CN114757962A - Overhead line forest fire identification method - Google Patents
Overhead line forest fire identification method Download PDFInfo
- Publication number
- CN114757962A CN114757962A CN202210406699.6A CN202210406699A CN114757962A CN 114757962 A CN114757962 A CN 114757962A CN 202210406699 A CN202210406699 A CN 202210406699A CN 114757962 A CN114757962 A CN 114757962A
- Authority
- CN
- China
- Prior art keywords
- area
- image
- forest fire
- color
- threshold
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
- 238000000034 method Methods 0.000 title claims abstract description 31
- 230000011218 segmentation Effects 0.000 claims abstract description 31
- 230000000877 morphologic effect Effects 0.000 claims abstract description 10
- 238000003672 processing method Methods 0.000 claims abstract description 9
- 230000007797 corrosion Effects 0.000 claims abstract description 7
- 238000005260 corrosion Methods 0.000 claims abstract description 7
- 238000001914 filtration Methods 0.000 claims description 11
- 238000012545 processing Methods 0.000 claims description 9
- 238000012544 monitoring process Methods 0.000 claims description 7
- 230000003068 static effect Effects 0.000 claims description 7
- 230000007547 defect Effects 0.000 abstract description 6
- 238000001514 detection method Methods 0.000 abstract description 6
- 238000005516 engineering process Methods 0.000 abstract description 4
- 230000004927 fusion Effects 0.000 abstract description 2
- 239000000284 extract Substances 0.000 abstract 1
- 230000005540 biological transmission Effects 0.000 description 14
- 230000000694 effects Effects 0.000 description 9
- 238000012806 monitoring device Methods 0.000 description 6
- 238000010586 diagram Methods 0.000 description 4
- 230000001788 irregular Effects 0.000 description 3
- 238000005259 measurement Methods 0.000 description 3
- 230000002159 abnormal effect Effects 0.000 description 2
- 239000003086 colorant Substances 0.000 description 2
- 238000007689 inspection Methods 0.000 description 2
- 230000002829 reductive effect Effects 0.000 description 2
- 230000009286 beneficial effect Effects 0.000 description 1
- 230000000903 blocking effect Effects 0.000 description 1
- 238000004364 calculation method Methods 0.000 description 1
- 238000013527 convolutional neural network Methods 0.000 description 1
- 238000013135 deep learning Methods 0.000 description 1
- 238000013136 deep learning model Methods 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 230000018109 developmental process Effects 0.000 description 1
- 230000003628 erosive effect Effects 0.000 description 1
- 238000003709 image segmentation Methods 0.000 description 1
- 230000036039 immunity Effects 0.000 description 1
- 230000000670 limiting effect Effects 0.000 description 1
- 238000012423 maintenance Methods 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 230000036961 partial effect Effects 0.000 description 1
- 238000005192 partition Methods 0.000 description 1
- 238000012216 screening Methods 0.000 description 1
- 238000010187 selection method Methods 0.000 description 1
- 238000006467 substitution reaction Methods 0.000 description 1
- 238000012360 testing method Methods 0.000 description 1
- 230000000007 visual effect Effects 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/10—Segmentation; Edge detection
- G06T7/136—Segmentation; Edge detection involving thresholding
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q50/00—Information and communication technology [ICT] specially adapted for implementation of business processes of specific business sectors, e.g. utilities or tourism
- G06Q50/06—Energy or water supply
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/10—Segmentation; Edge detection
- G06T7/187—Segmentation; Edge detection involving region growing; involving region merging; involving connected component labelling
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/60—Analysis of geometric attributes
- G06T7/62—Analysis of geometric attributes of area, perimeter, diameter or volume
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/90—Determination of colour characteristics
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Business, Economics & Management (AREA)
- Health & Medical Sciences (AREA)
- Economics (AREA)
- Water Supply & Treatment (AREA)
- Primary Health Care (AREA)
- Strategic Management (AREA)
- Tourism & Hospitality (AREA)
- Marketing (AREA)
- General Business, Economics & Management (AREA)
- Human Resources & Organizations (AREA)
- General Health & Medical Sciences (AREA)
- Public Health (AREA)
- Geometry (AREA)
- Image Analysis (AREA)
- Image Processing (AREA)
Abstract
The invention belongs to the field of intelligent operation and detection of overhead lines, and particularly relates to an identification method of mountain fire of an overhead line, which analyzes color components of an image background by using HSV color space and extracts a picture suspected of containing the mountain fire by using a specific color threshold value, thereby effectively overcoming the defect that color information of the picture at night is difficult to identify and improving the identification rate of the mountain fire at night; performing region segmentation by a symmetrical cross entropy fusion self-adaptive threshold segmentation technology, and removing interference of similar color components by using an expansion and corrosion morphological image processing method; and calculating the distribution differences of different areas through KL divergence to select connected domains, then fusing adjacent profiles, further removing the connected domains which do not meet the requirement of an area threshold, externally connecting the profiles of the rest specific areas with a minimum rectangle, returning to a rectangular coordinate, and accurately identifying the hidden danger areas of the mountain fire.
