CN117079167A - Image processing-based high-rise fire-fighting unmanned aerial vehicle monitoring method - Google Patents
Image processing-based high-rise fire-fighting unmanned aerial vehicle monitoring method Download PDFInfo
- Publication number
- CN117079167A CN117079167A CN202311346513.3A CN202311346513A CN117079167A CN 117079167 A CN117079167 A CN 117079167A CN 202311346513 A CN202311346513 A CN 202311346513A CN 117079167 A CN117079167 A CN 117079167A
- Authority
- CN
- China
- Prior art keywords
- curve
- convex hull
- obtaining
- end point
- sub
- 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.)
- Granted
Links
- 238000000034 method Methods 0.000 title claims abstract description 56
- 238000012544 monitoring process Methods 0.000 title claims abstract description 31
- 239000000779 smoke Substances 0.000 claims abstract description 95
- 238000004364 calculation method Methods 0.000 claims description 2
- 238000001514 detection method Methods 0.000 abstract description 22
- 230000001788 irregular Effects 0.000 description 8
- 238000009792 diffusion process Methods 0.000 description 5
- 239000003086 colorant Substances 0.000 description 4
- 239000000126 substance Substances 0.000 description 4
- 239000000463 material Substances 0.000 description 3
- 239000002245 particle Substances 0.000 description 3
- 238000010276 construction Methods 0.000 description 2
- 238000010586 diagram Methods 0.000 description 2
- 230000000694 effects Effects 0.000 description 2
- 238000005516 engineering process Methods 0.000 description 2
- 238000005192 partition Methods 0.000 description 2
- 230000000750 progressive effect Effects 0.000 description 2
- 230000000007 visual effect Effects 0.000 description 2
- 235000002918 Fraxinus excelsior Nutrition 0.000 description 1
- 208000027418 Wounds and injury Diseases 0.000 description 1
- 238000004458 analytical method Methods 0.000 description 1
- 239000002956 ash Substances 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 230000005540 biological transmission Effects 0.000 description 1
- 238000002485 combustion reaction Methods 0.000 description 1
- 230000006378 damage Effects 0.000 description 1
- 230000018109 developmental process Effects 0.000 description 1
- 238000003708 edge detection Methods 0.000 description 1
- 238000001914 filtration Methods 0.000 description 1
- 239000007789 gas Substances 0.000 description 1
- 208000014674 injury Diseases 0.000 description 1
- 238000004519 manufacturing process Methods 0.000 description 1
- 230000003014 reinforcing effect Effects 0.000 description 1
- 230000000630 rising effect Effects 0.000 description 1
- 230000011218 segmentation Effects 0.000 description 1
- 238000009423 ventilation Methods 0.000 description 1
Classifications
-
- G—PHYSICS
- G08—SIGNALLING
- G08B—SIGNALLING OR CALLING SYSTEMS; ORDER TELEGRAPHS; ALARM SYSTEMS
- G08B17/00—Fire alarms; Alarms responsive to explosion
- G08B17/12—Actuation by presence of radiation or particles, e.g. of infrared radiation or of ions
- G08B17/125—Actuation by presence of radiation or particles, e.g. of infrared radiation or of ions by using a video camera to detect fire or smoke
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/20—Image preprocessing
- G06V10/26—Segmentation of patterns in the image field; Cutting or merging of image elements to establish the pattern region, e.g. clustering-based techniques; Detection of occlusion
- G06V10/267—Segmentation of patterns in the image field; Cutting or merging of image elements to establish the pattern region, e.g. clustering-based techniques; Detection of occlusion by performing operations on regions, e.g. growing, shrinking or watersheds
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/40—Extraction of image or video features
- G06V10/46—Descriptors for shape, contour or point-related descriptors, e.g. scale invariant feature transform [SIFT] or bags of words [BoW]; Salient regional features
- G06V10/462—Salient features, e.g. scale invariant feature transforms [SIFT]
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V20/00—Scenes; Scene-specific elements
- G06V20/10—Terrestrial scenes
- G06V20/17—Terrestrial scenes taken from planes or by drones
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V20/00—Scenes; Scene-specific elements
- G06V20/10—Terrestrial scenes
- G06V20/176—Urban or other man-made structures
-
- G—PHYSICS
- G08—SIGNALLING
- G08B—SIGNALLING OR CALLING SYSTEMS; ORDER TELEGRAPHS; ALARM SYSTEMS
- G08B31/00—Predictive alarm systems characterised by extrapolation or other computation using updated historic data
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Multimedia (AREA)
- Theoretical Computer Science (AREA)
- Business, Economics & Management (AREA)
- Emergency Management (AREA)
- Computing Systems (AREA)
- Remote Sensing (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Image Analysis (AREA)
Abstract
The application relates to the technical field of image processing, in particular to a high-rise fire unmanned aerial vehicle monitoring method based on image processing, which comprises the steps of collecting a high-rise building image, obtaining an edge image of the high-rise building image, and obtaining the steepness of a convex hull of a first curve according to the connecting line slope and the position distance between the left end point, the right end point and an upper convex point of each convex hull curve in the edge image; obtaining the contour steepness of each sub-region according to the convex hull steepness of the first curve; obtaining the contour irregularity of each sub-region according to the contour steepness of each sub-region; obtaining a significant value of each sub-region according to the profile irregularity of each sub-region; and obtaining the smoke area in the high-rise building image according to the salient values of the subareas. Therefore, the smoke detection of the high-rise building is realized, the accuracy of the CA significance detection algorithm on the smoke detection generated by the fire disaster is improved, and the high-rise unmanned aerial vehicle monitoring precision is realized.
