CN115841731B - Infrared monitoring park fire early warning method - Google Patents
Infrared monitoring park fire early warning method Download PDFInfo
- Publication number
- CN115841731B CN115841731B CN202310123403.4A CN202310123403A CN115841731B CN 115841731 B CN115841731 B CN 115841731B CN 202310123403 A CN202310123403 A CN 202310123403A CN 115841731 B CN115841731 B CN 115841731B
- Authority
- CN
- China
- Prior art keywords
- flame
- infrared
- image
- video image
- images
- 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.)
- Active
Links
Images
Classifications
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02A—TECHNOLOGIES FOR ADAPTATION TO CLIMATE CHANGE
- Y02A40/00—Adaptation technologies in agriculture, forestry, livestock or agroalimentary production
- Y02A40/10—Adaptation technologies in agriculture, forestry, livestock or agroalimentary production in agriculture
- Y02A40/28—Adaptation technologies in agriculture, forestry, livestock or agroalimentary production in agriculture specially adapted for farming
Abstract
The invention provides an infrared monitoring park fire early warning method, which comprises the steps of collecting standard infrared images with normal working conditions in a park based on a thermal infrared imager. And acquiring an actual infrared image of the park site working process based on the thermal infrared imager, and comparing the actual infrared image with the standard infrared image. And extracting an abnormal infrared image according to the comparison result. And acquiring continuous frame video images based on the abnormal infrared images, and extracting flame shape features in different video images. And executing corresponding early warning processing according to the flame shape characteristics and the comprehensive change of the flame shape characteristics. The method for early warning the fire disaster in the park can find out some fire hazards in the working process of the park in time, meanwhile, the influence of some interference targets on the judgment result is eliminated, and the accuracy of early warning the fire disaster is further improved.
Description
Technical Field
The invention relates to the technical field of park fire early warning, in particular to an infrared monitoring park fire early warning method.
Background
At present, the campus fire prevention mainly is patrol through the campus staff, patrol through the campus staff to whole garden and discover some sources of fire, prevent the emergence of conflagration. But patrol through park staff and prevent the conflagration, on the one hand can waste a large amount of time of park staff, on the other hand if park staff is carelessly, do not discover some sources of fire in time, still can lead to the emergence of some conflagration, can't in time carry out early warning treatment to the conflagration.
Therefore, the invention provides an infrared monitoring park fire early warning method.
Disclosure of Invention
The invention provides an infrared monitoring park fire early warning method, which comprises the steps of comparing an infrared image obtained in the field working process with a standard infrared image under a normal working condition, extracting an abnormal infrared image, extracting flame shape characteristics in the abnormal infrared image, determining the occurrence of a fire according to the comprehensive change of the flame shape characteristics, and carrying out corresponding early warning treatment.
The invention provides an infrared detection park fire early warning method, which comprises the following steps:
step 1: collecting standard infrared images with normal working conditions in a park based on a thermal infrared imager;
step 2: acquiring an actual infrared image of the park site working process based on the thermal infrared imager, and comparing the actual infrared image with the standard infrared image;
step 3: extracting an abnormal infrared image according to the comparison result;
step 4: collecting continuous frame video images based on the abnormal infrared images, and extracting flame shape features in different video images;
step 5: and executing corresponding early warning processing according to the flame shape characteristics and the comprehensive change of the flame shape characteristics.
Preferably, based on the thermal infrared imager gathers the normal standard infrared image of all operating modes in the garden, include:
counting to obtain all working condition conditions in the working process in the park, collecting infrared images of each working condition based on the working process of the thermal infrared imager, and placing the collected infrared images of each working condition in the same corresponding infrared image set to obtain a plurality of sets for storing the infrared images;
judging the working condition corresponding to each set of stored infrared images, and if no fire disaster occurs in the working process of the working condition, making the working condition normal;
selecting all infrared image sets with normal working conditions, taking all standard infrared images in the infrared image sets with normal working conditions as first infrared images, and storing the first infrared images in a memory.
Preferably, the method for acquiring the actual infrared image of the park site working process based on the thermal infrared imager and comparing the actual infrared image with the standard infrared image comprises the following steps:
thermal imaging data acquisition is carried out on the working process of the park site based on the thermal infrared imager, and acquisition is carried out to obtain
As a second infrared image;
reducing the first infrared image and the second infrared image to a preset size and performing gray level conversion;
comparing the gray values of two adjacent pixels in each row in a first gray image, marking 1 pixel with a large gray value, and marking 0 pixel with a small gray value to obtain a first fingerprint of the first gray image;
simultaneously, a second fingerprint of the second gray level image is obtained;
comparing the first fingerprint with the second fingerprint, and counting inconsistent positions;
if the count is less than 4, the second infrared image is considered to be a normal infrared image;
if the count is greater than 8, the second infrared image is considered an anomalous infrared image.
Preferably, before extracting the flame shape feature in the different video images, the method further comprises: the flame identification model is constructed, and specifically comprises the following steps:
taking a first infrared image with normal working conditions as a first training sample, and constructing a first training sample set;
taking an infrared image of the working process of the fire disaster as a second training sample, and constructing a second training sample set;
extracting a second training sample without flame from the second training sample set, and constructing a third training sample set for the remaining second training samples with flame;
adding the extracted second training sample without flame into the first training sample set to obtain a fourth training sample set;
and training the third training sample set and the fourth training sample set by using the dert pre-training model to obtain a flame identification model.