Description
Technical Field
The invention belongs to the field of intelligent operation and inspection of overhead lines, and particularly relates to an overhead line forest fire identification method.
Background
Along with the upgrading of the maintenance technology of the power transmission line, the visual remote inspection of the power transmission line channel is widely applied, and on the power transmission line, the monitoring equipment regularly shoots static pictures to detect hidden dangers in the power transmission channel, but compared with other hidden dangers, the hidden danger of mountain fire is one of the important factors influencing the power transmission safety.
In recent years, with the development of deep learning technology, some overhead line forest fire identification methods based on convolutional neural networks appear, however, deep learning models need a large number of samples to be trained, and because a night scene is quite different from a day scene and is easily influenced by light, in addition, the characteristic expression of the forest fire at night in power transmission is also quite different from that of other scenes, the color of the flame center is white in an RGB space, the color of edge regions is between red and orange, and the scene samples at night are difficult to collect, so that the model has a poor night scene identification effect, and cannot achieve a good identification effect. In the early days, the color information under the RBG space is directly used for identifying the forest fire, but the identification effect is poor, and objects with the colors close to the forest fire cannot be distinguished.
In summary, how to provide a fast and reliable method for identifying the mountain fire of the overhead line to reduce the false alarm rate of identifying the hidden danger of the mountain fire at night is one of the problems to be solved urgently by the technical personnel in the field.
Disclosure of Invention
Aiming at the defects of the prior art, the invention aims to provide an identification method of the forest fire on the overhead line, wherein HSV color space is used for analyzing the color component of the image background, and a picture suspected of containing the forest fire is filtered and extracted by utilizing a specific color threshold, so that the defect that the color information of the picture at night is difficult to identify is effectively overcome, and the identification rate of the forest fire at night is improved; performing region segmentation by a symmetrical cross entropy fusion self-adaptive threshold segmentation technology, and removing interference of similar color components by using an expansion and corrosion morphological image processing method; the distribution differences of different regions are calculated through KL divergence to select the connected domains, adjacent profiles are fused, the connected domains which do not meet the requirement of an area threshold value can be further removed, the profiles of the rest specific regions are externally connected with a minimum rectangle, the rectangular coordinates are returned, and the region with the hidden danger of the mountain fire is accurately identified.