Description
Technical Field
The application relates to the technical field of image processing, in particular to a high-rise fire-fighting unmanned aerial vehicle monitoring method based on image processing.
Background
Along with the rapid innovation and development of unmanned aerial vehicle technology, production cost is continuously reduced, but unmanned aerial vehicle's performance is continuously improved, and the flight is more stable, convenient operation, and fly height is showing and is improving, and data signal processing, transmission also have obvious reinforcing, and modern unmanned aerial vehicle can accomplish the operation on various complex environment high efficiency. When emergency occurs, unmanned aerial vehicle has high advantage, can be easier, swift obtain effective information such as outside stranded personnel, smog, flame in the high-rise building, provide comparatively comprehensive detection visual angle, and high-rise building monitoring normal operational environment is dangerous in addition, and the injury risk is higher, uses unmanned aerial vehicle monitoring to avoid personnel to expose in dangerous environment, reduces potential risk.
When the saliency detection is carried out on the high-rise building collected by the unmanned aerial vehicle, as the traditional CA saliency algorithm is only suitable for the conditions that the color difference between a target area and a background area is large and the background color is single, in the high-rise building image, the background area is complex and the colors are not uniform, and more buildings and facilities with similar smoke colors can exist.
In summary, the application provides a high-rise fire-fighting unmanned aerial vehicle monitoring method based on image processing, which is characterized in that high-rise building images are collected, contour irregularity construction of each subarea is constructed according to smoke contour features, the salient values of each subarea are calculated according to the contour irregularity, whether smoke generated by fire disaster exists in the high-rise building images is judged according to the salient values of each subarea, and the high-rise fire-fighting unmanned aerial vehicle monitoring method has high detection precision.
Disclosure of Invention
In order to solve the technical problems, the application provides a high-rise fire unmanned aerial vehicle monitoring method based on image processing, which aims to solve the existing problems.
The application discloses a high-rise fire unmanned aerial vehicle monitoring method based on image processing, which adopts the following technical scheme:
the embodiment of the application provides a high-rise fire unmanned aerial vehicle monitoring method based on image processing, which comprises the following steps:
collecting a high-rise building image and obtaining a corresponding edge image; acquiring a convex hull curve in an edge image by a chain code method;
obtaining each suspected smoke convex hull curve in the edge image according to the length of the convex hull curve, and marking the curve as a first curve; acquiring a left end point, a right end point and an upper salient point on a first curve; obtaining a convex hull inclination index of the first curve according to the slopes of the connecting lines among the left end point, the right end point and the upper convex points; obtaining a convex hull sharp index of the first curve according to the position distances among the left end point, the right end point and the upper convex point; obtaining the steepness of the convex hull of the first curve according to the convex hull inclination index and the convex hull sharpness index of the first curve; equally dividing the high-rise building image into a plurality of sub-areas; obtaining the contour steepness of each subarea according to the convex hull steepness of the first curve in each subarea; obtaining the contour irregularity of each sub-region according to the contour steepness of each sub-region;
obtaining a significant value of each sub-region according to the profile irregularity of each sub-region; and obtaining the smoke area in the high-rise building image according to the salient values of the subareas.
Preferably, the obtaining each suspected smoke convex hull curve in the edge image according to the convex hull curve length specifically includes:
presetting a length threshold; taking the number of pixel points on each convex hull curve as the length of each convex hull curve; and taking the convex hull curve with the length larger than the length threshold value as a suspected smoke convex hull curve.
Preferably, the obtaining the left end point, the right end point and the upper bump on the first curve specifically includes:
taking the pixel point with the smallest abscissa on the first curve as a left endpoint; the pixel point with the largest abscissa is taken as the right endpoint; the pixel point with the largest ordinate is taken as the upper salient point.
Preferably, the convex hull inclination index of the first curve is obtained according to the slopes of the connecting lines among the left end point, the right end point and the upper convex point, and the method specifically includes:
recording the slope of the connecting line of the left end point and the right end point as a first slope; marking the slope of the connecting line of the left end point and the upper salient point as a second slope; marking the slope of the connecting line of the right end point and the upper salient point as a third slope; calculating an average value of absolute values of the second slope and the third slope; taking the product of the absolute value of the first slope and the average value as a convex hull inclination index of the first curve.
Preferably, the convex hull sharp index of the first curve is obtained according to the position distances among the left end point, the right end point and the upper convex point, and the method specifically includes:
taking the position distance between the left end point and the right end point as the width of the first curve; taking the position distance between the left end point and the upper salient point as the first height of a first curve; taking the position distance between the right end point and the upper salient point as the second height of the first curve; and calculating a sum value of the first height and the second height, and taking the ratio of the sum value to the width as a convex hull sharp index of the first curve.
Preferably, the steepness of the convex hull of the first curve is: and taking the product of the convex hull inclination index and the convex hull sharpness index of the first curve as the convex hull steepness of the first curve.
