Smoke and fire identification system

Abstract

The invention discloses a smoke and fire identification system, which comprises a smoke and fire image acquisition unit for acquiring a smoke and fire image of a smoke and fire generation area through image acquisition equipment; the smoke and fire surrounding environment data acquisition unit is used for acquiring smoke and fire image and acquiring position information generated by smoke and fire through the GPS positioning device at the same time; the smoke and fire image processing unit is used for splitting the smoke and fire image into a plurality of continuous picture frame data; the fire source diffusion identification unit is used for locating and inquiring whether flammable and explosive substances exist in surrounding buildings and weather data at locating positions in the urban BIM system according to the position information acquired by the GPS locating device; splitting the smoke and fire image frame data into a time sequence, and identifying weather data in the time sequence process; the smoke and fire combustion trend prediction unit is used for comprehensively predicting the smoke and fire combustion trend expansion possibility according to the wind direction data, the wind speed data value and the smoke and fire position direction of the flammable and combustible substances in the weather data; and the smoke and fire comprehensive early warning unit is used for performing early warning on the judgment result of the false expansion possibility of the smoke and fire combustion trend.

Description

Technical Field

[0001] The present invention belongs to the technical field of image processing, and specifically relates to a fireworks recognition system. Background Art

[0002] With the development of society, people's awareness of early disaster prevention is constantly improving, and the demand is also gradually increasing. For example, the prevention of fires is what we need. Whether we can provide early protection for a fire at its initial stage is crucial for life and property safety. However, the difficulty of fireworks recognition lies in the abstract features, diverse forms, unclear features, and few samples in the actual scene, making it difficult to obtain. Moreover, whether it evolves into a large-scale fire scene is not only related to features such as the grayscale, color, shape, texture, and change trend of fireworks, but may also be related to other relevant factors such as the environment, the combustible materials, and the fastest rescue speed.

[0003] The defect of the prior art is that it lacks consideration of the combustibles in the environment or other flammable and explosive substances in the environment, resulting in the inability of fireworks recognition to predict the change of the fire situation. Summary of the Invention

[0004] In view of the problem that the existing fireworks recognition system cannot predict the change of the fire situation, the present invention provides a fireworks recognition system.

[0005] To achieve the above technical purpose, the technical solution adopted by the present invention is as follows:

[0006] A fireworks recognition system includes a fireworks image acquisition unit, a fireworks surrounding environment data acquisition unit, a fireworks image processing unit, a fire source diffusion recognition unit, a fireworks combustion trend prediction unit, and a fireworks warning unit;

[0007] The fireworks image acquisition unit acquires the fireworks image of the fireworks generation area through an image acquisition device;

[0008] The fireworks surrounding environment data acquisition unit acquires the position information of the fireworks generation through a GPS positioning device while collecting the fireworks image;

[0009] The fireworks image processing unit splits the fireworks image into multiple consecutive picture frame data and scales the picture frame data to the same pixel size; it is convenient to subsequently compare the fireworks areas in the picture frame data of the same size.

[0010] The fire source diffusion recognition unit locates and queries whether there are flammable and explosive substances in the surrounding buildings and the weather data at the positioning location in the urban BIM system through the position information collected by the GPS positioning device; it splits the time series of the fireworks image picture frame data and identifies the weather data during this time series; it is convenient for subsequent prediction of the fireworks combustion trend.

[0011] The fireworks combustion trend prediction unit comprehensively predicts the possibility of the expansion of the fireworks combustion trend by combining the wind direction data, wind speed data value in the weather data, and the direction of the fireworks position where inflammable and explosive substances are located.

[0012] The fireworks comprehensive warning unit is used to give a warning about the judgment result of the incorrect expansion possibility of the fireworks combustion trend in the fireworks combustion trend prediction unit.

[0013] Furthermore, the detailed steps for predicting the incorrect expansion possibility of the fireworks combustion trend:

[0014] S101. Analyze the data of multiple consecutive picture frames of the fireworks image to obtain the combustion direction of the ignition point.