The technical problem to be solved by the invention is realized by adopting the following technical scheme: an overhead line forest fire identification method comprises the following steps:
a. acquiring a static image transmitted by monitoring equipment, converting a color space, and converting an RGB image into an HSV space; HSV is a relatively intuitive color model, according to the change of hue, brightness and color purity, the color change is expressed in a form which is easier to be perceived by human, and no direct relation exists between the brightness degree of the expressed color and the light intensity, so that in a night scene, the color characteristic is obvious, and HSV can represent the change of a more subtle color area; the HSV color space is converted through the representation space, the specific color area of the mountain fire at night is filtered, and the mountain fire color area can be well distinguished under the night scene;
b. calculating HSV average components of standard night pictures in a statistical manner, searching for the optimal parameter corresponding proportion, selecting specific color components of the mountain fire at night, carrying out color filtering on the static image and extracting the picture suspected to contain the mountain fire; the optimal parameter proportion is the color interval of the mountain fire in the HSV space, the specific selection method is calculated through experimental values, specifically, the average value is taken from multiple pictures, the color interval under the current power transmission night scene is found, and the purpose is to find the color interval corresponding to the mountain fire, so that the mountain fire area is divided according to the interval;
c. The method comprises the following steps of (1) eliminating an interference area in a picture suspected to contain mountain fire by improving an OTSU threshold segmentation mode through symmetrical cross entropy: performing threshold segmentation, and separating out a specific area suspected to contain the forest fire to obtain a binary image; processing the binary image by using a morphological processing method; because the OTSU calculates the threshold value according to an inter-class variance method, the threshold value has a good effect only on an image with an inter-class variance being a single peak, but when the size ratio of a target to a background is very different, for example, the background noise is too large, the edge shape of flame is influenced by various factors, and the like, the images are multimodal, and the symmetric cross entropy can better measure the degree of pixel probability distribution difference, so that the edge of the segmented image can be clearer by combining the symmetric cross entropy with the OTSU method to perform threshold segmentation;
d. drawing the contour of the binary image obtained in the step c, selecting a connected domain by utilizing KL divergence, extracting the contour of a specific region, calculating the area of an approximate contour through contour points to obtain the area S of a polygonal contour, removing the specific region of which the area S of the polygonal contour does not meet the requirement by utilizing an area threshold, circularly calculating all the contours, externally connecting the contours of the rest specific region with a minimum rectangle, returning to a rectangular coordinate to obtain a forest fire hidden danger region; extracting the contour of a specific filtering area by using a threshold segmentation method, calculating the pixel distribution of different neighborhoods by KL divergence due to the fact that the shape of the mountain fire is changeable, selecting a connected domain and calculating the area of the contour, finding the contour above an area threshold, deleting an interference contour, and returning a coordinate value by externally connecting a minimum rectangle, so that the position where the hidden danger target of the mountain fire occurs is accurately found, and the rapid identification is carried out;
Specifically, detecting the outline of the filtered binary image, and keeping all coordinate points of the approximate outline;
screening the extracted contour points by adopting an eight-neighborhood connected method, calculating KL divergence degrees of different regions in a blocking manner, and selecting a specific connected region; the KL divergence can measure distribution differences in different areas, a plurality of partition areas exist in the specific area separated in the step b, but adjacent areas can be the same target area, a connected area is selected by utilizing the KL divergence, namely if the distribution differences of the connected areas are small, the adjacent areas can be connected to serve as a whole area, so that the identification precision of the flame area is further improved, an eight-neighborhood connection method is adopted, the difference value is judged from 8 directions, and the precision is higher.
The technical scheme of the invention is as follows: in the step b, in an HSV (hue, saturation, value) representation space, the specific night mountain fire color component range P is [ (5, 12), (43, 255), (46, 255) ], color filtering is carried out according to the specific night mountain fire color component range P to obtain a filtering picture only in the component range, and the specific night mountain fire color component range is obtained by calculating HSV components of mountain fire regions in a plurality of power transmission night scene pictures through statistics.