Preferably, the obtaining the contour steepness of each subarea according to the convex hull steepness of the first curve in each subarea specifically includes:
for each subarea, calculating the ratio of the number of pixel points on each first curve in the subarea to the number of all pixel points on the corresponding complete convex hull curve; calculating the product of the steepness of the convex hulls of the first curves and the ratio; calculating the sum of the products of all the first curves in the region; the sum is taken as the contour steepness of the sub-region.
Preferably, the obtaining the profile irregularity of each sub-region according to the profile steepness of each sub-region specifically includes: for each subarea, taking the product of the first curve number and the contour steepness in the subarea as the contour irregularity of the subarea.
Preferably, the obtaining the salient value of each sub-area according to the contour irregularity of each sub-area specifically includes:
for the firstThe first region is obtained by CA algorithmAn average value of the degree of difference between each of the sub-regions and each of the other sub-regions; calculate the firstCalculating the product of the contour irregularity of each sub-region and the average value, and calculating the reciprocal of an exponential function taking the product as an index and taking a natural constant as a base; taking the difference between 1 and the calculated result as the firstSignificant values for individual sub-regions.
Preferably, the method obtains the smoke area in the high-rise building image according to the significant value of each subarea, specifically: presetting a significant threshold; taking the subarea with the significance value larger than the significance threshold value as a smoke area; a sub-region with a significance value less than the significance threshold is taken as a non-smoke region.
The application has at least the following beneficial effects:
according to the application, the unmanned aerial vehicle is used for collecting the image of the high-rise building, and the smoke possibly occurring in the image is detected by adopting the improved CA significance detection algorithm to detect the smoke, so that the problem that the smoke area in the image cannot be accurately detected due to the fact that the smoke is similar to the background color under the complex background can not be considered by the traditional CA algorithm is solved, the smoke detection accuracy is improved, and the smoke monitoring precision of the high-rise building is higher;
the application provides a high-rise fire unmanned aerial vehicle monitoring method based on image processing, which is characterized in that the steepness of a convex hull of a first curve is obtained according to the connecting line slope and the position distance between the left and right end points and the upper convex point of each convex hull curve in an edge image of a high-rise building image; obtaining the contour steepness of each sub-region according to the convex hull steepness of the first curve; obtaining the contour irregularity of each sub-region according to the contour steepness of each sub-region; obtaining the salient value of each subarea according to the outline irregularity of each subarea and a CA salient detection algorithm; and (3) obtaining the smoke areas in the high-rise building image according to the salient values of the areas, so that the smoke detection of the high-rise building is finished, and the accuracy of the CA salient detection algorithm on the smoke detection generated by the fire is improved.
Drawings
In order to more clearly illustrate the embodiments of the application or the technical solutions and advantages of the prior art, the following description will briefly explain the drawings used in the embodiments or the description of the prior art, and it is obvious that the drawings in the following description are only some embodiments of the application, and other drawings can be obtained according to the drawings without inventive effort for a person skilled in the art.
FIG. 1 is a flow chart of a high-rise fire unmanned aerial vehicle monitoring method based on image processing;
FIG. 2 is an exemplary diagram of a high-rise building image;
FIG. 3 is a diagram illustrating the positions of the left and right ends and the upper bump on the first curve.
Detailed Description
In order to further explain the technical means and effects adopted by the application to achieve the preset aim, the following is a detailed description of a specific implementation, structure, characteristics and effects of the high-rise unmanned aerial vehicle monitoring method based on image processing according to the application, which is provided by combining the accompanying drawings and the preferred embodiment. In the following description, different "one embodiment" or "another embodiment" means that the embodiments are not necessarily the same. Furthermore, the particular features, structures, or characteristics of one or more embodiments may be combined in any suitable manner.
Unless defined otherwise, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this application belongs.
The application provides a high-rise fire unmanned aerial vehicle monitoring method based on image processing, which is specifically described below with reference to the accompanying drawings.
The embodiment of the application provides a high-rise fire unmanned aerial vehicle monitoring method based on image processing.
Specifically, the following method for monitoring a high-rise fire unmanned aerial vehicle based on image processing is provided, please refer to fig. 1, and the method comprises the following steps:
and S001, collecting high-rise building images through the unmanned aerial vehicle.
Use unmanned aerial vehicle to shoot high-rise building image in the air, because high-rise air current is unstable, can lead to unmanned aerial vehicle to take photo by plane when the operation of taking photo by plane because of the impact of air current and take place the camera shake, simultaneously accelerate suddenly, the speed reduction also can lead to the organism to take place the shake when unmanned aerial vehicle, and the cloud platform camera on the market has better anti-shake function, consequently this embodiment uses the unmanned aerial vehicle of carrying on the cloud platform camera as image acquisition equipment, shoots urban high-rise building, obtains the RGB image of high-rise building. Converting the RGB image into a gray image, and denoising the gray image by adopting median filtering.
Step S002, obtaining a suspected smoke convex hull curve in the high-rise building image according to the smoke contour characteristics, and constructing contour irregularity of each subarea according to the steepness degree of the convex hull curve.
The internal condition of the high-rise building is complex, and fire inflammables are complex and various, but most of the high-rise building is a chemical product, and more substances such as particles, harmful gases, smoke and the like can be released during combustion. Smoke is one of the most obvious characteristics of a fire, and the size and the critical degree of the fire can be judged by evaluating the information such as the concentration, the diffusion degree and the like of the smoke.