Preferably, capturing successive frames of video images based on the abnormal infrared image, and extracting flame shape features in different video images, including:
extracting abnormal infrared images and arranging the abnormal infrared images according to the time sequence of the field working process to obtain
Successive frames of video images of the abnormal infrared image as a first video image;
filtering the obtained first video image to obtain a filtered first video image;
inputting the filtered first video image into a flame identification model, and determining whether each frame of video image has flame;
removing the flameless generated image in the first video image to obtain a second video image;
performing binarization processing on each frame of flame image in the second video image, and determining an initial boundary line in the binarization processing image;
analyzing the initial boundary line to obtain a final boundary line;
dividing the corresponding binarization processing image according to the final boundary line to obtain a target flame area;
and extracting index parameters related to preset indexes from each target flame area, and obtaining flame characteristics of the corresponding target flame area.
Preferably, inputting the filtered first video image into a flame identification model to determine whether each frame of video image has flames generated, including:
decomposing the filtered first video image into a plurality of abnormal infrared images of continuous frames, and respectively inputting each frame of abnormal infrared image into a constructed flame identification model for identification;
respectively calculating the similarity between each frame of abnormal infrared image and each flame category in the flame identification model, and taking the category of the maximum similarity as the category of the abnormal infrared image of the corresponding frame;
and determining whether flame is generated in each frame of video image according to the category of each frame of abnormal infrared image in the first video image.
Preferably, the analyzing the initial boundary line to obtain a final boundary line includes:
calibrating all the obtained initial boundary lines, and sequentially arranging all the calibrated initial boundary lines according to the profile trend;
respectively acquiring a first endpoint and a second endpoint of each initial boundary line, and determining a first tangent line of each first endpoint and a second tangent line of each second endpoint;
sequentially acquiring a first straight line of a second end point of a first boundary line in adjacent initial boundary lines and a first end point of a second boundary line in the adjacent initial boundary lines;
when the first straight line is smaller than the preset length, sequentially acquiring a second tangent line of a second endpoint of a first boundary line in the adjacent initial boundary lines and a first tangent line of a first endpoint of the second boundary line in the adjacent initial boundary lines, and determining whether an intersection point exists between the two tangent lines;
if so, judging whether the first distance from the midpoint of the first straight line to the intersection point is smaller than a preset distance, if so, reserving the intersection point, and establishing connection with two end points based on the intersection point;
if not, according to a first ratio of the first distance to the preset distance, carrying out first interception on the distance from the midpoint to the intersection point, reserving an interception point, and establishing connection with two endpoints based on the interception point;
when the first straight line is not smaller than the preset length, drawing a circle by taking half of the length of the first straight line as a radius and taking the middle point of the first straight line as a circle center to obtain a first circle;
meanwhile, judging whether a first point exists between the first tangent line and the second tangent line and the first circle or not;
if the first points exist, connecting the second end point, the two first points and the first end point in sequence according to the appearance position sequence of the first points;
if only one first point exists, acquiring an arc corresponding to the first point, and taking the middle point of the arc as a connecting point to realize sequential connection of the second endpoint, the connecting point and the first endpoint;
if the first point does not exist, determining a cutting direction of the first circle according to the intersection point direction of the first tangent line and the second tangent line, cutting the corresponding semicircle in the cutting direction in parallel with the diameter, acquiring two cutting points, and connecting the two cutting points and the first end point in sequence;
and constructing and obtaining a final boundary line based on all the connection results.
Preferably, extracting an index parameter related to a preset index from each target flame region, and obtaining a flame characteristic of the corresponding target flame region includes:
the area growth rate of the target flame area of the two continuous frames of video images is calculated according to the following formula:
wherein G is the area growth rate of the target flame region, S i+1 Is the area of the target flame region in the (i+1) th frame video image, S i Is the area of the target flame region in the ith frame of video image, T i+1 Is the time corresponding to the i+1st frame video image, T i Is the time corresponding to the video image of the ith frame;
and calculating flame characteristics of a target flame area of each frame of video image according to the following formula:
wherein, the liquid crystal display device comprises a liquid crystal display device,
representing a flame cluster characteristic value of a j1 st flame cluster in the video image of the corresponding frame; />
Representing the boundary perimeter of the j1 st flame cluster in the video image of the corresponding frame; />
Representing the area of the j1 st flame cluster in the video image of the corresponding frame; />
Representing a reference flame cluster characteristic value; />
Representing a color characteristic value of a j1 flame cluster in a video image of a corresponding frame; />
Representing a color pixel block contained in a j1 flame cluster in a video image of a corresponding frame; />
A color value representing a g1 st color pixel block in a j1 st flame cluster; />
Representing a reference color value; />
Representing a color ratio based on the color pixel block; />
Representing a minimum color ratio based on the color pixel block; />
Representing a maximum color ratio based on the color pixel block; m1 represents the number of flame clusters; />
Representing a set of flame cluster-based eigenvalues; />
Representing a set based on color feature values;
and obtaining the flame characteristics of the target flame region based on the set of flame cluster characteristic values and the set of color characteristic values.
Preferably, the executing corresponding early warning processing according to the flame shape feature and the comprehensive change of the flame shape feature includes:
recording the characteristic change of the target flame area in all the two continuous frames of video images;
and according to the recorded result, matching the early warning mode from the characteristic-early warning database, and carrying out early warning processing.
Additional features and advantages of the invention will be set forth in the description which follows, and in part will be obvious from the description, or may be learned by practice of the invention. The objectives and other advantages of the invention will be realized and attained by the structure particularly pointed out in the written description and claims thereof as well as the appended drawings.
The technical scheme of the invention is further described in detail through the drawings and the embodiments.