The technical scheme of the invention is as follows: in step c, the threshold segmentation formula after the symmetric cross entropy improves the OTSU threshold segmentation mode is as follows:
ζ(t)＝Pa(t)μa(t)log(μa(t))+Pb(t)μb(t)log(μb(t))(I)
in formula (I), ζ (t) represents the degree of difference between the target image and the background image, t represents the threshold value used for segmenting the image, Pa(t) probability of corresponding target region image, Pb(t) probability of corresponding to background region image, μa(t) and μb(t) are the gray level mean values corresponding to the target area image and the background area image respectively;
threshold segmentation is carried out through distance measurement, so that the robustness on targets with various shapes such as mountain fires and the like can be stronger, a specific area is separated, and noise interference is eliminated; eliminating interference areas by improving an OTSU threshold segmentation mode through symmetrical cross entropy to obtain a threshold T, specifically, counting the number of each pixel in a gray level in the whole image, calculating the probability distribution P of different areas in the whole image, performing traversal search on the gray level, calculating the probability between foreground and background classes under the current gray level, and finally calculating the threshold T corresponding to the intra-class variance and the inter-class variance, so that the threshold segmentation can be performed according to the threshold T to eliminate the interference areas; the difference between the foreground and the background in the image is calculated by using the symmetric cross entropy, then a threshold T capable of distinguishing the foreground from the background is found, the segmentation effect is more fine, some interference areas are removed, the edge area is better segmented, and compared with an OTSU threshold segmentation mode, the method has stronger noise immunity, and the OTSU threshold segmentation mode with the improved symmetric cross entropy can reduce the calculation complexity and can segment the image more quickly;
The morphological processing method specifically comprises the following steps of sequentially carrying out corrosion and expansion operations, firstly removing small areas with similar color components, and then expanding the outline of a target area; in order to remove the interference of light and other similar color components, an interference area is removed through image processing modes such as expansion and corrosion, the light of the similar color components is generally a small area as a noise point, the corrosion is to use a surrounding pixel area to remove the noise point, the overall outline is reduced and then expanded to the original size, the noise point is removed because a connected area is too small, and the noise point is not used during expansion, so that only a target area is reserved; a plurality of ignition points can be distinguished according to the image segmentation threshold value as long as the connected domains are not adjacent, and the ignition points can be divided when the condition above the threshold value T is met.
The technical scheme of the invention is as follows: in the step d, the area threshold comprises a minimum area threshold P1 of the forest fire and an area value LP of the street lamp, and the specific region of which the polygonal outline area S is smaller than the minimum area threshold P1 of the forest fire and the specific region equal to the area value LP of the street lamp are removed.
Although the interference area of part of light with similar color components can be removed through threshold segmentation and morphological processing, the color components of part of light and the color components of the forest fire are overlapped and cannot be completely removed; in addition, a part of mountain fire areas are too far away from monitoring equipment, so that specific positions cannot be accurately identified, namely, the current power transmission line cannot be threatened, and in order to further eliminate the interference factors, the minimum area threshold value P1 of mountain fire and the area value LP of the street lamp are obtained by calculating the outline areas of the mountain fire and the street lamps with similar colors in a plurality of power transmission night scenes in advance; on one hand, because the color components and the outline areas of all street lamps in a power transmission scene are basically not greatly different, and the positions of the street lamps are fixed, the outline can not be changed, the calculated current outline area is compared with a preset street lamp area value LP for judgment, and the outline of a light area which is the same as the street lamp area value LP and is overlapped with the color components of the mountain fires is removed; on the other hand, on the basis of an actual scene, each power transmission line has a certain monitoring device point position, as the monitoring device has a certain monitoring distance and precision, and an area too far away from one monitoring device is close to another monitoring device point position, a minimum area threshold value P1 of forest fire is set, when the monitoring area is too far away, namely the image area is smaller than the set threshold value, the monitoring device is filtered, the target area and the specific position of the target area can be identified more quickly, namely the monitoring device with the nearest distance is determined, repeated identification is avoided, the identification efficiency is improved, and the identification accuracy is improved.
The invention also provides an application of the identification method of the forest fire on the overhead line, which is used for the stage early warning of the forest fire, particularly, the area of the minimum rectangle externally connected with the hidden danger area of the forest fire is calculated, the severity grade is divided into general grade and emergency grade according to the area value, and the warning information is sent according to the severity grade.
The invention has the beneficial effects that:
according to the method, the night pictures are processed in the HSV space, and the pictures suspected of containing the forest fire are filtered and extracted through the specific color component of the forest fire at night, so that the defect that the color information of the night pictures is not easy to identify is effectively overcome, the number of the pictures processed at the later stage is reduced, and the efficiency of forest fire identification is improved.
According to the method, the OTSU threshold segmentation mode is improved through the symmetrical cross entropy to perform threshold segmentation on the picture suspected of containing the forest fire, the specific area suspected of containing the forest fire can be clearly separated, partial noise points are removed through a morphological processing method, the edge of the specific area suspected of containing the forest fire can be clearer, and the defect that the edge of the picture suspected of containing the forest fire is unclear and changeable is effectively overcome.