When a fire disaster occurs in a high-rise building, surrounding combustible materials can be rapidly ignited to form a big fire, meanwhile, as most of combustible materials are chemical products, a large amount of black particles, ashes and other visible substances can be generated by burning the combustible materials to form thick black smoke, the smoke can flow and spread along with air, at the moment, the smoke is ignited indoors to generate smoke, and then the smoke is spread outdoors through windows and ventilation openings to form a strong black smoke pillar, and the smoke is always upwardly changed and spread due to temperature and other reasons. The smoke clouds formed by the smoke can be overlapped and spread upwards one by one, the outline curve of the smoke inside is complex, most of the smoke clouds are irregular and symmetrical, the smoke particles are more, the color is heavier, the smoke clouds are more overlapped and present an upward spreading trend, the smoke clouds gradually go upwards and flow layer by layer, most of the outline curve of the smoke inside also has a rising and protruding trend, the outline curve is presented as a plurality of upward convex hulls on the outline curve on an image, and the outline of the convex hulls of the smoke is also irregular, as shown in fig. 2. The non-smoke areas in the image, such as high-rise buildings, windows and roofs, are relatively simple in outline, relatively regular in shape, and relatively symmetrical and regular as a whole, have no upward diffusion trend, and hardly have convex hull curves.
Firstly, carrying out edge detection on a gray level image of a high-rise building image through a Canny operator, and detecting smoke and outline parts of other objects in the image to obtain an edge image of the object in the high-rise building image. The smoke integrally shows an upward diffusion trend, the smoke cloud layer by layer is upward progressive, the smoke contour curve also has an upward and convex trend, so that the contour curve inside the smoke is provided with a plurality of curved upward smoke contour convex hull curves, the smoke contour is mostly a longer curve, and a large number of smoke contour convex hulls possibly exist on one curve, so that each upward convex curve in the edge image is obtained through a chain code method and used as each convex hull curve, and the chain code method is a known technology, and is not repeated in a specific process.
Because the number of shorter convex hull curves in the edge image is too large, in order to reduce the calculation amount, the obtained convex hull curves are processed as follows: setting a length threshold value by taking the number of pixel points on each convex hull curve as the length of each convex hull curveIt should be noted that, the present application,can be taken by a value implementer of (a)The embodiment is not particularly limited, and is set by itself. When the length of the convex hull curve is larger than the length threshold, the convex hull curve is considered longer and is used as a suspected smoke convex hull curve in the edge image; when the length of the convex hull curve is not greater than the length threshold, the convex hull curve is considered to be shorter and is not used as a suspected smoke convex hull curve in the edge image, the convex hull curve is eliminated, and the subsequent processing analysis is not carried out. And marking the suspected smoke convex hull curve in the edge image as a first curve.
Acquiring first curves in the edge image through the processing, respectively marked asWhereinFor the number of the first curves in the edge image, taking the pixel point with the minimum abscissa value on the first curve as the left end point of the first curveThe pixel point with the largest abscissa value is taken as the right end point of the first curveThe pixel point with the largest ordinate value is taken as the upper salient point of the first curveAs shown in fig. 3.
By the first in the edge imageA first curveFor example, curveLength is recorded as. The slope of the connecting line is obtained through the coordinates of the left endpoint and the right endpoint and is recorded as a first slopeThe slope of the connecting line obtained through the coordinates of the left end point and the upper salient point is recorded as a second slopeThe slope of the connecting line obtained through the coordinates of the right end point and the upper salient point is recorded as a third slopeThe method for calculating the slope is a well-known technique and will not be described here. The absolute value of the slope can reflect the inclination degree of the straight line according to mathematical knowledge, and the larger the absolute value of the slope is, the closer the straight line is to the vertical straight line, and the larger the inclination degree is; the smaller the absolute value of the slope, the closer the line is to a horizontal line, and the smaller the degree of tilt. In the first curve of the graph, the first curve,the slope of the connecting line representing the left and right end points is taken as the slope of the whole first curve,the larger the absolute value, the more vertical the whole of the first curve, the greater the inclination of the first curve, whileAndthe slopes of the connection lines of the left and right end points and the upper convex point on the first curve respectively reflect the inclination degree of the left and right end points and the upper convex point on the convex hull curve,andthe larger the absolute value of (c), the more vertical the upper convex hull is to the line of the endpoint, and the greater the inclination of the convex hull curve. Constructing convex hull inclination index of first curve according to slopes of three connecting lines：
In the method, in the process of the application,is the firstConvex hull inclination index of the first curve,is the firstThe slope of the line connecting the left and right end points of the first curve,is the slope of the connection line between the left end point and the upper convex point,is the slope of the connection line between the right end point and the upper convex point. Absolute value of slope between left and right end pointsThe larger the first curve is, the more vertical the whole is, the more inclined the convex hull curve is, and the convex hull inclination index isThe larger the slope of the connection line between the left and right end points on the convex hull curve and the upper convex pointThe larger the end point is, the more vertical the connecting line of the upper convex hull is, the more inclined the convex hull curve is, and the convex hull inclination index isThe larger.
