module (610), wherein the block point ambiguity determining module (610) is used for dividing abnormal images, dividing each abnormal image into blocks with the same number according to the same separation mode, determining the ambiguity degree of each block point, the output end of the block point ambiguity determining module (610) is connected with a characteristic point determining module (620), the characteristic point determining module (620) is used for determining the block position of the characteristic point of each abnormal image, the output end of the characteristic point determining module (620) is connected with a characteristic block point sharpening module (630), and the characteristic block point sharpening module (630) is used for sharpening the block with the characteristic point according to the block position of the characteristic of each abnormal image.

6. The natural disaster situation analysis system based on high-resolution remote sensing according to claim 5, wherein: the output end of the abnormal image sharpening processing unit (60) is connected with a data storage unit (70), and the data storage unit (70) is used for distributing a database and storing different types of data in a classified mode.

7. The natural disaster situation analysis system based on high-resolution remote sensing as set forth in claim 6, wherein: the high-resolution remote sensing monitoring system is characterized in that the output end of the high-resolution remote sensing monitoring unit (10) is connected with a shooting area position determining unit (80), the output end of the shooting area position determining unit (80) is connected with the input end of the point-division area shooting unit (20), and the shooting area position determining unit (80) is used for determining the geographic position of a monitoring point shot each time.

8. The natural disaster situation analysis system based on high-resolution remote sensing as set forth in claim 7, wherein: the output end of the shooting area position determining unit (80) is connected with the input end of the shooting point normal feature extracting unit (40), the output end of the shooting point normal feature extracting unit (40) is connected with the input end of the data storage unit (70), and the shooting area position determining unit (80) provides a monitoring point geographic position for the shooting point normal feature extracting unit (40) so as to determine image information of the monitoring point geographic position in a normal state.

9. The natural disaster situation analysis system based on high-resolution remote sensing as set forth in claim 8, wherein: the input end of the sub-point region shooting unit (20) is connected with a shooting interval time determining unit (90), the shooting interval time determining unit (90) is used for determining interval time of shooting each time, and the output end of the shooting interval time determining unit (90) is connected with the input end of the data storage unit (70).