And finally, drawing the outline of the specific area suspected to contain the forest fire, selecting connected domains through KL divergence for combination, calculating the area of the combined outline, further removing the connected domains which do not meet the requirement of an area threshold, externally connecting the outlines of the rest specific areas with a minimum rectangle, returning to a rectangular coordinate, accurately identifying the forest fire hidden danger area, and providing technical support for detection of the hidden dangers of the power transmission line.
The method also analyzes the severity of the forest fire through the area of the minimum rectangle externally connected with the forest fire hidden danger area, performs grading early warning, can provide more field information for subsequent processing, and is convenient for adjusting a processing scheme.
Drawings
Fig. 1 is a schematic flow diagram of an overhead line forest fire identification method according to the present invention;
FIG. 2 is a schematic diagram of an embodiment of an HSV color space of the present invention;
FIG. 3 is a color filter graph of an embodiment of the present invention;
figure 4 is a diagram of interference region processing according to an embodiment of the present invention,
FIG. 5 is a profile detection diagram of an embodiment of the present invention;
FIG. 6 is a graph showing the results of the detection according to the embodiment of the present invention.
Detailed Description
In order to make the technical solutions better understood by those skilled in the art, the technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the drawings in the embodiments of the present application, and it is obvious that the described embodiments are only a part of the embodiments of the present application, and not all embodiments. All other embodiments, which can be derived by a person skilled in the art from the embodiments given herein without making any creative effort, shall fall within the protection scope of the present application.
As shown in fig. 1, a method for identifying an overhead line forest fire includes the following steps:
a. And acquiring a static image transmitted by the monitoring equipment, converting a color space, and converting the RGB image into an HSV space.
b. And (3) counting and calculating HSV average components of the standard night pictures, searching the corresponding proportion of the optimal parameters, selecting specific color components of the mountain fire at night, carrying out color filtering on the static image, and extracting the pictures suspected to contain the mountain fire.
In the step b, in an HSV (hue, saturation, value) representation space, the specific color component range P of the mountain fire at night is [ (5, 12), (43, 255), (46, 255) ], and color filtering is performed according to the specific color component range P of the mountain fire at night to obtain a filtering picture only existing in the component range.
c. The method comprises the following steps of (1) eliminating an interference area in a picture suspected to contain mountain fire by improving an OTSU threshold segmentation mode through symmetrical cross entropy: and (4) performing threshold segmentation, and separating out a specific area suspected to contain the forest fire to obtain a binary image.
The threshold segmentation formula after the symmetric cross entropy improvement of the OTSU threshold segmentation mode is as follows:
ζ(t)＝Pa(t)μa(t)log(μa(t))+Pb(t)μb(t)log(μb(t))(I)
in formula (I), ζ (t) represents the degree of difference between the target image and the background image, t represents a threshold value used for segmenting the image, and Pa(t) probability of corresponding target region image, Pb(t) probability of corresponding background region image, μa(t) and μ bAnd (t) are the gray level mean values corresponding to the target area image and the background area image respectively. According to image binarization, an OTSU threshold segmentation mode is adopted, the defect that the shape of a target is insensitive is improved, symmetrical cross entropy is added for distance measurement, namely, the difference is measured by using the cross entropy, the asymmetrical cross entropy is symmetry which does not meet the distance measurement, the symmetrical cross entropy has better noise resistance and higher running speed compared with the asymmetrical cross entropy; the forest fire edge has motion characteristics, because the cross entropy can describe the degree of probability distribution difference, threshold segmentation is carried out based on the cross entropy, the flame shape is irregular, the OTSU only roughly separates a target from a background, the irregular target cannot well measure edge detail information, the symmetrical cross entropy compares images before and after segmentation, the robustness on the change of the motion region shape is good, and the motion target in a small region, namely the image with changeable and irregular shape edges, can be accurately extracted through the distribution probability of pixel point gray values.
