Video-based earthquake intensity calculation method, device and equipment

Abstract

The present invention provides a video-based earthquake intensity calculation method, device, and equipment. The method comprises: obtaining surveillance video of a target area where an earthquake occurred during a target time period before and after the earthquake, and calibrating a target object in the surveillance video; tracking the target object in the surveillance video and outputting the tracking result; if the tracking result indicates that the target object is lost in the surveillance video, recording the image of the target object in the surveillance video at the time of loss as the target image, performing full-image feature extraction on the target image, and determining a suspected target object; recording the image of the target object in the surveillance video before the target object was lost as a historical image, and calculating the motion prediction area of ​​the target object at the time of loss based on the historical image; re-determining the target object based on the suspected target object and the motion prediction area; and calculating the earthquake intensity of the target area based on the historical image and the target image after the target object was re-determined. The present invention can promptly and accurately calculate earthquake intensity.

Description

Technical Field

[0001] The present invention relates to the technical field of electronic digital data processing, and in particular to a video-based earthquake intensity calculation method, device and equipment. Background Art

[0002] Rapid determination of earthquake intensity is crucial for disaster relief efforts in small and medium-sized earthquakes. However, based on current technology, earthquake magnitude assessment after an earthquake is a complex, labor-intensive, and time-consuming task.

[0003] Existing assessment methods are mostly qualitative and lack scientific objectivity. Furthermore, existing earthquake intensity assessments are often based on microseismometers distributed across various regions. However, microseismometer signals rely on conventional communication methods and cannot be transmitted in a timely manner. Furthermore, these measurements are not accurate enough for earthquakes of relatively low magnitude. Summary of the Invention

[0004] The embodiments of the present invention provide a video-based earthquake intensity calculation method, apparatus, and device to solve the problem of inaccurate earthquake intensity calculation in the prior art.

[0005] In a first aspect, an embodiment of the present invention provides a video-based earthquake intensity calculation method, comprising:

[0006] Obtain surveillance video of the target area where the earthquake occurred in the target time period before and after the earthquake, and calibrate the target object in the surveillance video.

[0007] Track the target object in the surveillance video and output the tracking results.

[0008] If the tracking result indicates that the target object is lost in the surveillance video, the image in the surveillance video when the target object is lost is recorded as the target image, and full-image feature extraction is performed on the target image to determine the suspected target object.

[0009] The images before the target object is lost in the surveillance video are recorded as historical images, and the motion prediction area of ​​the target object when it is lost is calculated based on the historical images.

[0010] Based on the suspected target object and the motion prediction area, the target object in the target image is re-determined.

[0011] The seismic intensity of the target area where the earthquake occurred is calculated based on the historical image and the target image after the target object is re-determined.

[0012] In one possible implementation, performing full-image feature extraction on a target image to determine a suspected target object includes:

[0013] Perform full-image feature extraction on the target image to obtain the features of each object in the target image, which are recorded as the first features.

[0014] A first feature that is most similar to a feature of a target object is selected from each first feature, and an object corresponding to the most similar first feature is determined as a suspected target object.

[0015] In one possible implementation, calculating the motion prediction area of ​​the target object when it is lost based on historical images includes:

[0016] For each image in the historical images, a first area is determined based on the target object in the image and a preset range.

[0017] According to all the first regions, a motion prediction region of the target object when it is lost is obtained.

[0018] In one possible implementation, re-determining the target object in the target image based on the suspected target object and the motion prediction region includes:

[0019] Based on all the first regions, the occurrence probability of each type of first region is determined.

[0020] Based on the first areas of each type and the corresponding occurrence probabilities, a target area is determined.

[0021] The suspected target object in the target area is re-determined as the target object in the target image.

[0022] In one possible implementation, calculating the seismic intensity of the target area where the earthquake occurred based on the historical image and the target image after the target object is re-determined includes:

[0023] A peak ground acceleration and a peak ground velocity are calculated based on the target object in the historical image and the target object in the target image after the target object is re-determined.

[0024] Based on the peak ground acceleration and peak ground velocity, the seismic intensity of the target earthquake area is calculated.

[0025] In one possible implementation, the method further includes:

[0026] If the tracking result indicates that the target object is not lost in the surveillance video, the peak ground acceleration and the peak ground velocity are calculated based on the target object in the surveillance video.