In the first curve, the position distance between the left and right end points is obtained as the width of the first curve in the horizontal directionThe position distance between the left end point and the upper salient point is taken as a first height in the vertical direction of a first curveTaking the position distance between the right end point and the upper salient point as the second height in the vertical direction of the first curve. The closer the distance between the left and right end points is, the farther the distance between the left and right end points and the upper salient point is, the more the whole first curve is protruded and the sharper, thus constructing the convex hull sharpness index of the first curve：
In the method, in the process of the application,is the convex hull sharp index of the first curve,for a first height in the vertical direction of the first curve,a second height in the vertical direction of the first curve,is the width of the first curve in the horizontal direction. Distance between left and right end points of first curveThe smaller the width of the first curve is, and meanwhile, the distance between the left and right end points of the curve and the upper bump is、The larger the first curve, the higher the height of the first curve, the farther the upper convex point is from the two end points on the curve, the more the upper convex point is protruded, and the more the first curve is protruded and the sharper the convex hull sharp indexThe larger.
In the first curve, the slope of the line between the left and right end points and the upper bump reflects the inclination degree of the curve, and the distance between the end points and the upper bump reflects the sharpness degree of the curve. Building the steepness of the convex hull of the first curve：
In the method, in the process of the application,is the steepness of the convex hull of the first curve,is the firstConvex hull inclination index of the first curve,is the convex hull sharp index of the first curve. Convex hull inclination indexThe larger the first curve is, the more vertical the first curve is, the greater the degree of inclination, the steeper the first curve is in inclination angle; convex hull sharp indexThe larger the width of the first curve is, the smaller the height is, the higher the sharpness of the convex hull is, and the sharpness index of the convex hull isThe larger the first curve is, the steeper in height; in conclusion, the method comprises the steps of,the larger the convex hull the greater the steepness of the convex hull, the more likely the first curve is a convex hull of the smoke profile.
The environment in the high-rise building image is complex, the environment has the background of other high-rise buildings and the like, the background color is inconsistent and the difference is large, the image shot by the unmanned aerial vehicle is not a fixed scene, so that the shot image has the condition that the color and gray value of the background part and the target smoke part are similar, and if the threshold segmentation is simply adopted, the problem that the smoke cannot be distinguished from the background similar to the color is solved. In order to better detect high-rise smoke, the embodiment constructs the difference degree between smoke and background according to the characteristics of the smoke, and combines a CA visual saliency detection algorithm to detect a smoke area, specifically:
in high-rise building images, smoke is typically spread upward, forming a layer of well-contoured clouds that overlap one another. The clouds are also integrally upwardly diffused, so that a large number of the cloud profile curves have more convex hulls, and the non-smoke areas of the images, usually high-rise buildings, windows and roofs, are relatively regular, have no upward diffusion trend, and have almost no convex hull curves.
At this time, each suspected smoke convex hull curve in the high-rise building image has a convex hull steepness index, and the level of the CA saliency algorithm is regional, so that the image is divided intoThe number of sub-areas is equal to the number of sub-areas,the value of (2) can be set by the operator, and the present embodiment is not particularly limited. Each subarea is respectively marked as. Each region may have one or more convex hull curves, due to the division of the regions into convex hull curvesThe lines are obtained in different ways, and the convex hull curve in the region may be an incomplete convex hull curve. Counting the number of the first curves in each regionObtaining the number of pixel points on each first curveAnd obtaining the number of pixel points on the complete convex hull curve corresponding to each first curve. If the first image of high-rise buildingWithin each area, shareA first curve, wherein the first curve is in the regionThe first curves shareThe number of all the pixel points on the complete convex hull curve corresponding to the first curve isConstruction of the firstAbruptness of profile of individual sub-regions：
In the method, in the process of the application,is the firstThe steepness of the profile of the individual sub-regions,is the firstThe number of first curves in the individual zones,is the firstWithin the individual partitionThe number of pixels on the first curve,is the firstThe first curves correspond to the number of all pixel points on the complete convex hull curve,is the firstThe steepness of the convex hull of the complete convex hull curve corresponding to the first curve. The first curve in the region may be incomplete, with some incomplete first curve being proportional to the complete first curveThe larger the incomplete first curve should be given a larger steepness of the convex hull, so in the formula, the incomplete convex hull curve is the proportion of the complete convex hull, i.eAs an influencing factor, the larger the duty ratio is, and at the same time, the steepness of the convex hull of the curveThe larger theRegional profile abruptnessThe larger.
The more convex hull curves are provided in a region, the more complex the profile of each curve is, and the steeper the curve is, which means that the more irregular and symmetrical the profile in the region is, the higher the profile irregularity of the region is, and the more likely the region is a smoke region compared with other regions. Thus using the formula, a zoned profile irregularity is constructed：
In the method, in the process of the application,is the firstThe profile irregularities of the individual sub-areas,is the firstThe number of first curves in a partition,is the firstThe steepness of the profile of the individual sub-regions. The more the first curves, the greater the steepness of each convex hull curve in the region, and the steepness of the whole regionThe larger the convex hull curve, the more irregular and asymmetric the convex hull curve, the more irregular the contour of the subarea, and the irregular the contour of the subareaThe larger. The regional profile irregularityThe larger the relation between the contours of the subareas is, the more complex the relation is, the more irregular the contour of the convex hull curve in the subareas is, the more likely the subareas are smoke, and the higher the saliency of the subareas is when the smoke saliency of the image is detected.
And step S003, carrying out smoke detection on the high-rise building image according to the contour irregularity of each subarea and the CA saliency detection algorithm.
Because the traditional CA saliency algorithm is only suitable for the situations that the color difference between a target area and a background area is large and the background color is single, in the scene of the embodiment, the background area of an image is complex, the colors are not uniform, and more buildings and facilities with similar smoke colors exist. If the traditional CA algorithm is adopted, the problem that smoke and a background building cannot be distinguished exists, namely the error of the background building can be obviously enhanced, and the smoke and non-smoke areas cannot be distinguished accurately.
Therefore, according to the characteristic that the image smoke has upward diffusion trend, the smoke outline has a plurality of convex hulls, the convex hull outline is different from the background building outline, and the outline irregularity of each subarea is used for constructing the salient value of each subarea：
In the method, in the process of the application,is the firstThe significance of the individual sub-regions,is a high-rise buildingThe number of sub-areas in the image,、respectively the firstFirst, secondThe number of sub-areas is equal to the number of sub-areas,for the degree of difference between the two sub-regions acquired by the original CA algorithm,is the firstThe degree of irregularity of the profile of the individual sub-areas,to take the following measuresAn exponential function of the base. When dividing into areasThe greater the degree of irregularity of the profile, the more convex hulls the sub-region has, the more complex and irregular the profile within the sub-region, the more likely the sub-region is a smoke region, the more significant the sub-region needs to be enhanced, and the greater the significance of the sub-region.
After the CA algorithm significance detection, the significance value of each subarea is obtained. Setting a significance thresholdIt should be noted that, the present application,the value setting implementation of (a) can be set by the user, and the embodiment willThe value of (2) is set to 0.7. Judging the significance value of each sub-area, if the significance value of each sub-area is larger than a significance threshold value, indicating that smoke generated by a fire exists in the high-rise building image, wherein the sub-area with the significance value larger than the significance threshold value is a smoke area, and sending out an alarm through an alarm at the moment to inform a firefighter of extinguishing the fire; if the significance value of each sub-area is smaller than the significance threshold value, the fact that fire disasters do not exist in the high-rise building image is indicated that each sub-area is a non-smoke area, and the unmanned aerial vehicle continues to patrol other areas and does not give an alarm.
In summary, the embodiment of the application collects the image of the high-rise building through the unmanned aerial vehicle, detects the possible smoke in the image, and detects the smoke by adopting the improved CA significance detection algorithm, thereby avoiding the problem that the traditional CA algorithm cannot consider the problem that the smoke area in the image cannot be accurately detected due to the fact that the smoke is similar to the background color under the complex background, improving the accuracy of the smoke detection and having higher accuracy of monitoring the smoke of the high-rise building;
the embodiment provides a high-rise fire-fighting unmanned aerial vehicle monitoring method based on image processing, which is used for obtaining the steepness of convex hulls of a first curve according to the connecting line slope and the position distance between the left and right end points and the upper convex points of each convex hull curve in an edge image of a high-rise building image; obtaining the contour steepness of each sub-region according to the convex hull steepness of the first curve; obtaining the contour irregularity of each sub-region according to the contour steepness of each sub-region; obtaining the salient value of each subarea according to the outline irregularity of each subarea and a CA salient detection algorithm; and (3) obtaining the smoke areas in the high-rise building image according to the salient values of the areas, so that the smoke detection of the high-rise building is finished, and the accuracy of the CA salient detection algorithm on the smoke detection generated by the fire is improved.
It should be noted that: the sequence of the embodiments of the present application is only for description, and does not represent the advantages and disadvantages of the embodiments. And the foregoing description has been directed to specific embodiments of this specification. In addition, the processes depicted in the accompanying figures do not necessarily require the particular order shown, or sequential order, to achieve desirable results. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.
In this specification, each embodiment is described in a progressive manner, and the same or similar parts of each embodiment are referred to each other, and each embodiment mainly describes differences from other embodiments.
The above embodiments are only for illustrating the technical solution of the present application, and not for limiting the same; the technical solutions described in the foregoing embodiments are modified or some of the technical features are replaced equivalently, so that the essence of the corresponding technical solutions does not deviate from the scope of the technical solutions of the embodiments of the present application, and all the technical solutions are included in the protection scope of the present application.
Claims (10)
1. The method for monitoring the high-rise fire unmanned aerial vehicle based on image processing is characterized by comprising the following steps of:
collecting a high-rise building image and obtaining a corresponding edge image; acquiring a convex hull curve in an edge image by a chain code method;
obtaining each suspected smoke convex hull curve in the edge image according to the length of the convex hull curve, and marking the curve as a first curve; acquiring a left end point, a right end point and an upper salient point on a first curve; obtaining a convex hull inclination index of the first curve according to the slopes of the connecting lines among the left end point, the right end point and the upper convex points; obtaining a convex hull sharp index of the first curve according to the position distances among the left end point, the right end point and the upper convex point; obtaining the steepness of the convex hull of the first curve according to the convex hull inclination index and the convex hull sharpness index of the first curve; equally dividing the high-rise building image into a plurality of sub-areas; obtaining the contour steepness of each subarea according to the convex hull steepness of the first curve in each subarea; obtaining the contour irregularity of each sub-region according to the contour steepness of each sub-region;
obtaining a significant value of each sub-region according to the profile irregularity of each sub-region; and obtaining the smoke area in the high-rise building image according to the salient values of the subareas.