Drawings
The accompanying drawings are included to provide a further understanding of the invention and are incorporated in and constitute a part of this specification, illustrate the invention and together with the embodiments of the invention, serve to explain the invention. In the drawings:
fig. 1 is a flowchart of a method for early warning of a park fire by infrared monitoring according to an embodiment of the invention.
Detailed Description
The preferred embodiments of the present invention will be described below with reference to the accompanying drawings, it being understood that the preferred embodiments described herein are for illustration and explanation of the present invention only, and are not intended to limit the present invention.
The invention provides an infrared monitoring park fire early warning method, as shown in figure 1, comprising the following steps:
step 1: collecting standard infrared images with normal working conditions in a park based on a thermal infrared imager;
step 2: acquiring an actual infrared image of the park site working process based on the thermal infrared imager, and comparing the actual infrared image with the standard infrared image;
step 3: extracting an abnormal infrared image according to the comparison result;
step 4: collecting continuous frame video images based on the abnormal infrared images, and extracting flame shape features in different video images;
step 5: and executing corresponding early warning processing according to the flame shape characteristics and the comprehensive change of the flame shape characteristics.
In this embodiment, the working condition normally refers to a working process without fire occurrence, and all infrared images of the working process without fire occurrence are acquired and used as standard infrared images.
In the embodiment, a difference hash algorithm is adopted to perform similarity processing on the infrared image in the field working process and the standard infrared image, so as to find out an abnormal infrared image different from the standard infrared image.
In this embodiment, all the abnormal infrared images are extracted, and the first video images of successive frames are obtained in a time-series arrangement.
In this embodiment, the flame shape features in the different video images include the area growth rate of the target flame region, flame cluster feature values, flame cluster color feature values.
In this embodiment, the area growth rate of the target flame region is unchanged or becomes larger, the area of the target flame region is becoming larger, the flame cluster characteristic value indicates the degree of regularity of each flame cluster, the smaller the flame cluster characteristic value is, the more irregular the shape of the flame cluster is, the larger the flame cluster color characteristic value is, and the higher the temperature of the flame cluster is. And judging whether fire disaster occurs or not according to the comprehensive change of the three flame shape characteristics of the continuous video images.
In this embodiment, the changes of the three flame features of each frame of video image in the second video image are recorded, and the recorded changes of the three flame features are matched with a feature-early warning database to perform corresponding early warning processing.
The beneficial effects of the technical scheme are as follows: the infrared image obtained in the field working process is compared with the standard infrared image under the normal working condition, the abnormal infrared image is extracted, the flame shape characteristics in the abnormal infrared image are extracted, the occurrence of fire is determined according to the comprehensive change of the flame shape characteristics, and corresponding early warning treatment is carried out.
The invention provides an infrared monitoring park fire early warning method, which is based on that a thermal infrared imager collects standard infrared images with normal working conditions in a park, and comprises the following steps:
counting to obtain all working condition conditions in the working process in the park, collecting infrared images of each working condition based on the working process of the thermal infrared imager, and placing the collected infrared images of each working condition in the same corresponding infrared image set to obtain a plurality of sets for storing the infrared images;
judging the working condition corresponding to each set of stored infrared images, and if no fire disaster occurs in the working process of the working condition, making the working condition normal;
selecting all infrared image sets with normal working conditions, taking all standard infrared images in the infrared image sets with normal working conditions as first infrared images, and storing the first infrared images in a memory.
In this embodiment, all the working conditions include the normal working condition and the working condition where the fire disaster occurs, the infrared images of all the working conditions are put in different sets, and finally the infrared image set with the normal working condition is selected as the standard image.
In the embodiment, the acquired standard infrared image is used as a first infrared image and stored in the memory, so that the standard infrared image is conveniently compared with the infrared image acquired in the field working process.
The beneficial effects of the technical scheme are as follows: by collecting standard infrared images with normal working conditions, a foundation is provided for comparison between the infrared images in the later and field working processes.
The invention provides an infrared monitoring park fire early warning method, which is based on that an actual infrared image of a park site working process is acquired by an infrared thermal imager and compared with the standard infrared image, and comprises the following steps:
thermal imaging data acquisition is carried out on the working process of the park site based on the thermal infrared imager, and acquisition is carried out to obtain
As a second infrared image;
reducing the first infrared image and the second infrared image to a preset size and performing gray level conversion;
comparing the gray values of two adjacent pixels in each row in a first gray image, marking 1 pixel with a large gray value, and marking 0 pixel with a small gray value to obtain a first fingerprint of the first gray image;
simultaneously, a second fingerprint of the second gray level image is obtained;
comparing the first fingerprint with the second fingerprint, and counting inconsistent positions;
if the count is less than 4, the second infrared image is considered to be a normal infrared image;
if the count is greater than 8, the second infrared image is considered an anomalous infrared image.
In the embodiment, an infrared image acquired in the field working process is used as a second infrared image, and is compared with a first infrared image with normal working conditions.
In this embodiment, the first infrared image and the second infrared image are both reduced to 9*8, 72 pixels in total, and the two infrared images are converted into 64-level gray scale images.
In this embodiment, the first infrared image is subjected to gray level conversion to obtain a second gray level image, and the second infrared image is subjected to gray level conversion to obtain the second gray level image.
In this embodiment, there are 9 pixels in each line of each gray image, so after comparing the gray values of every two adjacent pixels and marking, 8 marks are obtained in each line, which are both 0,1. Eight lines of each gray level image are provided, and finally, 64 marking values are obtained for each gray level image.
In this embodiment, the fingerprint is unique, corresponding to the identity of the gray scale image. Each first gray level image has only one first fingerprint and each second gray level image has only one second fingerprint.
In this embodiment, when the first gray scale image and the second gray scale image are compared, it is seen how many marks are different from 64 marks in the first fingerprint and the second fingerprint corresponding to the first gray scale image and the second gray scale image. If the number of different marks is smaller than 4, the second infrared image is considered to be a normal infrared image, and if the number of different marks is larger than 8, the second infrared image is considered to be an abnormal infrared image, and the abnormal infrared image is extracted.
The beneficial effects of the technical scheme are as follows: and carrying out gray level conversion on the first infrared image and the second infrared image, determining whether the second infrared image is an abnormal infrared image or not by calculating the number of different marks in the first fingerprint and the second fingerprint, and extracting the abnormal infrared image.
The invention provides an infrared monitoring park fire early warning method, which comprises the following steps before extracting flame shape characteristics in different video images: the flame identification model is constructed, and specifically comprises the following steps:
taking a first infrared image with normal working conditions as a first training sample, and constructing a first training sample set;
taking an infrared image of the working process of the fire disaster as a second training sample, and constructing a second training sample set;
extracting a second training sample without flame from the second training sample set, and constructing a third training sample set for the remaining second training samples with flame;
adding the extracted second training sample without flame into the first training sample set to obtain a fourth training sample set;
and training the third training sample set and the fourth training sample set by using the dert pre-training model to obtain a flame identification model.
In the embodiment, the infrared images with normal working conditions are used as a training sample, a first training sample set is constructed, the infrared images obtained in the working process of fire disaster are used as a training sample, and a second training sample set is constructed.
In this embodiment, not all training samples in the second training sample set have flames, and the training samples without flames in the second training sample set are added to the first training sample set to obtain a third training sample set and a fourth training sample set respectively.
In this embodiment, the dert pre-training model trains training samples in the third training sample set and the fourth training sample set to obtain a flame recognition model.
The beneficial effects of the technical scheme are as follows: and constructing a flame identification model to identify each frame of video image in the first video image and judge whether flame is generated in each frame of video image.
The invention provides an infrared monitoring park fire early warning method, which is used for collecting continuous frame video images based on abnormal infrared images and extracting flame shape characteristics in different video images, and comprises the following steps:
extracting abnormal infrared images and arranging the abnormal infrared images according to the time sequence of the field working process to obtain
Successive frames of video images of the abnormal infrared image as a first video image;
filtering the obtained first video image to obtain a filtered first video image;
inputting the filtered first video image into a flame identification model, and determining whether each frame of video image has flame;
removing the flameless generated image in the first video image to obtain a second video image;
performing binarization processing on each frame of flame image in the second video image, and determining an initial boundary line in the binarization processing image;
analyzing the initial boundary line to obtain a final boundary line;
dividing the corresponding binarization processing image according to the final boundary line to obtain a target flame area;
and extracting index parameters related to preset indexes from each target flame area, and obtaining flame characteristics of the corresponding target flame area.
In this embodiment, the first video image is composed of temporally successive abnormal infrared images.
In this embodiment, the filtering is to perform noise reduction processing on the original image, so that most of noise is removed under the condition that main features of the original image are not affected, the subsequent image quality is determined, and the complexity of image binarization processing is greatly reduced.
In this embodiment, each frame of video image in the filtered first video image is input into the flame recognition model, and the video image without flame is removed, so as to obtain the second video image.
In this embodiment, the binarization process is to set the gray value of the pixel point on each frame of the flame image to 0 or 255, so that each frame of the flame image exhibits obvious visual effects of only black and white. Wherein the target flame area is white and the background area is black.
In this embodiment, the final boundary line is used to segment the target flame region and the background region.
In this embodiment, the target flame region is obtained by dividing the binarized image, and then the relevant index parameters are extracted to obtain the flame characteristics of the target flame region.
The beneficial effects of the technical scheme are as follows: and carrying out binarization processing and image segmentation on each frame of video image in the second video image to obtain a target flame area of each frame of video image, and extracting to obtain flame characteristics of the target flame area of each frame of video image.
The invention provides an infrared monitoring park fire early warning method, which inputs a filtered first video image into a flame identification model to determine whether each frame of video image has flame generation or not, and comprises the following steps:
decomposing the filtered first video image into a plurality of abnormal infrared images of continuous frames, and respectively inputting each frame of abnormal infrared image into a constructed flame identification model for identification;
respectively calculating the similarity between each frame of abnormal infrared image and each flame category in the flame identification model, and taking the category of the maximum similarity as the category of the abnormal infrared image of the corresponding frame;
and determining whether flame is generated in each frame of video image according to the category of each frame of abnormal infrared image in the first video image.
In this embodiment, to input each frame of the first video image into the flame recognition model for recognition, the first video image needs to be decomposed into successive abnormal infrared images.
In the embodiment, the flame type corresponding to each frame of video image is determined by calculating the similarity between each frame of video image and each flame type in the flame identification model, and the flame type with the largest similarity is the type of the video image, and whether flame is generated is determined according to the type of each frame of video image.
The beneficial effects of the technical scheme are as follows: and determining whether each frame of video image in the first video image has flame generation or not, and obtaining all video images generated by the flame when a fire disaster occurs, thereby constructing a second video image.