[0027] Based on the peak ground acceleration and peak ground velocity, the seismic intensity of the target earthquake area is calculated.

[0028] In one possible implementation, tracking a target object in a surveillance video and outputting a tracking result include:

[0029] The target object is tracked in the surveillance video through the kernel correlation filter or the channel and space reliability tracker to obtain the tracking result.

[0030] If the tracking result indicates that the target object is lost in the surveillance video, the time when the target object is lost is output.

[0031] In one possible implementation, obtaining surveillance video of a target area where an earthquake occurs during a target time period before and after the earthquake, and calibrating target objects in the surveillance video, includes:

[0032] Obtain surveillance video of the target area where the earthquake occurred in the target time period before and after the earthquake, and extract the first video frame image in the surveillance video, which is recorded as the starting video frame image.

[0033] Get the features of all objects in the starting video frame image.

[0034] For each object, the feature similarity between the object and the rest of the objects is calculated, and the sum of all feature similarities between the object and the rest of the objects is calculated to obtain the feature index corresponding to the object.

[0035] The object corresponding to the minimum value of the feature index is marked as the target object in the video.

[0036] In a second aspect, an embodiment of the present invention provides a video-based earthquake intensity calculation device, comprising:

[0037] The first processing module is used to obtain monitoring videos of the target area where the earthquake occurs in a target time period before and after the earthquake, and to calibrate target objects in the monitoring videos.

[0038] The second processing module is used to track the target object in the monitoring video and output the tracking result.

[0039] The third processing module is used to record the image of the target object in the surveillance video when the target object is lost as the target image if the tracking result indicates that the target object is lost in the surveillance video, and perform full-image feature extraction on the target image to determine the suspected target object.

[0040] The fourth processing module is used to record the images before the target object is lost in the surveillance video as historical images, and calculate the motion prediction area of ​​the target object when it is lost based on the historical images.

[0041] The fifth processing module is used to re-determine the target object in the target image based on the suspected target object and the motion prediction area.

[0042] The sixth processing module is used to calculate the seismic intensity of the target area where the earthquake occurred based on the historical image and the target image after the target object is re-determined.

[0043] In a third aspect, an embodiment of the present invention provides an electronic device comprising a memory and a processor, wherein the memory stores a computer program, and when the processor executes the computer program, the method in the first aspect or any possible implementation of the first aspect is implemented.

[0044] In an embodiment of the present invention, a reference object is determined by calibrating the target object in a video of the target area during an earthquake. The target object is then tracked using a target tracking algorithm to obtain tracking results. Based on the different tracking results, earthquake intensity calculation schemes for different scenarios are designed. The principle is that when the earthquake intensity is low, the target object will not be lost in the tracking results of the target tracking algorithm; when the earthquake intensity is high, the target object may be lost in the tracking results of the target tracking algorithm. Furthermore, during an earthquake, the ground vibration itself may cause other objects similar to the target object to appear in the surveillance video, resulting in errors in the results calculated by a simple target tracking algorithm. Similarly, because earthquake vibrations are irregular (in terms of direction), traditional methods for predicting the location of target objects based on historical data have excessive deviations and cannot meet the requirements for earthquake intensity calculation accuracy. Therefore, a method is designed to re-determine the target object by combining suspected target objects with motion prediction areas to ensure the accuracy of the re-determined target object and to ensure that the earthquake intensity of the target earthquake area can be accurately calculated even when the earthquake intensity is high. BRIEF DESCRIPTION OF THE DRAWINGS

[0045] Figure 1 This is a flowchart of an implementation method for calculating earthquake intensity based on video provided by an embodiment of the present invention;

[0046] Figure 2 1 is a schematic structural diagram of a video-based earthquake intensity calculation device provided by an embodiment of the present invention;

[0047] Figure 3 is a schematic diagram of an electronic device provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0048] The embodiments of the present invention will be described in detail below with reference to the accompanying drawings.

[0049] See also Figure 1 , which shows a flowchart of the implementation of the video-based earthquake intensity calculation method provided by an embodiment of the present invention, and is detailed as follows:

[0050] Step 101: Obtain surveillance video of a target area where an earthquake occurs during a target time period before and after the earthquake, and calibrate target objects in the surveillance video.

[0051] In one possible implementation, step 101 may include:

[0052] Obtain surveillance video of the target area where the earthquake occurred in the target time period before and after the earthquake, and extract the first video frame image in the surveillance video, which is recorded as the starting video frame image.

[0053] Get the features of all objects in the starting video frame image.

[0054] For each object, the feature similarity between the object and the rest of the objects is calculated, and the sum of all feature similarities between the object and the rest of the objects is calculated to obtain the feature index corresponding to the object.

[0055] The object corresponding to the minimum value of the feature index is marked as the target object in the video.

[0056] For example, in order to ensure that the target object is an easily detectable object and to ensure the accuracy of target tracking, an object that is least similar to other objects is selected in the starting video frame image, that is, the object corresponding to the minimum value of the feature index.

[0057] Step 102: Track the target object in the surveillance video and output the tracking result.

[0058] In one possible implementation, step 102 may include:

[0059] The target object is tracked in the surveillance video through the kernelized correlation filter (KCF) or the channel and spatial reliability tracker (CSRT) to obtain the tracking result.

[0060] If the tracking result indicates that the target object is lost in the surveillance video, the time when the target object is lost is output.

[0061] For example, the process of tracking a target object in a surveillance video using a kernel correlation filter and obtaining a tracking result may be as follows:

[0062] 1. Initialization:

[0063] Extract the initial position and appearance features (such as grayscale, HOG features) of the target object.

[0064] Construct target template and background template, and train the initial classifier through correlation filter.

[0065] 2. Tracking process:

[0066] Search Area: Set the search area centered on the target position in the next frame.

[0067] Feature extraction: Extract the same features as those used for initialization (such as HOG) from the search area.

[0068] Correlation operation: Use Fourier transform to calculate the correlation response between the target and the search area, and find the position of the maximum response as the target position.

[0069] Update filter: Update the filter according to the characteristics of the new position to adapt to the change of target position.

[0070] 3. Optimization and robustness enhancement:

[0071] Multi-scale estimation: Estimate the object scale variation by scaling the search region.

[0072] Re-detection mechanism: If tracking fails (e.g. the response value is too low), re-initialize and output the tracking result of the tracking failure.

[0073] Although this solution is simple (the algorithm structure is clear and easy to implement) and efficient (based on Fourier transform and GPU acceleration, suitable for real-time monitoring), it is sensitive to occlusion and rapid motion (it is easy to fail when the target is occluded or the appearance changes drastically) and relies on initial features (if the initial features are not selected properly, the tracking effect will degrade when the lighting changes).

[0074] For example, the target object is tracked in the surveillance video by using the channel and space reliability tracker, and the process of obtaining the tracking result may be specifically as follows:

[0075] 1. Feature extraction:

[0076] Extract multi-channel features of the target object (such as color histogram, HOG, depth map, etc.).

[0077] 2. Reliability calculation:

[0078] Channel reliability: Calculate the confidence of each feature channel (such as variance, gradient energy).

[0079] Spatial reliability: Calculate the weight of each pixel (such as Gaussian distribution) based on the position of the target object and the similarity with the background.

[0080] 3. Fusion and tracking:

[0081] The features are weighted and fused to construct a joint feature vector.

[0082] Use classifiers (such as SVM, random forest) or regression models (such as Kalman filter) to predict the target object.

[0083] 4. Update and retest:

[0084] Dynamically adjust reliability weights based on new observations.

[0085] If tracking fails (e.g. confidence is too low), re-detection is triggered and the tracking result of the tracking failure is output.

[0086] This solution has good adaptability to occlusion, illumination changes and background interference, and can be combined with multiple feature channels to adapt to complex scenes. It can make up for the shortcomings of the kernel correlation filter. Therefore, in the solution of this application, the target can also be tracked simultaneously by two solutions. When the tracking of the two solutions fails at the same time, the output tracking result is defined as the target object being lost in the monitoring video. This combination can be more suitable for earthquake intensity calculation, because earthquakes are uncertain, and different earthquake intensities will also lead to obvious changes in the environment. Therefore, it is easy to fail to calculate the earthquake intensity by only using the same tracking detection method.

[0087] In step 103 , if the tracking result indicates that the target object is lost in the surveillance video, the image in the surveillance video when the target object is lost is recorded as the target image, and full-image feature extraction is performed on the target image to determine a suspected target object.

[0088] In one possible implementation, performing full-image feature extraction on a target image to determine a suspected target object includes:

[0089] Perform full-image feature extraction on the target image to obtain the features of each object in the target image, which are recorded as the first features.

[0090] A first feature that is most similar to a feature of a target object is selected from each first feature, and an object corresponding to the most similar first feature is determined as a suspected target object.

[0091] Exemplarily, the features of the target object may be recorded simultaneously after the target object is selected, and used as a label when determining the first feature.

[0092] Step 104 : Record the images before the target object is lost in the surveillance video as historical images, and calculate the motion prediction area of ​​the target object when it is lost based on the historical images.

[0093] In one possible implementation, calculating the motion prediction area of ​​the target object when it is lost based on historical images includes:

[0094] For each image in the historical images, a first area is determined based on the target object in the image and a preset range.

[0095] According to all the first regions, a motion prediction region of the target object when it is lost is obtained.

[0096] For example, based on historical images, the approximate movement trajectory of the target object during the earthquake can be obtained. Combined with the preset range, it can ensure that a rough search range can be determined when the target object is lost, giving the range with the highest probability of successful search for the target object for re-determining the target object, thereby improving computing efficiency.

[0097] Step 105 : re-determine the target object in the target image based on the suspected target object and the motion prediction area.

[0098] In one possible implementation, step 105 may include:

[0099] Based on all the first regions, the occurrence probability of each type of first region is determined.

[0100] Based on the first areas of each type and the corresponding occurrence probabilities, a target area is determined.

[0101] The suspected target object in the target area is re-determined as the target object in the target image.

[0102] For example, the type of the first region refers to first regions at different locations formed with the target object at different locations as the center and a preset size as the radius. If the overlap between two first regions is greater than 95%, the two first regions can be considered to be of the same type.

[0103] Exemplarily, the specific process of determining the occurrence probability of each type of first region may include:

[0104] First, determine the total number N of all first areas. Assume that there are three types of first areas, A, B, and C. The number of first areas of type A is a, the number of first areas of type B is b, and the number of first areas of type C is c. Then the probability of occurrence of the first area of ​​type A is a / N, the probability of occurrence of the first area of ​​type B is b / N, and the probability of occurrence of the first area of ​​type C is c / N.

[0105] Exemplarily, the process of determining the target area may include:

[0106] For each type of first area, a second area is formed with the center of the first area as the center and the product of a preset size and the probability of occurrence of the first area of ​​the type as the radius;

[0107] The target region can be obtained by taking the union of all second regions.

[0108] Step 106 , calculating the seismic intensity of the target area where the earthquake occurred based on the historical image and the target image after the target object is re-determined.

[0109] In one possible implementation, step 106 may include:

[0110] A peak ground acceleration and a peak ground velocity are calculated based on the target object in the historical image and the target object in the target image after the target object is re-determined.

[0111] Based on the peak ground acceleration and peak ground velocity, the seismic intensity of the target earthquake area is calculated.

[0112] In one possible implementation, the method further includes:

[0113] If the tracking result indicates that the target object is not lost in the surveillance video, the peak ground acceleration and the peak ground velocity are calculated based on the target object in the surveillance video.

[0114] Based on the peak ground acceleration and peak ground velocity, the seismic intensity of the target earthquake area is calculated.

[0115] For example, the direction and displacement of the target object can be obtained by analyzing the position of the target object between multiple consecutive images, and the speed and acceleration of the target object can be calculated by combining the interval time between the multiple consecutive images.

[0116] After calculating the peak ground acceleration and the peak ground velocity, the seismic intensity of the target earthquake area can be calculated by combining the first formula and the second formula.

[0117] The first formula can be:

[0118]

[0119] The second formula can be:

[0120]

[0121] in, Indicates the earthquake intensity in the target earthquake area.

[0122] The video-based earthquake intensity calculation method establishes a reference by locating a target object in the video footage of the target area during an earthquake. The target object is then tracked using a target tracking algorithm to obtain tracking results. Seismic intensity calculation schemes for different scenarios are designed based on different tracking results. The principle is that when the earthquake intensity is low, the target object will not be lost in the tracking results of the target tracking algorithm; when the earthquake intensity is high, the target object may be lost in the tracking results of the target tracking algorithm. Furthermore, during an earthquake, the ground's inherent vibrations may cause other objects similar to the target object to appear in the surveillance video, resulting in errors in the results calculated by a simple target tracking algorithm. Similarly, because earthquake vibrations are irregular (in terms of direction), traditional methods that predict the location of target objects based on historical data have excessive bias and cannot meet the required accuracy for earthquake intensity calculation. Therefore, a method has been designed to re-identify the target object by combining suspected target objects with motion prediction areas. This ensures the accuracy of the re-identified target object and ensures accurate calculation of the seismic intensity in the target earthquake area even when the earthquake intensity is high.

[0123] It should be understood that the size of the serial numbers of the steps in the above embodiments does not mean the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of the present invention.

[0124] The following are device embodiments of the present invention. For details not fully described therein, reference may be made to the corresponding method embodiments described above.

[0125] Figure 2 The following is a schematic diagram of the structure of a video-based earthquake intensity calculation device provided by an embodiment of the present invention. For ease of explanation, only the parts related to the embodiment of the present invention are shown, which are detailed as follows:

[0126] like Figure 2 As shown, the video-based earthquake intensity calculation device includes:

[0127] The first processing module 201 is used to obtain surveillance video of a target area where an earthquake occurs in a target time period before and after the earthquake, and to calibrate target objects in the surveillance video.

[0128] The second processing module 202 is used to track the target object in the surveillance video and output the tracking result.

[0129] The third processing module 203 is configured to record the image in the surveillance video when the target object is lost as the target image if the tracking result indicates that the target object is lost in the surveillance video, and perform full-image feature extraction on the target image to determine a suspected target object.

[0130] The fourth processing module 204 is configured to record images before the target object is lost in the surveillance video as historical images, and calculate a motion prediction area of ​​the target object when it is lost based on the historical images.

[0131] The fifth processing module 205 is configured to re-determine the target object in the target image based on the suspected target object and the motion prediction area.

[0132] The sixth processing module 206 is configured to calculate the seismic intensity of the target area where the earthquake occurred based on the historical image and the target image after the target object is re-determined.

[0133] In a possible implementation, the third processing module 203 may be configured to:

[0134] Perform full-image feature extraction on the target image to obtain the features of each object in the target image, which are recorded as the first features.

[0135] A first feature that is most similar to a feature of a target object is selected from each first feature, and an object corresponding to the most similar first feature is determined as a suspected target object.

[0136] In a possible implementation, the fourth processing module 204 may be configured to:

[0137] For each image in the historical images, a first area is determined based on the target object in the image and a preset range.

[0138] According to all the first regions, a motion prediction region of the target object when it is lost is obtained.

[0139] In a possible implementation, the fifth processing module 205 may be configured to:

[0140] Based on all the first regions, the occurrence probability of each type of first region is determined.

[0141] Based on the first areas of each type and the corresponding occurrence probabilities, a target area is determined.

[0142] The suspected target object in the target area is re-determined as the target object in the target image.

[0143] In a possible implementation, the sixth processing module 206 may be configured to:

[0144] A peak ground acceleration and a peak ground velocity are calculated based on the target object in the historical image and the target object in the target image after the target object is re-determined.

[0145] Based on the peak ground acceleration and peak ground velocity, the seismic intensity of the target earthquake area is calculated.

[0146] In a possible implementation, the third processing module 203 may also be configured to:

[0147] If the tracking result indicates that the target object is not lost in the surveillance video, the peak ground acceleration and the peak ground velocity are calculated based on the target object in the surveillance video.

[0148] Based on the peak ground acceleration and peak ground velocity, the seismic intensity of the target earthquake area is calculated.

[0149] In a possible implementation, the second processing module 202 may be configured to:

[0150] The target object is tracked in the surveillance video through the kernel correlation filter or the channel and space reliability tracker to obtain the tracking result.

[0151] If the tracking result indicates that the target object is lost in the surveillance video, the time when the target object is lost is output.

[0152] In a possible implementation, the first processing module 201 may be configured to:

[0153] Obtain surveillance video of the target area where the earthquake occurred in the target time period before and after the earthquake, and extract the first video frame image in the surveillance video, which is recorded as the starting video frame image.

[0154] Get the features of all objects in the starting video frame image.

[0155] For each object, the feature similarity between the object and the rest of the objects is calculated, and the sum of all feature similarities between the object and the rest of the objects is calculated to obtain the feature index corresponding to the object.

[0156] The object corresponding to the minimum value of the feature index is marked as the target object in the video.

[0157] Figure 3 Schematic diagram of an electronic device provided by an embodiment of the present invention. Figure 3 As shown, the electronic device 5 of this embodiment includes: a processor 50 and a memory 51. The memory 51 stores a computer program 52. When the processor 50 executes the computer program 52, the steps of the above-mentioned method embodiments are implemented. Alternatively, when the processor 50 executes the computer program 52, the functions of the modules / units in the above-mentioned device embodiments are implemented.

[0158] For example, the computer program 52 may be divided into one or more modules / units, which are stored in the memory 51 and executed by the processor 50 to implement the present invention. The one or more modules / units may be a series of computer program instruction segments capable of implementing specific functions, and the instruction segments are used to describe the execution process of the computer program 52 in the electronic device 5.

[0159] The electronic device 5 may include, but is not limited to, a processor 50 and a memory 51. Those skilled in the art will appreciate that Figure 3 It is only an example of the electronic device 5 and does not constitute a limitation of the electronic device 5. It may include more or fewer components than shown in the figure, or a combination of certain components, or different components. For example, the electronic device 5 may also include input and output devices, network access devices, buses, etc.

[0160] The processor 50 may be a central processing unit (CPU), or other general-purpose processors, digital signal processors (DSP), application-specific integrated circuits (ASIC), field-programmable gate arrays (FPGA), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. The general-purpose processor may be a microprocessor or any conventional processor.

[0161] The memory 51 can be an internal storage unit of the electronic device 5, such as the hard drive or memory of the electronic device 5. The memory 51 can also be an external storage device of the electronic device 5, such as a plug-in hard drive, a Smart Media Card (SMC), a Secure Digital (SD) card, a flash memory card, etc. equipped on the electronic device 5. Furthermore, the memory 51 can include both the internal storage unit of the electronic device 5 and an external storage device. The memory 51 is used to store the computer program 52 and other programs and data required by the electronic device 5. The memory 51 can also be used to temporarily store data that has been output or is about to be output.

[0162] For the sake of convenience and brevity, the division of the above functional modules / units is only used as an example. In actual applications, the above functions can be assigned to different functional modules / units as needed. The above modules / units can be implemented in the form of hardware, software, or a combination of hardware and software.

[0163] An embodiment of the present invention further provides a computer-readable storage medium storing a computer program that, when executed by a processor, implements the methods in the above-mentioned method embodiments.

[0164] An embodiment of the present invention further provides a computer program product, including a computer program, which, when executed by a processor, implements the methods in the above-mentioned method embodiments.

[0165] The term "computer program" includes computer program code, which may be in source code form, object code form, executable file, or some intermediate form. Computer-readable media may include any entity or device capable of carrying computer program code, recording media, USB flash drives, removable hard drives, magnetic disks, optical disks, computer memory, read-only memory (ROM), random access memory (RAM), electrical carrier signals, telecommunications signals, and software distribution media.

[0166] In the above embodiments, the descriptions of each embodiment have their own focus. For parts not described or recorded in detail in one embodiment, please refer to the relevant descriptions of other embodiments. Unless otherwise specified or there is a logical conflict, the terms and / or descriptions between different embodiments are consistent and can be referenced to each other. The technical features of different embodiments can be combined to form new embodiments based on their inherent logical relationships.

[0167] The embodiments described above are only used to illustrate the technical solutions of the present invention, rather than to limit the same. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or make equivalent replacements for some of the technical features therein. These modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the various embodiments of the present invention, and should all be included in the scope of protection of the present invention.

Claims

1. A video-based earthquake intensity calculation method, characterized in that: include: Obtaining surveillance video of a target area where an earthquake occurred during a target time period before and after the earthquake, and calibrating target objects in the surveillance video; Tracking the target object in the surveillance video and outputting a tracking result; If the tracking result indicates that the target object is lost in the surveillance video, the image of the surveillance video at the time when the target object is lost is recorded as the target image, and full-image feature extraction is performed on the target image to determine a suspected target object; Recording an image of the target object in the surveillance video before it is lost as a historical image, and calculating a motion prediction area of ​​the target object at the time of loss based on the historical image; re-determining the target object in the target image based on the suspected target object and the motion prediction area; Calculating the seismic intensity of the target area where the earthquake occurred based on the historical image and the target image after the target object is re-determined; The calculating, based on the historical images, the motion prediction area of ​​the target object when it is lost includes: For each of the historical images, determining a first area based on the target object and a preset range on the image; Obtaining a motion prediction region of the target object when it is lost based on all of the first regions; The re-determining the target object in the target image based on the suspected target object and the motion prediction area includes: determining, based on all the first areas, an occurrence probability of each type of first area; determining a target area based on each type of first area and the corresponding occurrence probability; The suspected target object in the target area is re-determined as the target object in the target image.

2. The video-based earthquake intensity calculation method according to claim 1, characterized in that: The step of extracting features of the entire target image to determine a suspected target object includes: Performing full-image feature extraction on the target image to obtain a feature of each object in the target image, which is recorded as a first feature; A first feature that is most similar to the feature of the target object is selected from each of the first features, and an object corresponding to the most similar first feature is determined as a suspected target object.

3. The video-based earthquake intensity calculation method according to claim 1, characterized in that: Calculating the seismic intensity of the target area where the earthquake occurred based on the historical image and the target image after the target object is re-determined includes: Calculating a peak ground acceleration and a peak ground velocity based on the target object in the historical image and the target object in the target image after the target object is re-determined; The seismic intensity of the target earthquake area is calculated based on the peak ground acceleration and the peak ground velocity.

4. The video-based earthquake intensity calculation method according to claim 1, characterized in that: The method further comprises: If the tracking result indicates that the target object is not lost in the surveillance video, calculating a peak ground acceleration and a peak ground velocity based on the target object in the surveillance video; The seismic intensity of the target earthquake area is calculated based on the peak ground acceleration and the peak ground velocity.

5. The video-based earthquake intensity calculation method according to claim 1, characterized in that: Tracking the target object in the surveillance video and outputting the tracking result includes: Tracking the target object in the surveillance video using a kernel correlation filter or a channel and spatial reliability tracker to obtain a tracking result; If the tracking result indicates that the target object is lost in the surveillance video, the time when the target object is lost is output.

6. The video-based earthquake intensity calculation method according to claim 1, characterized in that: The step of obtaining a surveillance video of a target area where an earthquake occurs in a target time period before and after the earthquake, and calibrating a target object in the surveillance video, includes: Obtaining a surveillance video of a target area where an earthquake occurs in a target time period before and after the earthquake, and extracting the first video frame image in the surveillance video, recording it as the starting video frame image; Acquire features of all objects in the starting video frame image; For each object, calculate the feature similarity between the object and the rest of the objects, and calculate the sum of all feature similarities between the object and the rest of the objects to obtain the feature index corresponding to the object; The object corresponding to the minimum value of the feature index is calibrated as the target object in the video.

7. A video-based earthquake intensity calculation device, characterized in that: include: The first processing module is used to obtain a surveillance video of a target area where an earthquake occurs in a target time period before and after the earthquake, and to calibrate a target object in the surveillance video; A second processing module is used to track the target object in the monitoring video and output a tracking result; a third processing module, configured to, if the tracking result indicates that the target object is lost in the surveillance video, record the image of the target object in the surveillance video at the time of loss as a target image, and perform full-image feature extraction on the target image to determine a suspected target object; a fourth processing module, configured to record an image of the target object in the surveillance video before the target object is lost as a historical image, and calculate a motion prediction area of ​​the target object at the time of loss based on the historical image; a fifth processing module, configured to re-determine the target object in the target image based on the suspected target object and the motion prediction area; a sixth processing module, configured to calculate the seismic intensity of the target area where the earthquake occurred based on the historical image and the target image after the target object is re-determined; The fourth processing module is further configured to: For each of the historical images, determining a first area based on the target object and a preset range on the image; Obtaining a motion prediction region of the target object when it is lost based on all of the first regions; The fifth processing module is further configured to: determining, based on all the first areas, an occurrence probability of each type of first area; determining a target area based on each type of first area and the corresponding occurrence probability; The suspected target object in the target area is re-determined as the target object in the target image.

8. An electronic device, characterized in that: The method comprises a memory and a processor, wherein the memory stores a computer program, and when the processor executes the computer program, the method according to any one of claims 1 to 6 is implemented.

Images