And processing the binary image by using a morphological processing method. The morphological processing method specifically comprises the steps of sequentially carrying out corrosion and expansion operations, removing small areas with similar color components, and then expanding the outline of a target area.
d. D, drawing contours of the binary image obtained in the step c, selecting a connected domain by using KL divergence, extracting the contour of a specific region, calculating the area of an approximate contour through contour points to obtain a polygonal contour area S, removing the specific region of which the polygonal contour area S does not meet the requirement by using an area threshold, circularly calculating all contours, then externally connecting the contours of the rest specific region with a minimum rectangle, returning to a rectangular coordinate, and obtaining the mountain fire hidden danger region. In the step d, the area threshold comprises a minimum area threshold P1 of the forest fire and an area value LP of the street lamp, and the specific area of the polygonal outline area S smaller than the minimum area threshold P1 of the forest fire and the specific area equal to the area value LP of the street lamp are removed.
The application of the identification method of the forest fire on the overhead line is used for the stage early warning of the forest fire, specifically, the area of a minimum rectangle externally connected with a forest fire hidden danger area is calculated, the severity level is divided into general and emergency according to the area value, and warning information is sent according to the severity level.
Example 1
In an offline area of a power transmission line in a certain province, mountain fire hidden dangers occur at night, and hidden danger detection is carried out on a series of abnormal images collected at the front end.
The abnormal picture testing process is as follows:
To-be-detected picture name: 99000843200301_20210304163003.jpg
1) The HSV color space is transformed, the color components of a particular target are analyzed, and the effect map of the processed image is shown in fig. 2.
2) The color component filtering is continued for the specific target area, and the OTSU threshold segmentation is performed on the image, and the effect graph of processing the image is shown in fig. 3.
3) The effect graph of the divided image is shown in fig. 4, in which the divided image is processed by erosion and expansion to remove the interference of the irrelevant area.
4) And detecting the outer contour, keeping all contour points, selecting a connected domain, drawing the outer contour, and extracting the contour, wherein an effect graph is shown in fig. 5.
And calculating the area of the outer contour, finding out the minimum circumscribed rectangle of the specific target region, and identifying the hidden danger of the mountain fire in the original image, wherein the detection result is shown in fig. 6.
The above embodiments are only used for illustrating the technical solutions of the present application, and not for limiting the same; although the present application has been described in detail with reference to the foregoing embodiments, it should be understood by those of ordinary skill in the art that: the technical solutions described in the foregoing embodiments may still be modified, or some technical features may be equivalently replaced; and such modifications or substitutions do not depart from the spirit and scope of the corresponding technical solutions in the embodiments of the present application.
Claims (5)
1. An overhead line forest fire identification method is characterized by comprising the following steps:
a. acquiring a static image transmitted by monitoring equipment, converting a color space, and converting an RGB image into an HSV space;
b. calculating HSV average components of standard night pictures in a statistical manner, selecting specific color components of the mountain fire at night to perform color filtering on the static image, and extracting the picture suspected to contain the mountain fire;
c. the method comprises the following steps of (1) eliminating an interference area in a picture suspected of containing mountain fires by improving an OTSU threshold segmentation mode through symmetrical cross entropy: performing threshold segmentation, and separating out a specific area suspected to contain the forest fire to obtain a binary image; processing the binary image by using a morphological processing method;
d. drawing the contour of the binary image obtained in the step c, selecting a connected domain by utilizing KL divergence, extracting the contour of a specific region, calculating the area of an approximate contour through contour points to obtain the area S of a polygonal contour, removing the specific region of which the area S of the polygonal contour does not meet the requirement by utilizing an area threshold value, circularly calculating all the contours, externally connecting the contours of the rest specific region with a minimum rectangle, and returning to a rectangular coordinate to obtain the forest fire hidden danger region.
2. The overhead line forest fire identification method according to claim 1, characterized in that: in the step b, in an HSV (hue, saturation, value) representation space, the specific color component range P of the mountain fire at night is [ (5, 12), (43, 255), (46, 255) ], and color filtering is performed according to the specific color component range P of the mountain fire at night to obtain a filtering picture only existing in the component range.