2. The method for monitoring the high-rise unmanned aerial vehicle based on image processing according to claim 1, wherein the method for obtaining each suspected smoke convex hull curve in the edge image according to the convex hull curve length is specifically as follows:
presetting a length threshold; taking the number of pixel points on each convex hull curve as the length of each convex hull curve; and taking the convex hull curve with the length larger than the length threshold value as a suspected smoke convex hull curve.
3. The method for monitoring the high-rise unmanned aerial vehicle based on image processing according to claim 1, wherein the steps of obtaining the left end point, the right end point and the upper salient point on the first curve are as follows:
taking the pixel point with the smallest abscissa on the first curve as a left endpoint; the pixel point with the largest abscissa is taken as the right endpoint; the pixel point with the largest ordinate is taken as the upper salient point.
4. The method for monitoring the high-rise unmanned aerial vehicle for fire control based on image processing according to claim 1, wherein the convex hull inclination index of the first curve is obtained according to the slopes of the connecting lines among the left end point, the right end point and the upper convex point, specifically comprising:
recording the slope of the connecting line of the left end point and the right end point as a first slope; marking the slope of the connecting line of the left end point and the upper salient point as a second slope; marking the slope of the connecting line of the right end point and the upper salient point as a third slope; calculating an average value of absolute values of the second slope and the third slope; taking the product of the absolute value of the first slope and the average value as a convex hull inclination index of the first curve.
5. The method for monitoring the high-rise unmanned aerial vehicle based on image processing according to claim 1, wherein the method for obtaining the convex hull sharpness index of the first curve according to the position distances among the left end point, the right end point and the upper convex point comprises the following steps:
taking the position distance between the left end point and the right end point as the width of the first curve; taking the position distance between the left end point and the upper salient point as the first height of a first curve; taking the position distance between the right end point and the upper salient point as the second height of the first curve; and calculating a sum value of the first height and the second height, and taking the ratio of the sum value to the width as a convex hull sharp index of the first curve.
6. The method for monitoring the high-rise unmanned aerial vehicle based on image processing according to claim 1, wherein the steepness of the convex hull of the first curve is: and taking the product of the convex hull inclination index and the convex hull sharpness index of the first curve as the convex hull steepness of the first curve.
7. The method for monitoring the high-rise unmanned aerial vehicle based on image processing according to claim 1, wherein the step of obtaining the contour steepness of each subarea according to the convex hull steepness of the first curve in each subarea comprises the following steps:
for each subarea, calculating the ratio of the number of pixel points on each first curve in the subarea to the number of all pixel points on the corresponding complete convex hull curve; calculating the product of the steepness of the convex hulls of the first curves and the ratio; calculating the sum of the products of all the first curves in the region; the sum is taken as the contour steepness of the sub-region.
8. The method for monitoring the high-rise unmanned aerial vehicle based on image processing according to claim 1, wherein the method for obtaining the profile irregularity of each subarea according to the profile steepness of each subarea comprises the following steps: for each subarea, taking the product of the first curve number and the contour steepness in the subarea as the contour irregularity of the subarea.
9. The method for monitoring the high-rise unmanned aerial vehicle based on image processing according to claim 1, wherein the method for obtaining the salient values of the subareas according to the profile irregularity of the subareas comprises the following steps:
for the firstIndividual sub-regions, obtain +.>An average value of the degree of difference between each of the sub-regions and each of the other sub-regions; calculate->Calculating the product of the contour irregularity of each sub-region and the average value, and calculating the reciprocal of an exponential function taking the product as an index and taking a natural constant as a base; taking the difference between 1 and the calculation result as +.>Significant values for individual sub-regions.
10. The method for monitoring the high-rise unmanned aerial vehicle based on image processing according to claim 1, wherein the method for obtaining the smoke area in the high-rise building image according to the significant value of each sub-area is specifically as follows: presetting a significant threshold; taking the subarea with the significance value larger than the significance threshold value as a smoke area; a sub-region with a significance value less than the significance threshold is taken as a non-smoke region.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202311346513.3A CN117079167B (en) | 2023-10-18 | 2023-10-18 | Image processing-based high-rise fire-fighting unmanned aerial vehicle monitoring method |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202311346513.3A CN117079167B (en) | 2023-10-18 | 2023-10-18 | Image processing-based high-rise fire-fighting unmanned aerial vehicle monitoring method |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN117079167A true CN117079167A (en) | 2023-11-17 |