[0015] S102. Combine the wind direction data and wind speed data value in the weather data to judge whether there is a trend to increase the combustion direction of the ignition point. If there is an increasing trend, go to the next step S103; if there is no increasing trend, judge that the fire situation is stable.

[0016] S103. Determine whether the combustion direction of the ignition point is consistent with the direction of the fireworks position where inflammable and explosive substances are located, and further judge whether there is a possibility of a large-scale fire. If so, go to the next step S104; if not, judge it as a small-scale fire.

[0017] S104. Send it to the fireworks comprehensive warning unit and issue a large-scale fire warning message to timely control the growth trend of the fire.

[0018] Furthermore, after the acquisition of the fireworks image and the image data of the substances appearing in the surrounding environment of the fireworks, the first acquisition method:

[0019] First, use an industrial camera to obtain a video of the location where the fireworks are generated.

[0020] Then split the video into picture frames.

[0021] Finally, after calculating the grayscale value of the entire picture frame, split the same picture frame into a fireworks picture, weather data, and other environmental factors, and perform the splitting using different grayscale values.

[0022] Furthermore, for the acquisition of the fireworks image and the image data of the substances appearing in the surrounding environment of the fireworks, the second acquisition method:

[0023] First, use an industrial camera to obtain a video of the location where the fireworks are generated.

[0024] Then split the video into picture frames.

[0025] Finally, according to the gray-scale values of the preset fireworks picture part in the entire picture frame, it is brought into the picture frame for comparison. First, the picture parts within the gray-scale value range of the preset fireworks picture part are screened out as the fireworks picture part. Then, the parts where the gray-scale values of other environmental factors are the same as those of the weather at the positioning location are split out, and the remaining part is the weather data part.

[0026] Further, in the fireworks image processing unit, since the industrial camera captures the fireworks image under the condition of a dynamically rotating and stretching lens, the obtained picture frame data needs to be transformed to the same perspective or the picture frame data needs to be scaled to the same pixel condition of the same size.

[0027] Further, it also includes a deep learning model prediction unit. After calculating and capturing the fireworks area in the picture frame data processed by the fireworks image processing unit, it is input into the deep learning model to compare whether the continuous change threshold of the fireworks area in the historical fire occurrence is less than the change threshold of the fireworks area in the current picture frame data.

[0028] Further, in the deep learning model prediction unit, the area change data value of the gray-scale value of the fireworks part in the picture frame data is input into the deep neural network model for deep learning. The input layer is the average value of the area of the initial first section of the fireworks picture part in the picture frame data of the fireworks image, and the hidden layer is the average value of the area of the second section of the fireworks picture part in the picture frame data of the fireworks image; the output layer is the average value of the area of the third section of the fireworks picture part in the picture frame data of the fireworks image; after comparing the three continuous change values of the fireworks area in the historical fire occurrence one by one to obtain the maximum change threshold, it is judged whether the change threshold obtained by comparing the three continuous change values of the fireworks in the current picture frame data one by one is greater than or equal to the maximum change threshold in the historical fire occurrence. If it is greater than or equal to, it is determined that a fire may be triggered at the current fireworks occurrence location. If it is less than the maximum change threshold in the historical fire occurrence, the possibility of triggering a fire is relatively small.

[0029] Further, in the same fireworks environment, the area change trend of the same weather data is inversely proportional to the fireworks area in the picture frame data; that is, the growth ratio of the change threshold of the fireworks area in the picture frame data is larger than the maximum change threshold in the historical fire occurrence. At the same time, if the reduction ratio of the area change threshold of the same weather data in the same fireworks environment is faster or larger than the maximum change threshold in the historical fire occurrence, it is determined that the same weather data in the same fireworks environment has an effect of assisting combustion; conversely, when it is directly proportional, there is no effect of assisting combustion.

[0030] The present invention has the following beneficial effects compared with the prior art:

[0031] By obtaining the fireworks image, the surrounding environment of the fireworks, and the weather data, and inputting them into the deep learning model to judge the growth trend of the fireworks area; the area of the fireworks picture frame and the area of the weather data around the fireworks are input into the time series prediction model to judge whether the continuous change value of the fireworks in the current picture frame data exceeds the historical threshold, and to judge whether the weather data in the fireworks environment has an effect of assisting combustion, so as to achieve the warning level of comprehensively judging the change trend of the fireworks image. Brief Description of the Drawings

[0032] Figure 1 It is the overall structural block diagram of a fireworks recognition system in an embodiment of the present invention. Detailed Embodiment

[0033] For the convenience of understanding by those skilled in the art, the present invention will be further described below in conjunction with the embodiments and the drawings. The content mentioned in the embodiments does not limit the present invention.

[0034] As Figure 1 shown, this embodiment provides a fireworks recognition system, including a fireworks image acquisition unit, a fireworks surrounding environment data acquisition unit, a fireworks image processing unit, a fire source diffusion recognition unit, a fireworks combustion trend prediction unit, and a fireworks warning unit;

[0035] The fireworks image acquisition unit acquires the fireworks image of the fireworks generation area through an image acquisition device;

[0036] The fireworks surrounding environment data acquisition unit acquires the position information of the fireworks generation through a GPS positioning device while acquiring the fireworks image;

[0037] The fireworks image processing unit splits the fireworks image into multiple consecutive picture frame data and scales the picture frame data to the same pixel size; it is convenient to grab the fireworks area in the picture frame data for comparison of the same size later.

[0038] The fire source diffusion recognition unit locates and queries whether there are flammable and explosive substances in the surrounding buildings and the weather data at the positioning location in the urban BIM system through the position information collected by the GPS positioning device; the weather data is recognized in the time series of the split picture frame data of the fireworks image; it is convenient for subsequent prediction of the fireworks combustion trend.

[0039] The fireworks combustion trend prediction unit comprehensively predicts the possibility of the expansion of the fireworks combustion trend by combining the wind direction data, wind speed data value in the weather data, and the direction of the position of the flammable and explosive substances in the fireworks;

[0040] The fireworks comprehensive warning unit is used to give a warning about the judgment result of the possible mis-expansion of the fireworks combustion trend in the fireworks combustion trend prediction unit.

[0041] Detailed steps for predicting the possible mis-expansion of the fireworks combustion trend:

[0042] S101. Analyze the data of multiple consecutive picture frames of the fireworks image to obtain the burning direction of the ignition point.

[0043] S102. Combine the wind direction data and wind speed data values in the weather data to determine whether there is a trend of increasing the burning direction of the ignition point. If there is an increasing trend, proceed to the next step S103; if there is no increasing trend, determine that the fire situation is stable.

[0044] S103. Further determine whether there is a possibility of a large - scale fire by checking whether the burning direction of the ignition point is consistent with the direction of the location of flammable and explosive substances in the fireworks. If so, proceed to the next step S104; if not, determine it as a small - scale fire.

[0045] S104. Send it to the integrated fireworks warning unit and issue a large - scale fire warning message to timely control the increasing trend of the fire.

[0046] After the acquisition of the data of the fireworks image and the substance image that appears in the surrounding environment of the fireworks, the first acquisition method:

[0047] First, use an industrial camera to obtain a video of the location where the fireworks are generated.

[0048] Then, split the video into picture frames.

[0049] Finally, after calculating the grayscale values of the entire picture frame, split the same picture frame into fireworks pictures, weather data, and other environmental factors, and perform the split by different grayscale values. Usually, the grayscale values of other environmental factors are the same as the grayscale values of the weather at the positioning location. In each frame of the picture, except for other environmental factors, the remaining part is the fireworks picture and weather data part, and then they are split through different grayscale values to finally obtain the fireworks image of the fireworks picture part and the data of the substance image that appears in the surrounding environment of the fireworks in the weather data part.

[0050] The acquisition of the data of the fireworks image and the substance image that appears in the surrounding environment of the fireworks, the second acquisition method:

[0051] First, use an industrial camera to obtain a video of the location where the fireworks are generated.

[0052] Then, split the video into picture frames.

[0053] Finally, compare by substituting the preset grayscale value of the fireworks picture part into the picture frame. First, screen out the picture part within the range of the preset grayscale value of the fireworks picture part as the fireworks picture part, and then split out the part where the grayscale value of other environmental factors is the same as the grayscale value of the weather at the positioning location. The remaining part is the weather data part.

[0054] In the firework image processing unit, since the industrial camera captures firework images under the condition of a dynamically rotating and stretching lens, the acquired picture frame data needs to be transformed to the same perspective or the picture frame data needs to be scaled to the same pixel condition with the same size.

[0055] It also includes a deep learning model prediction unit. After calculating and capturing the firework area in the picture frame data processed by the firework image processing unit, it is input into the deep learning model to compare whether the continuous change threshold of the firework area in historical fire occurrences is less than the change threshold of the firework area in the current picture frame data.

[0056] In the deep learning model prediction unit, the area change data value of the gray value of the firework part in the picture frame data is input into the deep neural network model for deep learning. The input layer is the average value of the area of the initial first section of the firework picture part in the picture frame data of the firework image, and the hidden layer is the average value of the area of the second section of the firework picture part in the picture frame data of the firework image; the output layer is the average value of the area of the third section of the firework picture part in the picture frame data of the firework image; after comparing the three-section continuous change values of the firework area in historical fire occurrences one by one to obtain the maximum change threshold, it is judged whether the change threshold obtained by comparing the three-section continuous change values of the firework in the current picture frame data one by one is greater than or equal to the maximum change threshold in historical fire occurrences. If it is greater than or equal to, it is determined that the current firework occurrence location may cause a fire. If it is less than the maximum change threshold in historical fire occurrences, the possibility of causing a fire is relatively small.

[0057] In the same firework environment with the same weather data, the area change trend is inversely proportional to the firework area in the picture frame data; that is, the growth ratio of the change threshold of the firework area in the picture frame data is larger than the maximum change threshold in historical fire occurrences. At the same time, if the reduction ratio of the change threshold of the area of the same weather data in the same firework environment is faster or larger than the maximum change threshold in historical fire occurrences, it is determined that the same weather data in the same firework environment has an effect of assisting combustion; otherwise, when it is directly proportional, there is no effect of assisting combustion.

[0058] Compared with the prior art, the present invention has the following beneficial effects:

[0059] By acquiring firework images, the surrounding environment and weather data of the firework, and inputting them into the deep learning model to judge the growth trend of its firework area; the area of the firework picture frame and the area of the weather data around the firework are input into the time series prediction model to judge whether the continuous change value of the firework in the current picture frame data exceeds the historical threshold, and to judge whether the weather data in the firework environment has an effect of assisting combustion, so as to achieve the warning level of comprehensively judging the change trend of the firework image.

[0060] The above has provided a detailed introduction to a fireworks recognition system of the present application. The description of the specific embodiments is only used to help understand the method and its core idea of the present application. It should be noted that for those of ordinary skill in the art of this technology, without departing from the principle of the present application, several improvements and modifications can be made to the present application, and these improvements and modifications also fall within the protection scope of the claims of the present application.

Claims

1. A fireworks recognition system, characterized in that: It includes a fireworks image acquisition unit, a fireworks surrounding environment data acquisition unit, a fireworks image processing unit, a fire source diffusion identification unit, a fireworks burning trend prediction unit and a fireworks early warning unit; A fireworks image acquisition unit, which acquires fireworks images of the fireworks generation area through an image acquisition device; The fireworks surrounding environment data acquisition unit acquires the location information of the fireworks through the GPS positioning device while collecting the fireworks image; A fireworks image processing unit, which splits the fireworks image into a plurality of continuous picture frame data and scales the picture frame data to pixels of a uniform size; The fire source spread identification unit uses the location information collected by the GPS positioning device to locate and query whether there are flammable and explosive materials in the surrounding buildings and the weather data at the location in the urban BIM system; the fireworks image frame data is split into time series and the weather data during the time series is identified; this facilitates the subsequent prediction of fireworks combustion trends. The firework burning trend prediction unit comprehensively predicts the possibility of the firework burning trend expansion by combining the wind direction data, wind speed data values ​​and the location and direction of the firework where the flammable and explosive materials are located in the weather data; The comprehensive fireworks warning unit is used to warn of the incorrect judgment result of the possibility of the fireworks burning trend expanding in the fireworks burning trend prediction unit.

2. A fireworks identification system according to claim 1, characterized in that: Detailed steps to predict the possibility of incorrect expansion of the burning trend of fireworks: S101, analyzing multiple continuous picture frame data of fireworks images to obtain the burning direction of the ignition point; S102, judging whether there is a trend of increasing the burning direction of the ignition point in combination with the wind direction data and wind speed data values ​​in the weather data, if there is an increasing trend, proceeding to the next step S103, if there is no increasing trend, judging that the fire is stable; S103, whether the burning direction of the ignition point is consistent with the direction of the pyrotechnic position of the flammable and explosive material, further determine whether there is a possibility of a large fire, if yes, proceed to the next step S104; if not, determine it to be a small fire; S104: Send to the fire and smoke integrated warning unit and issue a large-scale fire warning information to control the fire growth trend in time.

3. A fireworks identification system according to claim 2, characterized in that: After collecting the image data of fireworks and the material image data appearing in the surrounding environment of fireworks, the first acquisition method is: First, an industrial camera is used to obtain a video of the location where the fireworks are generated; Then split the video into picture frames; Finally, after calculating the grayscale value of the entire picture frame, the same picture frame is segmented into fireworks pictures, weather data, and other environmental factors, and the segmentation is performed using different grayscale values.

4. A fireworks identification system according to claim 2, characterized in that: The second acquisition method is to collect the image data of fireworks and the materials appearing in the surrounding environment of fireworks: First, an industrial camera is used to obtain a video of the location where the fireworks are generated; Then split the video into picture frames; Finally, the grayscale value of the fireworks picture part preset in the entire picture frame is brought into the picture frame for comparison. First, the picture part within the preset fireworks picture part grayscale value range is screened out as the fireworks picture part, and then the part with the grayscale value of other environmental factors consistent with the weather grayscale value of the positioning location is split out, and the remaining part is the weather data part.

5. A fireworks identification system according to claim 3 or 4, characterized in that: In the fireworks image processing unit, since the industrial camera collects fireworks images under the condition of dynamically rotating and stretching the lens, the acquired picture frame data needs to be converted to the same viewing angle or the picture frame data needs to be scaled to pixels of the same size.

6. A fireworks identification system according to claim 5, characterized in that: It also includes a deep learning model prediction unit, which captures and calculates the fireworks area in the picture frame data processed by the fireworks image processing unit and inputs it into the deep learning model to compare whether the continuous change threshold of the fireworks area caused by historical fires is smaller than the fireworks area change threshold in the current picture frame data.

7. A fireworks identification system according to claim 6, characterized in that: The grayscale value area change data value of the fireworks part in the picture frame data in the deep learning model prediction unit is input into the deep network model for deep learning, the input layer is the average value of the area of ​​the fireworks picture part in the initial first segment of the picture frame data in the fireworks image, and the hidden layer is the average value of the area of ​​the fireworks picture part in the second segment of the picture frame data in the fireworks image; the output layer is the average value of the area of ​​the fireworks picture part in the third segment of the picture frame data in the fireworks image; the maximum change threshold is obtained by comparing the three consecutive change values ​​of the fireworks area of ​​historical fires one by one, and it is determined whether the change threshold obtained by comparing the three consecutive change values ​​of the fireworks in the current picture frame data one by one is greater than or equal to the maximum change threshold of historical fires. If it is greater than or equal to, it is determined that the current location of the fireworks may cause a fire. If it is less than the maximum change threshold of historical fires, the possibility of causing a fire is small.

Images