The invention provides an infrared monitoring park fire early warning method, which is used for analyzing an initial boundary line to obtain a final boundary line and comprises the following steps:
calibrating all the obtained initial boundary lines, and sequentially arranging all the calibrated initial boundary lines according to the profile trend;
respectively acquiring a first endpoint and a second endpoint of each initial boundary line, and determining a first tangent line of each first endpoint and a second tangent line of each second endpoint;
sequentially acquiring a first straight line of a second end point of a first boundary line in adjacent initial boundary lines and a first end point of a second boundary line in the adjacent initial boundary lines;
when the first straight line is smaller than the preset length, sequentially acquiring a second tangent line of a second endpoint of a first boundary line in the adjacent initial boundary lines and a first tangent line of a first endpoint of the second boundary line in the adjacent initial boundary lines, and determining whether an intersection point exists between the two tangent lines;
if so, judging whether the first distance from the midpoint of the first straight line to the intersection point is smaller than a preset distance, if so, reserving the intersection point, and establishing connection with two end points based on the intersection point;
if not, according to a first ratio of the first distance to the preset distance, carrying out first interception on the distance from the midpoint to the intersection point, reserving an interception point, and establishing connection with two endpoints based on the interception point;
when the first straight line is not smaller than the preset length, drawing a circle by taking half of the length of the first straight line as a radius and taking the middle point of the first straight line as a circle center to obtain a first circle;
meanwhile, judging whether a first point exists between the first tangent line and the second tangent line and the first circle or not;
if the first points exist, connecting the second end point, the two first points and the first end point in sequence according to the appearance position sequence of the first points;
if only one first point exists, acquiring an arc corresponding to the first point, and taking the middle point of the arc as a connecting point to realize sequential connection of the second endpoint, the connecting point and the first endpoint;
if the first point does not exist, determining a cutting direction of the first circle according to the intersection point direction of the first tangent line and the second tangent line, cutting the corresponding semicircle in the cutting direction in parallel with the diameter, acquiring two cutting points, and connecting the two cutting points and the first end point in sequence;
and constructing and obtaining a final boundary line based on all the connection results.
In this embodiment, the intercepting direction is consistent with the intersecting point direction of the first tangent line and the second tangent line, if there is no intersecting point, random straight line interception of the upper semicircle or the lower semicircle is performed, and the obtained two points are taken as intercepting points.
In this embodiment, each initial boundary line is calibrated and ordered in a counter-clockwise direction starting from the uppermost one.
In this embodiment, the leftmost end point of each initial boundary line is a first end point, the rightmost end point is a second end point, and a tangent line of the first end point is obtained as a first tangent line, and a tangent line of the second end point is obtained as a second tangent line.
In this embodiment, two adjacent initial boundary lines are acquired. The left initial boundary line is a first boundary line, the right initial boundary line is a second boundary line, and the second end point of the first boundary line and the first end point of the second boundary line are connected by a straight line.
In this embodiment, the preset length is set in advance and set to 2mm.
In this embodiment, when the first straight line length between the second end point and the first end point of the adjacent initial boundary line is smaller than 2mm, the first tangent line and the second tangent line of the adjacent initial boundary line are obtained, and whether there is an intersection point between the first tangent line and the second tangent line is determined.
In this embodiment, the preset distance is a comparison value set in advance, and is set to 1mm.
In this embodiment, if there is an intersection point between two tangent lines in the adjacent initial boundary lines, it is determined whether the distance from the first straight line in the adjacent initial boundary lines to the intersection point of the two tangent lines is smaller than a preset distance. If the distance is smaller than the preset distance, reserving the intersection point of the two tangent lines, and connecting the intersection point with the two end points.
In this embodiment, if the distance from the first straight line to the intersection point of the two tangent lines in the adjacent initial boundary lines is greater than a preset distance, a first ratio of the first distance to the preset distance is obtained, the distance from the midpoint of the first straight line to the intersection point of the two tangent lines is intercepted according to the magnitude of the first ratio, and the obtained interception point is connected with the two end points.
In this embodiment, the greater the first ratio, the greater the truncated distance.
In this embodiment, when the length of the first straight line between the second end point and the first end point of the adjacent initial boundary line is greater than 2mm, a circle is drawn by using the midpoint of the first straight line as the center of a circle and the distance from the midpoint to any one of the end points as the radius, thereby obtaining a first circle.
In this embodiment, it is determined whether the obtained tangent line of the first circle and the two end points has the first point.
In this embodiment, the first point refers to the intersection of two tangent lines with the first circle.
In this embodiment, if both tangent lines have a first point with the first circle, the second end point, the two first points, and the first end point are connected in sequence in the order from left to right.
In this embodiment, if there is only a first point between the two tangents and the first circle, it is determined which point is the intersection of the first circle and which tangent. If the first point is the intersection point of the first circle and the first tangent line, acquiring an arc between the first end point and the first point, and obtaining the middle point of the arc. And taking the middle point of the arc as a connecting point, connecting the second end point and the connecting point, and sequentially connecting the first end points. If the first point is the intersection point of the first circle and the second tangent line, acquiring the arc between the second end point and the first point, and obtaining the middle point of the arc. And taking the middle point of the arc as a connecting point, connecting the second end point and the connecting point, and sequentially connecting the first end points.
In this embodiment, if the two tangent lines have no first point with the first circle, it is determined that the intersection point of the two tangent lines intercepts the first circle with respect to the position of the first circle. And cutting a half of the first circle by the intersection point of the two tangent lines against the direction of the first circle, and obtaining two cutting points. And connecting the second end point, the two interception points and the first end point in sequence.
In this embodiment, the final boundary line is obtained based on the connection result of every two adjacent initial boundary lines.
The beneficial effects of the technical scheme are as follows: and obtaining a connection result according to every two adjacent initial boundary lines, and finally obtaining a final boundary line based on all connection results, wherein the obtained final boundary line is more accurate, and the obtained target flame area is more complete.
The invention provides an infrared monitoring park fire early warning method, which extracts index parameters related to preset indexes from each target flame area and obtains flame characteristics of the corresponding target flame area, and comprises the following steps:
the area growth rate of the target flame area of the two continuous frames of video images is calculated according to the following formula:
wherein G is the area growth rate of the target flame region, S i+1 Is the area of the target flame region in the (i+1) th frame video image, S i Is the area of the target flame region in the ith frame of video image, T i+1 Is the time corresponding to the i+1st frame video image, T i Is the time corresponding to the video image of the ith frame;
and calculating flame characteristics of a target flame area of each frame of video image according to the following formula:
wherein, the liquid crystal display device comprises a liquid crystal display device,
representing a flame cluster characteristic value of a j1 st flame cluster in the video image of the corresponding frame; />
Representing corresponding frame videoThe perimeter of the boundary of the j1 st flame cluster in the image; />
Representing the area of the j1 st flame cluster in the video image of the corresponding frame; />
Representing a reference flame cluster characteristic value; />
Representing a color characteristic value of a j1 flame cluster in a video image of a corresponding frame; />
Representing a color pixel block contained in a j1 flame cluster in a video image of a corresponding frame; />
A color value representing a g1 st color pixel block in a j1 st flame cluster; />
Representing a reference color value; />
Representing a color ratio based on the color pixel block; />
Representing a minimum color ratio based on the color pixel block; />
Representing a maximum color ratio based on the color pixel block; m1 represents the number of flame clusters; />
Representing a set of flame cluster-based eigenvalues; />
Representing a set based on color feature values;
and obtaining the flame characteristics of the target flame region based on the set of flame cluster characteristic values and the set of color characteristic values.
In the embodiment, the area growth rate of the target flame area in two continuous frames of video images, the flame cluster characteristic value and the flame cluster color characteristic value are taken as the flame characteristics of the target flame area.
In this embodiment, the target flame area in each frame of video image has a plurality of flame clusters, each flame cluster corresponding to a color feature value.
In this embodiment, one set of flame cluster feature values includes all flame cluster feature values for a target flame region in a frame of video image, and one set of color feature values includes all flame cluster feature values for the target flame region in a frame of video image.
The beneficial effects of the technical scheme are as follows: by observing the comprehensive change of the extracted flame shape characteristics, whether fire disaster occurs in two continuous frames of video images or not can be determined, and whether fire disaster occurs in the whole field working process or not can be further determined.
The invention provides an infrared monitoring park fire early warning method, which executes corresponding early warning processing according to the flame shape characteristics and the comprehensive change of the flame shape characteristics, and comprises the following steps:
recording the characteristic change of the target flame area in all the two continuous frames of video images;
and according to the recorded result, matching the early warning mode from the characteristic-early warning database, and carrying out early warning processing.
In this embodiment, the feature-early warning database stores different corresponding early warning modes under different changes of flame shape features.
In the embodiment, two continuous frames of video images simultaneously meet the requirement that the area of a target flame area is enlarged, the flame cluster characteristic value in the second frame of video image corresponding flame cluster characteristic value set is larger than that of the first frame of video image, and the color characteristic value in the second frame of video image color characteristic value set is larger than that of the first frame of video image, and the second frame of video image is matched with the fire early warning in the characteristic-early warning database to carry out fire early warning. Otherwise, the fire disaster early warning device is possibly influenced by interference targets such as incandescent lamps, candles and the like, and does not perform fire disaster early warning.
The beneficial effects of the technical scheme are as follows: according to the comprehensive change of the flame shape characteristics in the continuous video images, whether fire disaster occurs or not is judged, corresponding early warning is carried out, the influence of some interference targets on the judgment result can be eliminated, and the early warning accuracy is improved.
It will be apparent to those skilled in the art that various modifications and variations can be made to the present invention without departing from the spirit or scope of the invention. Thus, it is intended that the present invention also include such modifications and alterations insofar as they come within the scope of the appended claims or the equivalents thereof.
Claims (6)
1. An infrared monitoring park fire early warning method is characterized by comprising the following steps:
step 1: collecting standard infrared images with normal working conditions in a park based on a thermal infrared imager;
step 2: acquiring an actual infrared image of the park site working process based on the thermal infrared imager, and comparing the actual infrared image with the standard infrared image;
step 3: extracting an abnormal infrared image according to the comparison result;
step 4: collecting continuous frame video images based on the abnormal infrared images, and extracting flame shape features in different video images;
step 5: executing corresponding early warning processing according to the flame shape characteristics and the comprehensive change of the flame shape characteristics;
the method for extracting flame shape features in different video images comprises the following steps of:
extracting abnormal infrared images and arranging the abnormal infrared images according to the time sequence of the field working process to obtain
Successive frames of video images of the abnormal infrared image as a first video image;
filtering the obtained first video image to obtain a filtered first video image;
inputting the filtered first video image into a flame identification model, and determining whether each frame of video image has flame;
removing the flameless generated image in the first video image to obtain a second video image;
performing binarization processing on each frame of flame image in the second video image, and determining an initial boundary line in the binarization processing image;
analyzing the initial boundary line to obtain a final boundary line;
dividing the corresponding binarization processing image according to the final boundary line to obtain a target flame area;
extracting index parameters related to preset indexes from each target flame area, and obtaining flame characteristics of the corresponding target flame area;
analyzing the initial boundary line to obtain a final boundary line, including:
calibrating all the obtained initial boundary lines, and sequentially arranging all the calibrated initial boundary lines according to the profile trend;
respectively acquiring a first endpoint and a second endpoint of each initial boundary line, and determining a first tangent line of each first endpoint and a second tangent line of each second endpoint;
sequentially acquiring a first straight line of a second end point of a first boundary line in adjacent initial boundary lines and a first end point of a second boundary line in the adjacent initial boundary lines;
when the first straight line is smaller than the preset length, sequentially acquiring a second tangent line of a second endpoint of a first boundary line in the adjacent initial boundary lines and a first tangent line of a first endpoint of the second boundary line in the adjacent initial boundary lines, and determining whether an intersection point exists between the two tangent lines;
if so, judging whether the first distance from the midpoint of the first straight line to the intersection point is smaller than a preset distance, if so, reserving the intersection point, and establishing connection with two end points based on the intersection point;
if not, according to a first ratio of the first distance to the preset distance, carrying out first interception on the distance from the midpoint to the intersection point, reserving an interception point, and establishing connection with two endpoints based on the interception point;
when the first straight line is not smaller than the preset length, drawing a circle by taking half of the length of the first straight line as a radius and taking the middle point of the first straight line as a circle center to obtain a first circle;
meanwhile, judging whether a first point exists between the first tangent line and the second tangent line and the first circle or not;
if the first points exist, connecting the second end point, the two first points and the first end point in sequence according to the appearance position sequence of the first points;
if only one first point exists, acquiring an arc corresponding to the first point, and taking the middle point of the arc as a connecting point to realize sequential connection of the second endpoint, the connecting point and the first endpoint;
if the first point does not exist, determining a cutting direction of the first circle according to the intersection point direction of the first tangent line and the second tangent line, cutting the corresponding semicircle in the cutting direction in parallel with the diameter, acquiring two cutting points, and connecting the two cutting points and the first end point in sequence;
constructing and obtaining a final boundary line based on all connection results;
the method for extracting the index parameters related to the preset index from each target flame area and obtaining the flame characteristics of the corresponding target flame area comprises the following steps:
the area growth rate of the target flame area of the two continuous frames of video images is calculated according to the following formula:
and calculating flame characteristics of a target flame area of each frame of video image according to the following formula:
and obtaining the flame characteristics of the target flame region based on the set of flame cluster characteristic values and the set of color characteristic values.
2. The infrared monitoring method for early warning of a campus fire according to claim 1, wherein the step of collecting standard infrared images of all normal working conditions in the campus based on the thermal infrared imager comprises the steps of:
counting to obtain all working condition conditions in the working process in the park, collecting infrared images of each working condition based on the working process of the thermal infrared imager, and placing the collected infrared images of each working condition in the same corresponding infrared image set to obtain a plurality of sets for storing the infrared images;
judging the working condition corresponding to each set of stored infrared images, and if no fire disaster occurs in the working process of the working condition, making the working condition normal;
selecting all infrared image sets with normal working conditions, taking all standard infrared images in the infrared image sets with normal working conditions as first infrared images, and storing the first infrared images in a memory.
3. The infrared monitoring method for early warning of a campus fire according to claim 1, wherein acquiring an actual infrared image of a working process of the site of the campus based on a thermal infrared imager and comparing the actual infrared image with the standard infrared image comprises:
thermal imaging data acquisition is carried out on the working process of the park site based on the thermal infrared imager, and acquisition is carried out to obtain
As a second infrared image;
reducing the first infrared image and the second infrared image to a preset size and performing gray level conversion;
comparing the gray values of two adjacent pixels in each row in a first gray image, marking 1 pixel with a large gray value, and marking 0 pixel with a small gray value to obtain a first fingerprint of the first gray image;
simultaneously, a second fingerprint of a second gray level image is obtained;
comparing the first fingerprint with the second fingerprint, and counting inconsistent positions;
if the count is less than 4, the second infrared image is considered to be a normal infrared image;
if the count is greater than 8, the second infrared image is considered an anomalous infrared image.
4. The infrared-monitored campus fire early warning method of claim 1, further comprising, prior to extracting flame shape features from the different video images: the flame identification model is constructed, and specifically comprises the following steps:
taking a first infrared image with normal working conditions as a first training sample, and constructing a first training sample set;
taking an infrared image of the working process of the fire disaster as a second training sample, and constructing a second training sample set;
extracting a second training sample without flame from the second training sample set, and constructing a third training sample set for the remaining second training samples with flame;
adding the extracted second training sample without flame into the first training sample set to obtain a fourth training sample set;
and training the third training sample set and the fourth training sample set by using the dert pre-training model to obtain a flame identification model.
5. The infrared-monitored campus fire early warning method of claim 1, wherein inputting the filtered first video image into the flame identification model, determining whether each frame of video image has flames generated, comprises:
decomposing the filtered first video image into a plurality of abnormal infrared images of continuous frames, and respectively inputting each frame of abnormal infrared image into a constructed flame identification model for identification;
respectively calculating the similarity between each frame of abnormal infrared image and each flame category in the flame identification model, and taking the category of the maximum similarity as the category of the abnormal infrared image of the corresponding frame;
and determining whether flame is generated in each frame of video image according to the category of each frame of abnormal infrared image in the first video image.
6. The infrared-monitored campus fire early warning method of claim 1, wherein the performing the corresponding early warning process according to the flame shape characteristic and the integrated change of the flame shape characteristic comprises:
recording the characteristic change of the target flame area in all the two continuous frames of video images;
and according to the recorded result, matching the early warning mode from the characteristic-early warning database, and carrying out early warning processing.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202310123403.4A CN115841731B (en) | 2023-02-16 | 2023-02-16 | Infrared monitoring park fire early warning method |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202310123403.4A CN115841731B (en) | 2023-02-16 | 2023-02-16 | Infrared monitoring park fire early warning method |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN115841731A CN115841731A (en) | 2023-03-24 |
| CN115841731B true CN115841731B (en) | 2023-05-05 |
Family
ID=85579793
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202310123403.4A Active CN115841731B (en) | 2023-02-16 | 2023-02-16 | Infrared monitoring park fire early warning method |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN115841731B (en) |
Families Citing this family (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN116740653A (en) * | 2023-08-14 | 2023-09-12 | 山东创亿智慧信息科技发展有限责任公司 | Distribution box running state monitoring method and system |
Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JPH08212354A (en) * | 1995-02-06 | 1996-08-20 | Toyota Motor Corp | Object recognizing method |
| CN102156751A (en) * | 2011-04-26 | 2011-08-17 | 深圳市迅雷网络技术有限公司 | Method and device for extracting video fingerprint |
| CN103473788A (en) * | 2013-07-31 | 2013-12-25 | 中国电子科技集团公司第三十八研究所 | Indoor fire and flame detection method based on high-definition video images |
| CN105046868A (en) * | 2015-06-16 | 2015-11-11 | 苏州华启智能科技股份有限公司 | Fire early warning method based on infrared thermal imager in narrow environment |
| CN112115766A (en) * | 2020-07-28 | 2020-12-22 | 辽宁长江智能科技股份有限公司 | Flame identification method, device, equipment and storage medium based on video picture |
| CN115601919A (en) * | 2022-10-28 | 2023-01-13 | 青鸟消防股份有限公司（Cn） | Fire alarm method based on Internet of things equipment and video image comprehensive identification |
-
2023
- 2023-02-16 CN CN202310123403.4A patent/CN115841731B/en active Active
Patent Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JPH08212354A (en) * | 1995-02-06 | 1996-08-20 | Toyota Motor Corp | Object recognizing method |
| CN102156751A (en) * | 2011-04-26 | 2011-08-17 | 深圳市迅雷网络技术有限公司 | Method and device for extracting video fingerprint |
| CN103473788A (en) * | 2013-07-31 | 2013-12-25 | 中国电子科技集团公司第三十八研究所 | Indoor fire and flame detection method based on high-definition video images |
| CN105046868A (en) * | 2015-06-16 | 2015-11-11 | 苏州华启智能科技股份有限公司 | Fire early warning method based on infrared thermal imager in narrow environment |
| CN112115766A (en) * | 2020-07-28 | 2020-12-22 | 辽宁长江智能科技股份有限公司 | Flame identification method, device, equipment and storage medium based on video picture |
| CN115601919A (en) * | 2022-10-28 | 2023-01-13 | 青鸟消防股份有限公司（Cn） | Fire alarm method based on Internet of things equipment and video image comprehensive identification |
Also Published As
| Publication number | Publication date |
|---|---|
| CN115841731A (en) | 2023-03-24 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN114549522A (en) | Textile quality detection method based on target detection | |
| CN103034838B (en) | A kind of special vehicle instrument type identification based on characteristics of image and scaling method | |
| CN106295124B (en) | The method of a variety of image detecting technique comprehensive analysis gene subgraph likelihood probability amounts | |
| WO2021139258A1 (en) | Image recognition based cell recognition and counting method and apparatus, and computer device | |
| CN115841731B (en) | Infrared monitoring park fire early warning method | |
| CN112819093A (en) | Man-machine asynchronous recognition method based on small data set and convolutional neural network | |
| CN113674273B (en) | Optical detection method and system based on product defects and readable storage medium | |
| CN106056597A (en) | Object visual detection method and device | |
| CN104700424B (en) | Medical color fujinon electronic video endoscope dead pixel points of images detection means | |
| CN106355584A (en) | Automatic detection method for microaneurysm in eye fundus image on basis of local entropy determining threshold | |
| CN115994907B (en) | Intelligent processing system and method for comprehensive information of food detection mechanism | |
| CN114898261A (en) | Sleep quality assessment method and system based on fusion of video and physiological data | |
| TWI628625B (en) | Method of producing imaging biomarker for brain disease | |
| CN115937085B (en) | Nuclear cataract image processing method based on neural network learning | |
| CN112036391B (en) | Electronic inspection method and system based on shale gas field production station | |
| CN114065798A (en) | Visual identification method and device based on machine identification | |
| CN115239663A (en) | Method and system for detecting defects of contact lens, electronic device and storage medium | |
| CN113989588A (en) | Self-learning-based intelligent evaluation system and method for pentagonal drawing test | |
| CN117576100B (en) | Surface defect grading detection and evaluation method for FPC connector | |
| CN116168223B (en) | Multi-mode-based peanut leaf spot disease grade detection method | |
| CN115082865B (en) | Bridge crane intrusion dangerous behavior early warning method and system based on visual image recognition | |
| CN117197139B (en) | Tongue diagnosis image multi-label classification method based on AI | |
| CN117314899B (en) | Carbon fiber plate quality detection method based on image characteristics | |
| CN117115172B (en) | Donkey-hide gelatin quality detection method and system based on machine vision | |
| CN116311248A (en) | Detection method for paper fox spots |
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 |