3. The overhead line forest fire identification method according to claim 1, characterized in that: in step c, the threshold segmentation formula after the symmetric cross entropy improves the OTSU threshold segmentation mode is as follows:
ζ(t)＝Pa(t)μa(t)log(μa(t))+Pb(t)μb(t)log(μb(t)) (I)
in formula (I), ζ (t) represents the degree of difference between the target image and the background image, t represents a threshold value used for segmenting the image, and Pa(t) probability of corresponding target region image, Pb(t) probability of corresponding background region image, μa(t) and μb(t) are the gray level mean values corresponding to the target area image and the background area image respectively;
the morphological processing method specifically comprises the steps of sequentially carrying out corrosion and expansion operations, removing small areas with similar color components, and then expanding the outline of a target area.
4. The overhead line forest fire identification method according to claim 1, characterized in that: in the step d, the area threshold comprises a minimum area threshold P1 of the forest fire and an area value LP of the street lamp, and the specific region of which the polygonal outline area S is smaller than the minimum area threshold P1 of the forest fire and the specific region equal to the area value LP of the street lamp are removed.
5. An overhead line forest fire identification method as claimed in any one of claims 1 to 4, wherein the method is used for forest fire grading early warning, specifically, the area of the minimum rectangle circumscribing the hidden danger area of forest fire is calculated, the severity grade is divided according to the area value, the forest fire is divided into general grade and emergency grade, and warning information is sent according to the severity grade.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210406699.6A CN114757962A (en) | 2022-04-18 | 2022-04-18 | Overhead line forest fire identification method |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210406699.6A CN114757962A (en) | 2022-04-18 | 2022-04-18 | Overhead line forest fire identification method |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN114757962A true CN114757962A (en) | 2022-07-15 |
Family
ID=82330864
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202210406699.6A Pending CN114757962A (en) | 2022-04-18 | 2022-04-18 | Overhead line forest fire identification method |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN114757962A (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN116863333A (en) * | 2023-06-28 | 2023-10-10 | 深圳市名通科技股份有限公司 | AI intelligent detection method for FSU equipment working state |
-
2022
- 2022-04-18 CN CN202210406699.6A patent/CN114757962A/en active Pending
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN116863333A (en) * | 2023-06-28 | 2023-10-10 | 深圳市名通科技股份有限公司 | AI intelligent detection method for FSU equipment working state |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN107085714B (en) | Forest fire detection method based on video | |
| CN106373320B (en) | Method for recognizing fire disaster based on flame color dispersion and sequential frame image similarity | |
| CN103208126B (en) | Moving object monitoring method under a kind of physical environment | |
| CN106780438A (en) | Defects of insulator detection method and system based on image procossing | |
| CN113139521B (en) | Pedestrian boundary crossing monitoring method for electric power monitoring | |
| CN104331710B (en) | On off state identifying system | |
| CN107590499B (en) | Video-based equipment LED indicator lamp state monitoring method and system | |
| CN112507865B (en) | Smoke identification method and device | |
| CN110675588B (en) | Forest fire detection device and method | |
| CN103442209A (en) | Video monitoring method of electric transmission line | |
| CN107730517B (en) | A kind of fire video image analysis algorithm based on adaption brightness segmentation | |
| CN103761529A (en) | Open fire detection method and system based on multicolor models and rectangular features | |
| CN106056139A (en) | Forest fire smoke/fog detection method based on image segmentation | |
| CN114202646A (en) | Infrared image smoking detection method and system based on deep learning | |
| CN107169966B (en) | Power transmission line forest fire identification method based on temperature distribution | |
| CN107341806A (en) | Based on the detection of the indoor indicator lamp of color and shape facility and condition discrimination method | |
| CN114757962A (en) | Overhead line forest fire identification method | |
| CN114387269A (en) | Fiber yarn defect detection method based on laser | |
| CN114677406A (en) | Method for identifying petroleum flame from video stream by using accumulated frame difference and color statistics | |
| CN113221603A (en) | Method and device for detecting shielding of monitoring equipment by foreign matters | |
| CN109800693B (en) | Night vehicle detection method based on color channel mixing characteristics | |
| CN113298027B (en) | Flame detection method and device, electronic equipment and storage medium | |
| CN115810161A (en) | Transformer substation fire identification method and system | |
| Lai et al. | Advanced real time fire detection in video surveillance system | |
| CN107590418A (en) | A kind of video smoke recognition methods based on behavioral characteristics |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination |