Intelligent alarm method and system based on infrared thermal imaging technology

Abstract

The invention relates to the technical field of smart home, and particularly discloses an intelligent alarm method and system based on an infrared thermal imaging technology, and the method comprises the steps: collecting an infrared thermal image sequence of a home environment based on an infrared detector; performing threshold segmentation on each frame of image in the infrared thermal image sequence based on a preset temperature threshold to extract a high-temperature region image sequence; forming a cross-frame high-temperature region tracking trajectory between continuous frames of the high-temperature region image sequence by using a target tracking algorithm; obtaining time sequence change parameter information of each high-temperature region according to the high-temperature region tracking trajectory and the high-temperature region image sequence; distinguishing the target type of each high-temperature area according to the time sequence change parameter information; if the high-temperature area of the biological target type exists, generating an alarm signal; the method has the capability of accurately distinguishing the biological target and the sunlight target, the reliability and accuracy of detection are effectively improved, and the false alarm rate can be effectively reduced.

Description

Technical Field

[0001] This application relates to the technical field of smart home, and specifically, to an intelligent alarm method and system based on infrared thermal imaging technology. Background Art

[0002] As a non-contact detection means, infrared thermal imaging technology plays an increasingly important role in the field of security monitoring, especially in home security monitoring, because it can work under conditions of no light or weak light and can reflect the temperature information of objects. By collecting infrared thermal images of the home environment, the system can sense potential targets with temperature characteristics, thereby realizing security warnings.

[0003] However, in the actual application of the home environment, the infrared thermal imaging alarm system faces various interference sources. One common and difficult-to-effectively-solve problem is sunlight interference. Sunlight enters the room through windows, glass doors, etc., and will form areas with relatively high temperatures on the ground or walls, that is, sunlight heat areas. The temperature of such sunlight heat areas is often higher than the ambient background temperature and is easily misidentified by the system as potential threat targets. More challenging is that the characteristics of sunlight heat areas are complex and dynamically changing. Its position will slowly drift and move with the change of the sun's angle; its shape and size may change due to factors such as window structure and indoor item occlusion; at the same time, its temperature and brightness will also fluctuate with environmental factors such as external light intensity and cloud changes.

[0004] This complex interference situation makes it difficult for traditional infrared thermal imaging alarm systems to accurately distinguish real biological targets (such as humans or pets) from non-biological high-temperature areas caused by sunlight. Due to the high-temperature characteristics and dynamic changes of sunlight heat areas, the system is prone to misjudge them as biological targets, resulting in a large number of false alarms. These false alarms will not only frequently interrupt the normal life of users, reduce the trust and satisfaction of users with the alarm system, but also more likely lead to a decrease in the trust of users in the alarm system, thus ignoring the alarm signal in the event of a real security incident, causing serious consequences.

[0005] In view of the above problems, there is currently no effective technical solution. Summary of the Invention

[0006] The purpose of this application is to provide an intelligent alarm method and system based on infrared thermal imaging technology to accurately distinguish biological targets and sunlight targets, improve the reliability and accuracy of detection, and effectively reduce the false alarm rate.

[0007] In a first aspect, this application provides an intelligent alarm method based on infrared thermal imaging technology, which is applied in a home environment. The method includes the following steps: S1. Collect an infrared thermal image sequence of the home environment based on an infrared detector; S2. Perform threshold segmentation on each frame image in the infrared thermal image sequence based on a preset temperature threshold to extract a high-temperature region image sequence; S3. Between consecutive frames of the high-temperature region image sequence, use an object tracking algorithm to associate the high-temperature regions between the frames of the high-temperature region image sequence to form a cross-frame high-temperature region tracking trajectory; S4. Obtain the temporal change parameter information of each high-temperature region according to the high-temperature region tracking trajectory and the high-temperature region image sequence; S5. Distinguish the target types of each high-temperature region according to the temporal change parameter information, where the target types are biological targets or sunlight targets; S6. If there is a high-temperature region with a biological target type, generate an alarm signal.

[0008] The intelligent alarm method based on infrared thermal imaging technology of this application obtains a cross-frame high-temperature region tracking trajectory by associating high-temperature regions based on time series, sequentially tracks and analyzes to obtain the temporal change parameter information of the high-temperature region, and distinguishes biological targets and sunlight targets by analyzing the temporal change parameter information of the high-temperature region. It has the ability to accurately distinguish biological targets and sunlight targets, effectively improves the reliability and accuracy of detection, and can effectively reduce the false alarm rate.

[0009] For the intelligent alarm method based on infrared thermal imaging technology described above, the temporal change parameter information includes a sequence of change rates of the central position, a sequence of change rates of the region area, a sequence of changes in shape parameters, a sequence of change rates of the average temperature of the region, and a sequence of changes in the internal temperature distribution of the region.

[0010] In this example, the sequence of change rates of the central position reflects the moving speed and direction of the high-temperature region; the sequence of change rates of the region area reflects the change in the size of the high-temperature region; the sequence of changes in shape parameters reflects the change in the shape of the high-temperature region; the sequence of change rates of the average temperature of the region reflects the change trend of the average temperature of the high-temperature region; and the sequence of changes in the internal temperature distribution of the region reflects the change in the internal heat distribution of the high-temperature region.

[0011] For the intelligent alarm method based on infrared thermal imaging technology described above, between step S2 and step S3, there is also a step: SA. Perform morphological filtering processing on the high-temperature region image sequence.

[0012] In this example, the morphological filtering processing improves the quality of the high-temperature region image sequence, provides more suitable data for subsequent object tracking, reduces tracking errors caused by regional morphology problems, and thus reduces the possibility of false alarms.

[0013] The intelligent alarm method based on infrared thermal imaging technology, wherein the morphological filtering process includes erosion and dilation operations, and the erosion operation includes: SA1. Based on a pre-determined set of structural elements and the first number of iterations, perform an erosion operation on each frame image in the high-temperature region image sequence to eliminate small noise points and isolated regions; The dilation operation includes: SA2. Based on a pre-determined set of structural elements and the second number of iterations, perform a dilation operation on each frame image in the eroded high-temperature region image sequence to smooth the region edges and fill the holes inside the regions.

[0014] The intelligent alarm method based on infrared thermal imaging technology, wherein the combination of structural elements includes square structural elements, circular structural elements, and linear structural elements, and step SA1 includes: SA11. For each frame image in the high-temperature region image sequence, perform an erosion operation using a square structural element, a circular structural element, and a linear structural element respectively to obtain multiple eroded frame images; SA12. Calculate the edge roughness of each eroded frame image, and select the frame image with the lowest edge roughness as the erosion result of the frame image in the corresponding high-temperature region image sequence to obtain the eroded high-temperature region image sequence.

[0015] The intelligent alarm method based on infrared thermal imaging technology, wherein step SA2 includes: SA21. For each frame image in the eroded high-temperature region image sequence, perform a dilation operation using a square structural element, a circular structural element, and a linear structural element respectively to obtain multiple dilated frame images; SA22. Calculate the area change rate between the high-temperature region in each dilated frame image and the high-temperature region in the corresponding frame image in the high-temperature region image sequence before erosion, and select the frame image with the lowest area change rate as the dilation result of the frame image in the corresponding high-temperature region image sequence to obtain the high-temperature region image sequence after morphological filtering processing.

[0016] The intelligent alarm method based on infrared thermal imaging technology, wherein step S5 includes: S51. For each high-temperature region, based on a pre-established parameter feature library of biological targets and sunlight targets, calculate the similarity score between the temporal change parameter information of each high-temperature region and the corresponding parameter features in the parameter feature library of biological targets and sunlight targets to obtain the similarity scores of the high-temperature region belonging to biological targets and sunlight targets; S52. According to the similarity scores of biological targets and sunlight targets and a preset classification rule, determine the target types of each high-temperature region.

[0017] The intelligent alarm method based on infrared thermal imaging technology, wherein the target tracking algorithm includes the KCF algorithm and the Kalman filtering algorithm, and step S3 includes: S31. Initialize the target tracker for each high-temperature area based on the KCF algorithm, so as to use the target tracker to track the target of the frame image of the high-temperature area image, predict the new position frame by frame and calculate the confidence score. If the confidence score is higher than the preset score threshold, update the target tracker of the corresponding high-temperature area and associate the cross-frame trajectory. If the confidence score is lower than the preset score threshold, predict the potential position through the Kalman filtering algorithm and re-initialize the target tracker of the corresponding high-temperature area. If the initialization is successful, continue to track the high-temperature area, otherwise end the tracking of the high-temperature area; S32. Generate cross-frame high-temperature area tracking trajectories according to the tracking results of each target tracker.

[0018] The intelligent alarm method based on infrared thermal imaging technology, wherein step S2 includes: S21. For each frame image in the infrared thermal image sequence, compare the temperature value of the pixel point with the preset temperature threshold, extract the pixel points with temperatures higher than the preset temperature threshold, and form a binary image; S22. Perform connected component division on the binary image to form an initial high-temperature area; S23. Filter out the initial high-temperature areas in the binary image that are smaller than the preset minimum area, and determine each frame image in the high-temperature area image sequence.

[0019] In a second aspect, the present application also provides an intelligent alarm system based on infrared thermal imaging technology, which is applied in a home environment. The system includes: An infrared detector for collecting an infrared thermal image sequence of the home environment; An alarm component for performing an alarm operation; A control component for performing the intelligent alarm method based on infrared thermal imaging technology provided in the first aspect to generate an alarm signal to trigger the alarm component to perform an alarm operation.

[0020] The intelligent alarm system based on infrared thermal imaging technology of the present application can analyze the temporal change characteristics of the high-temperature area, and the system can identify sunlight targets with complex change rules, avoid misjudging them as intruders, and thus reduce the false alarm rate caused by sunlight interference.

[0021] As can be seen from the above, the present application provides an intelligent alarm method and system based on infrared thermal imaging technology. Among them, the intelligent alarm method based on infrared thermal imaging technology in the present application obtains a cross-frame high-temperature area tracking trajectory by associating high-temperature areas based on time series, sequentially tracks and analyzes to obtain the time series change parameter information of the high-temperature area, and distinguishes biological targets and sunlight targets by analyzing the time series change parameter information of the high-temperature area. It has the ability to accurately distinguish biological targets and sunlight targets, effectively improves the reliability and accuracy of detection, and can effectively reduce the false alarm rate. BRIEF DESCRIPTION OF THE DRAWINGS

[0022] Figure 1 It is a flowchart of an intelligent alarm based on infrared thermal imaging technology provided by some embodiments of the present application.

[0023] Figure 2 It is a flowchart of an intelligent alarm based on infrared thermal imaging technology provided by some other embodiments of the present application.

[0024] Figure 3 It is a simplified flowchart of an intelligent alarm based on infrared thermal imaging technology provided by an embodiment of the present application.

[0025] Figure 4 It is a schematic diagram of the electronic control structure of an intelligent alarm system based on infrared thermal imaging technology provided by an embodiment of the present application.

[0026] Reference numerals: 201, infrared detector; 202, alarm component; 203, control component. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0027] Next, the technical solutions in the embodiments of the present application will be clearly and completely described in conjunction with the drawings in the embodiments of the present application. Obviously, the described embodiments are only a part of the embodiments of the present application, rather than all the embodiments. Usually, the components of the embodiments of the present application described and illustrated herein can be arranged and designed in various different configurations. Therefore, the following detailed description of the embodiments of the present application provided in the drawings is not intended to limit the scope of the present application to be protected, but only represents the selected embodiments of the present application. Based on the embodiments of the present application, all other embodiments obtained by those skilled in the art without creative efforts belong to the scope of protection of the present application.

[0028] It should be noted that: similar reference numerals and letters represent similar items in the following drawings. Therefore, once an item is defined in one drawing, it does not need to be further defined and explained in subsequent drawings. At the same time, in the description of the present application, terms such as "first" and "second" are only used for distinguishing descriptions, and cannot be understood as indicating or implying relative importance.

[0029] In the first aspect, please refer toFigures 1 - 3 , some embodiments of the present application provide an intelligent alarm method based on infrared thermal imaging technology for a home environment. When there is a non-biological high-temperature area formed by sunlight passing through a window, it distinguishes biological targets from the non-biological high-temperature area. The steps of the method include: S1. Collect an infrared thermal image sequence of the home environment based on an infrared detector; S2. Perform threshold segmentation on each frame image in the infrared thermal image sequence based on a preset temperature threshold to extract a high-temperature area image sequence; S3. Between consecutive frames of the high-temperature area image sequence, use a target tracking algorithm to associate the high-temperature areas between each frame image of the high-temperature area image sequence to form a cross-frame high-temperature area tracking trajectory; S4. Obtain the temporal variation parameter information of each high-temperature area according to the high-temperature area tracking trajectory and the high-temperature area image sequence; S5. Distinguish the target type of each high-temperature area according to the temporal variation parameter information, and the target type is a biological target or a sunlight target; S6. If there is a high-temperature area with a biological target type, generate an alarm signal.

[0030] Specifically, the infrared detector is preferably an uncooled infrared detector. In step S1, the infrared detector obtains temperature distribution data of the home environment changing with time. Step S2 limits the analysis scope to potential heat sources. Step S3 is used to track and obtain the dynamic behavior of each high-temperature area. To distinguish these heat sources, the method performs target tracking on the high-temperature areas between consecutive image frames. Through tracking, the association of each high-temperature area in time can be established to form a high-temperature area tracking trajectory. The temporal variation parameter information obtained in step S4 quantifies the characteristics of the high-temperature area changing with time, such as moving speed, area size change, temperature change, etc., so that step S5 can distinguish the target type corresponding to the high-temperature area according to the temporal variation parameter information, and then determine whether it is necessary to execute step S6 to trigger the alarm behavior. Thus, the high-temperature areas caused by non-biological heat sources such as sunlight will not trigger an alarm, thereby reducing the false alarm rate.

[0031] More specifically, the core idea of the intelligent alarm method based on infrared thermal imaging technology in the embodiments of the present application is to distinguish biological targets and sunlight areas by analyzing the dynamic characteristics of high-temperature areas over time. In step S2, all high-temperature areas with temperatures higher than a preset threshold are extracted to form a high-temperature area image sequence, and these high-temperature areas may contain biological targets and sunlight areas. To distinguish them, the method performs target tracking on these high-temperature areas between consecutive frames to establish the tracking trajectories of each high-temperature area over time. Thus, the system can track changes such as the movement, size, shape, and temperature of each high-temperature area. Then, based on these tracking trajectories and the original image data, the temporal change parameter information of each high-temperature area is calculated, such as the moving speed, area change rate, temperature change trend, etc. These parameters are the key features for distinguishing biological targets and sunlight areas because biological targets usually exhibit specific movement patterns and temperature changes, while sunlight areas have different temporal characteristics. Finally, based on this temporal change parameter information, the method determines whether each high-temperature area is a biological target or a sunlight target. If a biological target is determined to exist, an alarm signal is generated. Through this discrimination method based on temporal change parameters, the system can effectively identify true biological targets and significantly reduce false alarms caused by sunlight areas.

[0032] More specifically, if a biological target is determined to exist, the alarm component can send an alarm message to the user's mobile application through the network interface according to the alarm information.

[0033] The intelligent alarm method based on infrared thermal imaging technology in the embodiments of the present application obtains the cross-frame high-temperature area tracking trajectory by temporally associating high-temperature areas, sequentially tracks and analyzes to obtain the temporal change parameter information of the high-temperature area, and distinguishes biological targets and sunlight targets by analyzing the temporal change parameter information of the high-temperature area. It has the ability to accurately distinguish biological targets and sunlight targets, effectively improves the reliability and accuracy of detection, and can effectively reduce the false alarm rate.

[0034] In some preferred embodiments, the temporal change parameter information includes the change rate sequence of the central position, the change rate sequence of the regional area, the change sequence of the shape parameter, the change rate sequence of the regional average temperature, and the change sequence of the internal temperature distribution of the region.

[0035] Specifically, the change rate sequence of the central position reflects the moving speed and direction of the high-temperature area; the change rate sequence of the regional area reflects the change in the size of the high-temperature area; the change sequence of the shape parameter reflects the change in the shape of the high-temperature area; the change rate sequence of the regional average temperature reflects the change trend of the average temperature of the high-temperature area; and the change sequence of the internal temperature distribution of the region reflects the change in the internal heat distribution of the high-temperature area.

[0036] More specifically, in step S5, by analyzing the multi-dimensional temporal change parameter information, the dynamic characteristics of the high-temperature region can be characterized. There are differences between the change patterns of the movement, size, shape, temperature, and internal heat distribution of biological targets and those of the drift, size, shape, temperature, and internal heat distribution of the sunlight region. Based on these differences, each high-temperature region can be distinguished as a biological target or a sunlight target. Through this method of multi-parameter threshold determination, biological targets with mobility and relatively stable body temperature can be effectively distinguished from sunlight regions that change with light and move slowly, improving the accuracy of distinguishing biological targets in an environment with sunlight interference and reducing the false alarm rate.

[0037] In some preferred embodiments, step S4 includes: Traverse each high-temperature region tracking trajectory and perform the following operations on the high-temperature region image sequence: S41. Extract the high-temperature regions corresponding to the currently selected high-temperature region tracking trajectory from the high-temperature region image sequence to form a high-temperature region set; S42. Extract the central position coordinate set of the high-temperature regions according to the high-temperature region set, calculate the Euclidean distance between the central position coordinates of adjacent frames to obtain the central position change amount sequence, and then perform a first-order difference on the central position change amount sequence to obtain the central position change rate sequence; S43. Count the number of pixels in the high-temperature regions according to the high-temperature region set to obtain the region area sequence, and then perform a first-order difference on the region area sequence to obtain the region area change rate sequence; S44. Calculate and extract the shape parameter sequence of the high-temperature regions according to the high-temperature region set. The shape parameters of the shape parameter sequence include circularity, rectangularity, and ellipticity, and then perform a first-order difference on the shape parameter sequence to obtain the shape parameter change sequence; S45. Extract the temperature values of all pixels in the high-temperature regions according to the high-temperature region set, calculate to obtain the average temperature sequence, and then perform a first-order difference on the average temperature sequence to obtain the region average temperature change rate sequence; S46. Count the number of pixels with each temperature value in the high-temperature regions according to the high-temperature region set to obtain the temperature histogram set, and then calculate the similarity of the temperature histograms between adjacent frames using the Bhattacharyya distance to obtain the change sequence of the internal temperature distribution of the region.

[0038] Specifically, the above processing method traverses each tracked high-temperature region trajectory, collects all high-temperature region images corresponding to this trajectory from the high-temperature region image sequence, and forms a high-temperature region set, providing a basis for subsequent analysis of the time-varying characteristics of this specific high-temperature region.

[0039] More specifically, the process of obtaining the central position change rate sequence is as follows: extract the central position coordinates of the high-temperature region in each frame of the image from the high-temperature region concentration, calculate the Euclidean distance of the central position between adjacent frames to obtain the position change amount sequence. Perform a first-order difference on the position change amount sequence to obtain the central position change rate sequence. This change rate sequence reflects the moving speed and direction of the high-temperature region on the image plane.

[0040] More specifically, the process of obtaining the region area change rate sequence is as follows: count the number of pixels contained in the high-temperature region in each frame of the image from the high-temperature region concentration to obtain the region area sequence. Perform a first-order difference on the region area sequence to obtain the region area change rate sequence. This change rate sequence reflects the change speed of the size of the high-temperature region over time.

[0041] More specifically, the process of obtaining the change sequence of shape parameters is as follows: calculate the shape parameters of the high-temperature region in each frame of the image from the high-temperature region concentration, including circularity, rectangularity, and ellipticity, to obtain the shape parameter sequence. Perform a first-order difference on the shape parameter sequence to obtain the shape parameter change sequence. This change sequence reflects the change speed and degree of the shape of the high-temperature region over time.

[0042] More specifically, the process of obtaining the region average temperature change rate sequence is as follows: extract the temperature values of all pixels within the high-temperature region in each frame of the image from the high-temperature region concentration, calculate the average temperature to obtain the average temperature sequence. Perform a first-order difference on the average temperature sequence to obtain the region average temperature change rate sequence. This change rate sequence reflects the change speed of the overall temperature of the high-temperature region over time.

[0043] More specifically, the process of obtaining the change sequence of the internal temperature distribution of the region is as follows: count the number of pixels with each temperature value within the high-temperature region in each frame of the image from the high-temperature region concentration to obtain a temperature histogram set. Use the Bhattacharyya distance to calculate the similarity of the temperature histograms between adjacent frames to obtain the change sequence of the internal temperature distribution of the region. This change sequence reflects the degree of change in the internal heat distribution of the high-temperature region over time.

[0044] More specifically, the above processing method extracts the temporal change parameter information of the high-temperature region from multiple dimensions such as position, area, shape, average temperature, and internal temperature distribution, and focuses on extracting the change rate or degree of change of these parameters. These dynamic features provide a data basis for subsequent accurate target type discrimination and help reduce the false alarm rate caused by sunlight interference.

[0045] In some preferred embodiments, between step S2 and step S3, the following step is further included: SA. Perform morphological filtering on the high-temperature region image sequence.

[0046] Specifically, after extracting the high-temperature region image sequence through threshold segmentation, there may be small noise points or isolated regions in the image. The edges of the high-temperature region may not be smooth, and there may be holes inside the region. These image quality problems will affect the recognition and association of the high-temperature region by subsequent target tracking algorithms, and may lead to tracking interruption or errors. Therefore, the intelligent alarm method based on infrared thermal imaging technology in the embodiments of the present application introduces morphological filtering processing to optimize the high-temperature region image sequence. Morphological filtering processing is an image processing technology that can act on binary images to remove noise points, smooth region edges, fill holes inside the region, and connect disconnected regions. By applying morphological filtering before target tracking, the morphology of the high-temperature region can be improved, making it more suitable for target tracking. Morphological filtering processing improves the quality of the high-temperature region image sequence, provides more suitable data for subsequent target tracking, reduces tracking errors caused by regional morphology problems, and thus reduces the possibility of false alarms.

[0047] In some preferred embodiments, the morphological filtering processing includes erosion and dilation operations. The erosion operation includes: SA1. Based on a pre-determined set of structuring elements and the first number of iterations, perform an erosion operation on each frame of the high-temperature region image sequence to eliminate small noise points and isolated regions; The dilation operation includes: SA2. Based on a pre-determined set of structuring elements and the second number of iterations, perform a dilation operation on each frame of the eroded high-temperature region image sequence to smooth the region edges and fill the holes inside the region.

[0048] Specifically, the erosion operation uses a pre-determined set of structuring elements and the first number of iterations to process the high-temperature region image. Its function is to shrink the high-temperature region, thereby eliminating small noise points and isolated false high-temperature regions. The dilation operation is performed on the eroded image, using a pre-determined set of structuring elements and the second number of iterations. Its function is to expand the high-temperature region, which can smooth the region edges and fill small holes inside the region. The set of structuring elements and the number of iterations are parameters that control the effects of erosion and dilation. The first number of iterations and the second number of iterations can be set according to usage requirements, and the first number of iterations is preferably equal to or less than the second number of iterations.

[0049] More specifically, after threshold segmentation of the infrared thermal image sequence based on a preset temperature threshold to extract the high-temperature region image sequence, the high-temperature region image sequence may contain small noise points, isolated regions, or uneven region edges, and there may be holes inside the regions. These problems will interfere with subsequent target tracking. First, an erosion operation is performed. Based on a pre-determined set of structuring elements and the first number of iterations, each frame image in the high-temperature region image sequence is processed. The erosion operation shrinks the high-temperature region by removing boundary pixels, thereby effectively eliminating small noise points and isolated regions in the image. Then, a dilation operation is performed. Based on a pre-determined set of structuring elements and the second number of iterations, each frame image in the high-temperature region image sequence after erosion processing is processed. The dilation operation expands the high-temperature region by adding pixels to the boundary, thereby smoothing the region edges and filling the holes inside the regions. By processing with erosion first and then dilation, the noise in the image can be removed, and the shape of the high-temperature region becomes more regular and continuous. After morphological filtering processing of the high-temperature region image sequence, its region boundary is clearer, the interior is more complete, and there is less noise, which provides a more reliable input for subsequent target tracking algorithms and improves the accuracy and stability of tracking. After such processing, the high-temperature regions in the frame images of the high-temperature region image sequence are more suitable for subsequent target tracking.

[0050] In some preferred embodiments, the set of structuring elements includes a square structuring element, a circular structuring element, and a linear structuring element; Step SA1 includes: SA11. For each frame image in the high-temperature region image sequence, perform erosion operations using a square structuring element, a circular structuring element, and a linear structuring element respectively to obtain multiple eroded frame images; SA12. Calculate the edge roughness of each eroded frame image, and select the frame image with the lowest edge roughness as the erosion result of the frame image corresponding to the high-temperature region image sequence to obtain the eroded high-temperature region image sequence. Specifically, using a single type of structural element for erosion operations may not be able to handle noise points and isolated regions of different shapes in the image. To solve this problem, in step SA11 of the intelligent alarm method based on infrared thermal imaging technology in the embodiments of the present application, for each frame of the high-temperature region image sequence, erosion operations are performed using multiple types of structural elements respectively, thereby obtaining multiple different erosion results. Each structural element has a different erosion effect on the image and can produce different responses to image features of different shapes. Then, in step SA12, edge roughness is further introduced as a criterion for selecting the optimal erosion result. Edge roughness reflects the smoothness of the region boundary. In the erosion operation, the ideal effect is to eliminate small noise points and isolated regions while retaining the shape and boundary of the main region. If the erosion operation fails to eliminate noise or overly erodes the main region, it may result in an insufficiently smooth edge. By calculating the edge roughness of the image after erosion using different structural elements and selecting the result with the lowest edge roughness, this solution can determine an erosion result that removes small noise points and isolated regions while keeping the boundary of the main region relatively smooth.

[0051] More specifically, the combination of structural elements includes a square structural element, a circular structural element, and a linear structural element. Each structural element has a different impact on features of different shapes in the image (such as dot noise, thin lines, and sharp corners). The square structural element is sensitive to features in the horizontal and vertical directions, the circular structural element is sensitive to isotropic features, and the linear structural element is sensitive to features in a specific direction. Thus, by using multiple structural elements, different erosion results for the same input image can be obtained, and each result performs differently in eliminating noise of a specific shape.

[0052] More specifically, in step SA12, edge roughness is used as the evaluation criterion. Edge roughness can be obtained by calculating the gradient change of the boundary pixel points of the region or the ratio of the boundary perimeter to the area, etc. An image with low edge roughness usually means that noise points and small structures have been effectively removed, while the boundary of the main region has been well preserved. By comparing the edge roughness of different erosion results, the result with the lowest roughness is selected as the final erosion output. This selection process enables the erosion operation to adaptively select the most suitable structural element or its combined effect according to the specific situation of the image content, thereby improving the effect of eliminating small noise points and isolated regions and providing a more accurate basis for subsequent dilation operations and region analysis.

[0053] In some preferred embodiments, step SA2 includes: SA21. For each frame of the high-temperature region image sequence after erosion, perform dilation operations using a square structural element, a circular structural element, and a linear structural element respectively to obtain multiple dilated frame images; SA22. Calculate the area change rate between the high-temperature area in each expanded frame image and the high-temperature area of ​​the corresponding frame image in the high-temperature area image sequence before corrosion, and select the frame image with the lowest area change rate as the expansion result of the frame image in the corresponding high-temperature area image sequence to obtain the high-temperature area image sequence after morphological filtering processing.

[0054] Specifically, the above processing method aims to solve the problem of how to select the expansion result when the high temperature area after the corrosion is expanded so as to achieve the purpose of smoothing the edges and filling the holes while keeping the original shape and area of ​​the high temperature area as much as possible and avoiding deformation. Step SA21 performs an expansion operation on the image after the corrosion by using a variety of structural elements of different shapes. Different structural elements have different expansion effects on the image area, which provides a variety of possible expansion results. Step SA22 introduces the area change rate as a selection criterion to determine the expansion result. This step calculates the area change rate of the high temperature area obtained after the expansion of each structural element and the original high temperature area before the corrosion. The area change rate reflects the degree of influence of the expansion operation on the area size. Selecting an expansion result with a low area change rate means selecting an expansion method that changes the area of ​​the original area less. Compared with the corrosion processing operation, the application of each structural element will produce a different expansion result image. Comparing the area change rates of the three expansion results and selecting the expanded image with a low change rate as the final expansion result of the frame image can ensure that the expansion operation can minimize the impact on the original size and shape of the high temperature area while smoothing the edges and filling the holes, avoiding excessive expansion or shape distortion, that is, maximally retaining the original size and shape characteristics of the high temperature area. Thus, a high-temperature area image sequence after morphological filtering is obtained, and the high-temperature area in the sequence is closer to the original target, which is beneficial to subsequent analysis and processing, thereby improving the accuracy of the entire alarm method.

[0055] In some preferred embodiments, step S5 comprises: S51, for each high temperature area, based on the pre-established parameter feature library of biological targets and sunlight targets, calculate the similarity score between the time series variation parameter information of each high temperature area and the corresponding parameter features in the parameter feature library of biological targets and sunlight targets, and obtain the similarity score of the high temperature area belonging to the biological target and sunlight target; S52. Determine the target type of each high temperature area according to the similarity scores of the biological targets and the sunlight targets and the preset classification rules.

[0056] Specifically, the parameter features of the pre-established parameter feature library of biological targets and sunlight targets include a center position change rate feature sequence, a regional area change rate feature sequence, a shape parameter change feature sequence, a regional average temperature change rate feature sequence, and a regional internal temperature distribution change feature sequence.

[0057] More specifically, the above processing method utilizes the temporal variation parameter information of the high-temperature region and realizes the determination of the target type by comparing the similarity with the parameter characteristics of known target types. Specifically, the pre-established parameter feature library stores the typical temporal variation laws of biological targets and sunlight targets under different circumstances. For example, biological targets usually exhibit relatively fast and non-linear changes in moving speed, fluctuations in regional area and shape, and relatively stable or slow changes in temperature, while sunlight targets may show slow and linear drifts, slow changes in area and shape, and obvious changes in temperature with the light intensity. Calculating the similarity score is a process of quantitatively comparing the temporal variation information of the current high-temperature region with these typical characteristics. For example, methods such as Euclidean distance, cosine similarity, or correlation coefficient can be used to calculate the similarity between each temporal variation parameter sequence of the current high-temperature region and the corresponding feature sequence in the feature library. These similarity scores reflect the matching degree between the current high-temperature region and the characteristics of biological targets and sunlight targets. Subsequently, according to the preset classification rules, such as setting thresholds based on the difference or ratio of similarity scores, or using classifiers such as support vector machines and neural networks, the final type determination of the high-temperature region is carried out. Thus, this method can effectively utilize the dynamic characteristics of the high-temperature region for discrimination, improving the determination accuracy.

[0058] More specifically, specifically, after obtaining the temporal variation parameter information of the high-temperature region, for each tracked high-temperature region, the sequences of the central position change rate, regional area change rate, shape parameter change, regional average temperature change rate, and regional internal temperature distribution change of the high-temperature region are respectively compared with the corresponding feature sequences in the pre-established biological target parameter feature library and sunlight target parameter feature library. For example, the similarity score between the sequence of the central position change rate of the current high-temperature region and the central position change rate feature sequence of the biological target, as well as the similarity score between the sequence of the central position change rate of the current high-temperature region and the central position change rate feature sequence of the sunlight target, can be calculated. Similar similarity calculations are performed for all temporal variation parameter information, so as to obtain the overall similarity score of the high-temperature region with the characteristics of the biological target (which can be the average value or weighted average of the similarity scores corresponding to all sequences) and the overall similarity score of the high-temperature region with the characteristics of the sunlight target. These similarity scores comprehensively reflect the matching degree between the dynamic characteristics of the high-temperature region in multiple dimensions and the known target types. Finally, based on these similarity scores, the preset classification rules are applied for determination. For example, if the similarity score with the characteristics of the biological target is significantly higher than the similarity score with the characteristics of the sunlight target, it is determined as a biological target; conversely, if the similarity score with the characteristics of the sunlight target is significantly higher than the similarity score with the characteristics of the biological target, it is determined as a sunlight target. Through this determination method based on multi-dimensional temporal feature similarity comparison, it is possible to more robustly cope with the complex and changeable characteristics of sunlight targets, thereby effectively distinguishing biological targets from sunlight targets and reducing the false alarm rate.

[0059] In some preferred embodiments, step S52 includes: S521. Calculate the difference between the biological target similarity score and the sunlight target similarity score for each high-temperature region to obtain a similarity difference; S522. Determine the target type of each high-temperature region according to the similarity difference and a preset classification rule. The classification rule is as follows: SB1. Compare the similarity difference with a first preset difference threshold and a second preset difference threshold. The first preset difference threshold is greater than the second preset difference threshold; SB2. If the similarity difference is greater than the first preset difference threshold, determine that the corresponding high-temperature region is a biological target. If the similarity difference is less than the second preset difference threshold, determine that the corresponding high-temperature region is a sunlight target. If the similarity difference is between the second preset difference threshold and the first preset difference threshold, execute step SB3; SB3. Obtain the average moving speed of the center position of the corresponding high-temperature region according to the high-temperature region image sequence, compare the average moving speed with a preset speed threshold. If the average moving speed is greater than the preset speed threshold, determine that the corresponding high-temperature region is a biological target; otherwise, determine it as a sunlight target.

[0060] Specifically, the above processing method can preliminarily judge the target type by calculating the similarity difference between the biological target similarity score and the sunlight target similarity score. Among them, the similarity difference can be a positive number or a negative number, and the second preset difference threshold is a negative value. When the absolute value of the difference is large enough, it indicates a high classification confidence, and the classification is directly determined according to the positive or negative of the difference. The above classification rule is based on the first preset difference threshold and the second preset difference threshold to determine whether the classification confidence is high enough.

[0061] More specifically, when the similarity difference is between the second preset difference threshold and the first preset difference threshold, it indicates that it is difficult to distinguish the target type of the high-temperature region only by the similarity difference. At this time, the average moving speed of the center position of the high-temperature region is introduced as an auxiliary judgment basis. Biological targets usually have a relatively high moving speed, while the high-temperature regions formed by sunlight have a relatively low moving speed. By comparing the average moving speed with the preset speed threshold, the motion characteristics of the target are used for differentiation. This hierarchical logic combining similarity judgment and motion speed judgment improves the classification accuracy in the case of similar similarity scores and reduces the false alarm rate.

[0062] In some preferred embodiments, the target tracking algorithm includes the KCF algorithm and the Kalman filtering algorithm. Step S3 includes: S31. Initialize the object trackers for each high-temperature area based on the KCF algorithm to use the object trackers to track the objects in the frame images of the high-temperature area images, predict the new positions frame by frame and calculate the confidence scores. If the confidence score is higher than the preset score threshold, update the object tracker corresponding to the high-temperature area and associate the cross-frame trajectory. If the confidence score is lower than the preset score threshold, predict the potential position through the Kalman filter algorithm and re-initialize the object tracker corresponding to the high-temperature area. If the initialization is successful, continue to track the high-temperature area, otherwise end the tracking of the high-temperature area; S32. Generate cross-frame high-temperature area tracking trajectories according to the tracking results of each object tracker.

[0063] Specifically, in step S3, the KCF algorithm and the Kalman filter algorithm are used in combination in the high-temperature area tracking step. The KCF algorithm is used to initialize the object tracker and perform frame-by-frame tracking, predict the position of the object in the new frame and calculate the confidence score of the tracking result. The confidence score reflects the reliability evaluation of the KCF algorithm for the current tracking result. When the confidence score of the KCF algorithm is higher than the preset score threshold, it indicates that the KCF tracking result is reliable. At this time, accept the position predicted by KCF as the position of the object in the new frame, update the model of the KCF tracker (object tracker), and associate the tracking result of the current frame with the historical trajectory. When the confidence score of the KCF algorithm is lower than the preset score threshold, it indicates that there may be problems with KCF tracking, such as changes in the appearance of the object, occlusion, or tracking drift. At this time, the Kalman filter algorithm is introduced. The Kalman filter algorithm is based on the motion model of the object and predicts its potential position in the current frame according to the motion state of the object in the previous frame. Use the position predicted by the Kalman filter to try to re-initialize the KCF tracker. If the KCF tracker can be successfully re-initialized based on the predicted position, it is considered that the object may be located near the predicted position, and the tracking is resumed and continued. If the re-initialization fails, it is considered that the tracking of this high-temperature area has been interrupted or the object has disappeared, and the tracking of this high-temperature area is stopped. Through this way of cooperation between KCF and the Kalman filter, even when the appearance of the object changes or is temporarily occluded, the motion prediction information can be used to assist in tracking recovery, so as to generate more stable and accurate cross-frame high-temperature area tracking trajectories. The accurate tracking trajectory provides reliable data for subsequent extraction of the temporal change parameter information of the high-temperature area, thereby improving the accuracy of distinguishing biological objects from sunlight objects and reducing the false alarm rate.

[0064] In some preferred embodiments, step S2 includes: S21. For each frame image in the infrared thermal image sequence, compare the temperature value of the pixel point with the preset temperature threshold, extract the pixel points with temperatures higher than the preset temperature threshold, and form a binary image; S22. Perform connected component labeling on the binary image to form initial high-temperature regions; S23. Filter out the initial high-temperature regions in the binary image that are smaller than a preset minimum region area, and determine each frame image in the high-temperature region image sequence.

[0065] Specifically, in step S21, the pixel points in the image are divided into two categories: pixel points with a temperature higher than the threshold and pixel points with a temperature not higher than the threshold. The pixel points higher than the threshold are marked, for example, assigned a value of 1, to form a binary image that only contains the contour of the high-temperature region. In step S22, connected component labeling is performed on the binary image to form initial high-temperature regions. This operation identifies the sets of connected high-temperature pixel points in the binary image and groups them into independent regions. Each connected region is regarded as a potential high-temperature entity. In step S23, the initial high-temperature regions in the binary image that are smaller than the preset minimum region area are filtered out to determine each frame image in the high-temperature region image sequence. This operation removes those connected regions with too small an area by setting an area lower limit. These small regions are usually sensor noise or tiny and unimportant heat sources. By removing them, a cleaner and more accurate high-temperature region image sequence can be obtained. Each remaining region represents an independent high-temperature entity with a certain area. Thus, the extracted high-temperature region image sequence is clearer, reducing the interference of noise and irregular regions, providing a more reliable basis for subsequent target tracking and extraction of temporal variation parameters, thereby improving the accuracy of distinguishing biological targets from sunlight targets and reducing the false alarm rate.

[0066] In a second aspect, please refer to Figure 4 , some embodiments of the present application further provide an intelligent alarm system based on infrared thermal imaging technology, which is applied in a home environment. The system includes: An infrared detector 201 for collecting an infrared thermal image sequence of the home environment; An alarm component 202 for performing an alarm operation; A control component 203 for performing the intelligent alarm method provided in the first aspect to generate an alarm signal to trigger the alarm component 202 to perform the alarm operation.

[0067] Specifically, the infrared detector 201 is configured to obtain an infrared thermal image sequence of the home environment. The alarm component 202 is configured to perform an alarm action when receiving an alarm signal. The control component 203 is configured to process the infrared thermal image sequence, perform the intelligent alarm method, and generate an alarm signal according to the processing result. The control component 203 is connected to the infrared detector 201 and the alarm component 202, receives image data, and sends a control signal to the alarm component 202. Thus, the system realizes the intelligent alarm function through the collaborative work of these components.

[0068] Specifically, the system is designed to solve the problem of false alarms caused by non-biological heat areas formed by sunlight passing through windows in the home environment. The infrared detector 201 continuously collects the infrared thermal image sequence of the home environment and inputs this image data into the control component 203. The control component 203 executes an intelligent alarm method internally. This method first processes the input infrared thermal image sequence, such as extracting the high-temperature areas in the image. Subsequently, it tracks these high-temperature areas to form cross-frame tracking trajectories. Based on the tracking trajectories and the image information of the high-temperature areas, it calculates the time-series change parameter information of each high-temperature area. Using this time-series change parameter information, the control component 203 distinguishes whether the high-temperature area is a biological target or a sunlight target. Only when it is recognized that the high-temperature area belongs to a biological target, the control component 203 generates an alarm signal. This alarm signal is sent to the alarm component 202 to trigger the alarm component 202 to perform an alarm operation.

[0069] Through analyzing the time-series change characteristics of the high-temperature areas, the intelligent alarm system according to the embodiment of the present application can identify sunlight targets with complex change rules, avoid misjudging them as intruders, and thus reduce the false alarm rate caused by sunlight interference.

[0070] In this article, relational terms such as first and second are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any actual relationship or order between these entities or operations.

[0071] The above description is only for the embodiments of the present application and is not intended to limit the protection scope of the present application. For those skilled in the art, various changes and modifications can be made to the present application. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present application shall be included in the protection scope of the present application.

Claims

1. An intelligent alarm method based on infrared thermal imaging technology, applied in a home environment, characterized in that, The method includes the following steps: S1. Collect an infrared thermal image sequence of the home environment based on an infrared detector; S2. Perform threshold segmentation on each frame image in the infrared thermal image sequence based on a preset temperature threshold to extract a high-temperature region image sequence; S3. Between consecutive frames of the high-temperature region image sequence, use an object tracking algorithm to associate the high-temperature regions between each frame image of the high-temperature region image sequence to form a cross-frame high-temperature region tracking trajectory; S4. Obtain the time-series change parameter information of each high-temperature region according to the high-temperature region tracking trajectory and the high-temperature region image sequence; S5. Distinguish the target types of each high-temperature region according to the time-series change parameter information, and the target type is a biological target or a sunlight target; S6. If there is a high-temperature region with a biological target type, generate an alarm signal.

2. The intelligent alarm method based on infrared thermal imaging technology according to claim 1, wherein The time-series change parameter information includes a sequence of change rates of the central position, a sequence of change rates of the region area, a sequence of change of shape parameters, a sequence of change rates of the regional average temperature, and a sequence of change of the internal temperature distribution of the region.

3. The intelligent alarm method based on infrared thermal imaging technology according to claim 1, wherein, Between step S2 and step S3, there is also a step: SA. Perform morphological filtering processing on the high-temperature region image sequence.

4. The intelligent alarm method based on infrared thermal imaging technology according to claim 3, wherein The morphological filtering processing includes erosion and dilation operations, and the erosion operation includes: SA1. Based on a pre-determined set of structuring elements and a first number of iterations, perform an erosion operation on each frame image in the high-temperature region image sequence to eliminate small noise points and isolated regions; The dilation operation includes: SA2. Based on a pre-determined set of structuring elements and a second number of iterations, perform a dilation operation on each frame image in the eroded high-temperature region image sequence to smooth the region edges and fill the holes inside the region.

5. The intelligent alarm method based on infrared thermal imaging technology according to claim 4, wherein The set of structuring elements includes a square structuring element, a circular structuring element, and a linear structuring element. Step SA1 includes: SA11. For each frame image in the high-temperature region image sequence, perform an erosion operation using a square structuring element, a circular structuring element, and a linear structuring element respectively to obtain a plurality of eroded frame images; SA12. Calculate the edge roughness of each eroded frame image, and select the frame image with the lowest edge roughness as the erosion result of the frame image in the corresponding high-temperature region image sequence to obtain the eroded high-temperature region image sequence.

6. The intelligent alarm method based on infrared thermal imaging technology according to claim 5, wherein Step SA2 includes: SA21. For each frame image in the eroded high-temperature region image sequence, perform a dilation operation using a square structuring element, a circular structuring element, and a linear structuring element respectively to obtain a plurality of dilated frame images; SA22. Calculate the area change rate between the high-temperature region in each dilated frame image and the high-temperature region of the corresponding frame image in the high-temperature region image sequence before erosion, and select the frame image with the lowest area change rate as the dilation result of the frame image in the corresponding high-temperature region image sequence to obtain the high-temperature region image sequence after morphological filtering processing.

7. An intelligent alarm method based on infrared thermal imaging technology according to claim 2, characterized in that, Step S5 includes: S51. For each high-temperature area, based on the pre-established parameter feature library of biological targets and sunlight targets, calculate the similarity score between the parameter information of the temporal variation of each high-temperature area and the corresponding parameter features in the parameter feature library of biological targets and sunlight targets, and obtain the similarity score of the high-temperature area belonging to biological targets and sunlight targets. S52. Determine the target type of each high-temperature area according to the similarity scores of biological targets and sunlight targets and the preset classification rules.

8. An intelligent alarm method based on infrared thermal imaging technology according to claim 1, characterized in that, The target tracking algorithm includes the KCF algorithm and the Kalman filter algorithm. Step S3 includes: S31. Initialize the target tracker for each high-temperature area based on the KCF algorithm, use the target tracker to track the target in the frame image of the high-temperature area image, predict the new position frame by frame and calculate the confidence score. If the confidence score is higher than the preset score threshold, update the target tracker of the corresponding high-temperature area and associate the cross-frame trajectory. If the confidence score is lower than the preset score threshold, predict the potential position through the Kalman filter algorithm and re-initialize the target tracker of the corresponding high-temperature area. If the initialization is successful, continue to track the high-temperature area, otherwise end the tracking of the high-temperature area. S32. Generate cross-frame high-temperature area tracking trajectories according to the tracking results of each target tracker.

9. The intelligent alarm method based on infrared thermal imaging technology according to claim 1, characterized in that Step S2 includes: S21. For each frame image in the infrared thermal image sequence, compare the temperature value of the pixel point with the preset temperature threshold, extract the pixel points with temperatures higher than the preset temperature threshold, and form a binary image. S22. Perform connected component division on the binary image to form initial high-temperature areas. S23. Filter out the initial high-temperature areas in the binary image that are smaller than the preset minimum area, and determine each frame image in the high-temperature area image sequence.

10. An intelligent alarm system based on infrared thermal imaging technology, which is applied in a home environment, is characterized in that, The system includes: An infrared detector for collecting an infrared thermal image sequence of the home environment. An alarm component for performing an alarm operation. A control component for executing the intelligent alarm method based on infrared thermal imaging technology according to any one of claims 1-9 to generate an alarm signal to trigger the alarm component to perform an alarm operation.

Images