| CN117079167B CN117079167B (en) | 2024-01-09 |
Family
ID=88715749
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202311346513.3A Active CN117079167B (en) | 2023-10-18 | 2023-10-18 | Image processing-based high-rise fire-fighting unmanned aerial vehicle monitoring method |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN117079167B (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN118053531A (en) * | 2024-04-16 | 2024-05-17 | 大连杰伍科技有限公司 | Intelligent management method and system for clinical data of medical examination |
Citations (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| KR20200113391A (en) * | 2019-03-25 | 2020-10-07 | (주)엠시오 | Autonomous driving drone for fire extinguish and method for fire suppression using the same |
| CN113989516A (en) * | 2021-10-19 | 2022-01-28 | 北京牧之科技有限公司 | Smoke dynamic identification method and related device |
| US20220206162A1 (en) * | 2020-12-30 | 2022-06-30 | Zoox, Inc. | Object contour determination |
| CN115797356A (en) * | 2023-02-09 | 2023-03-14 | 山东第一医科大学附属省立医院(山东省立医院) | Nuclear magnetic resonance tumor region extraction method |
| CN115937160A (en) * | 2022-12-16 | 2023-04-07 | 西北核技术研究所 | Explosion fireball contour detection method based on convex hull algorithm |
| CN116229359A (en) * | 2023-02-13 | 2023-06-06 | 杭电(丽水)研究院有限公司 | Smoke identification method based on improved classical optical flow method model |
| CN116259005A (en) * | 2023-02-04 | 2023-06-13 | 国网浙江省电力有限公司嘉兴供电公司 | Intelligent monitoring system based on roof photovoltaic fire control |
| CN116862916A (en) * | 2023-09-05 | 2023-10-10 | 常熟理工学院 | Production detection method and system based on image processing |
-
2023
- 2023-10-18 CN CN202311346513.3A patent/CN117079167B/en active Active
Patent Citations (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| KR20200113391A (en) * | 2019-03-25 | 2020-10-07 | (주)엠시오 | Autonomous driving drone for fire extinguish and method for fire suppression using the same |
| US20220206162A1 (en) * | 2020-12-30 | 2022-06-30 | Zoox, Inc. | Object contour determination |
| CN113989516A (en) * | 2021-10-19 | 2022-01-28 | 北京牧之科技有限公司 | Smoke dynamic identification method and related device |
| CN115937160A (en) * | 2022-12-16 | 2023-04-07 | 西北核技术研究所 | Explosion fireball contour detection method based on convex hull algorithm |
| CN116259005A (en) * | 2023-02-04 | 2023-06-13 | 国网浙江省电力有限公司嘉兴供电公司 | Intelligent monitoring system based on roof photovoltaic fire control |
| CN115797356A (en) * | 2023-02-09 | 2023-03-14 | 山东第一医科大学附属省立医院(山东省立医院) | Nuclear magnetic resonance tumor region extraction method |
| CN116229359A (en) * | 2023-02-13 | 2023-06-06 | 杭电(丽水)研究院有限公司 | Smoke identification method based on improved classical optical flow method model |
| CN116862916A (en) * | 2023-09-05 | 2023-10-10 | 常熟理工学院 | Production detection method and system based on image processing |
Non-Patent Citations (4)
| Title |
|---|
| XUAN ZHAA ET.AL.: ""Fire Smoke Detection Based on Contextual Object Detection"", 《2018 IEEE 3RD INTERNATIONAL CONFERENCE ON IMAGE, VISION AND COMPUTING (ICIVC)》 * |
| 史劲亭;袁非牛;夏雪;: "视频烟雾检测研究进展", 中国图象图形学报, vol. 23, no. 03, pages 303 - 319 * |
| 时佳琦: ""真实场景下黑烟车烟雾检测算法的研究"", 《中国优秀硕士学位论文全文数据库 工程科技Ⅱ辑》, vol. 2023, no. 02, pages 034 - 1607 * |
| 李小坤: ""面向自然场景的火焰烟雾视觉检测方法研究"", 《中国优秀硕士学位论文全文数据库 工程科技Ⅰ辑》, vol. 2022, no. 03, pages 026 - 38 * |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN118053531A (en) * | 2024-04-16 | 2024-05-17 | 大连杰伍科技有限公司 | Intelligent management method and system for clinical data of medical examination |
Also Published As
| Publication number | Publication date |
|---|---|
| CN117079167B (en) | 2024-01-09 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN117079167B (en) | Image processing-based high-rise fire-fighting unmanned aerial vehicle monitoring method | |
| CN109961455B (en) | Target detection method and device | |
| JP4705090B2 (en) | Smoke sensing device and method | |
| US7609856B2 (en) | Smoke detection method based on video processing | |
| US9047515B2 (en) | Method and system for wildfire detection using a visible range camera | |
| CN107437318B (en) | Visible light intelligent recognition algorithm | |
| CN107347151A (en) | binocular camera occlusion detection method and device | |
| US8159539B2 (en) | Smoke detecting method and system | |
| CN106991418B (en) | Winged insect detection method and device and terminal | |
| CN103514430B (en) | The method and apparatus of detection flame | |
| CN109360370B (en) | Robot-based smoke and fire detection method | |
| CN111461013B (en) | Unmanned aerial vehicle-based real-time fire scene situation awareness method | |
| CN115115595B (en) | Real-time calibration method of airborne laser radar and infrared camera for forest fire monitoring | |
| CN114120171A (en) | Fire smoke detection method, device and equipment based on video frame and storage medium | |
| CN105139429A (en) | Fire detecting method based on flame salient picture and spatial pyramid histogram | |
| CN106815567B (en) | Flame detection method and device based on video | |
| WO2022091577A1 (en) | Information processing device and information processing method | |
| CN111144465A (en) | Multi-scene-oriented smoke detection algorithm and electronic equipment applying same | |
| CN112949536B (en) | Fire alarm method based on cloud platform | |
| CN117789394A (en) | Early fire smoke detection method based on motion history image | |
| CN113283332A (en) | Fire-fighting robot flame identification method, equipment and storage medium | |
| CN112668389A (en) | High-altitude parabolic target detection method, device, system and storage medium | |
| CN117079212A (en) | Smoke detection method and device, electronic equipment and storage medium | |
| CN114549978A (en) | Mobile robot operation method and system based on multiple cameras | |
| CN113870513A (en) | Smoke and fire identification and early warning method based on GIS |
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 | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |