Explosion-proof intelligent video monitoring control system

Abstract

The invention discloses an explosion-proof intelligent video monitoring control system, and relates to the technical field of intelligent security and protection, and the system comprises a collection module which is used for collecting multispectral video data, and generating a partition time sequence original data set; the feature extraction module is used for extracting pixel intensity change and constructing a native fluctuation spectrum; the linkage factor construction module is used for constructing a multi-channel disturbance parameter matrix based on the multi-channel native fluctuation spectrum in the same area and extracting a multi-dimensional eddy current spectrum linkage factor; the anomaly judgment module is used for carrying out high-dimensional density peak clustering, identifying an abnormal high condensation region and calculating a spectrum gathering factor; the path modeling module is used for performing time sequence tracking on the abnormally high condensation area and establishing a multi-dimensional diffusion path model; and the risk early warning module is used for setting a multi-level response threshold value and realizing graded risk early warning of the abnormal region. Through multi-channel disturbance modeling, abnormal accumulation area discrimination and diffusion tracking, the ability of sensitive perception and dynamic evolution monitoring of early risks in a complex scene is improved.

Description

Technical Field

[0001] The present invention relates to the field of intelligent security technology, in particular to an explosion-proof intelligent video monitoring and control system. Background Art

[0002] With the continued growth in demand for industrial automation and intelligent security, key explosion-proof sites such as the petrochemical industry, natural gas, coal mining, military industry, and hazardous materials storage are increasingly demanding real-time monitoring and risk warnings. Traditional explosion-proof video surveillance systems, which often rely on single sensors or visible light video monitoring, struggle to dynamically capture multimodal, multi-regional, and rapidly changing situations in complex environments. Their intelligent recognition of subtle disturbances, danger signs, and the early spread of abnormalities is also limited.

[0003] The existing explosion-proof monitoring technology has the following main shortcomings: it only relies on single-channel or low-frame-rate video information, and lacks the ability to three-dimensionally identify multi-source early risk characteristics such as abnormal disturbances, smoke, flames, and gas leaks; it is difficult to effectively analyze the coordinated anomalies and spatial energy agglomeration effects of multiple points and channels within the monitoring area, which can easily lead to missed detections of abnormal events, false alarms and delayed warnings; it is unable to combine multiple parameters such as temporal dynamic characteristics, regional multi-dimensional linkage factors and spatial diffusion paths to conduct graded warnings and response linkages for risks, and lacks closed-loop control and intelligent adjustment mechanisms.

[0004] Therefore, there is an urgent need for an explosion-proof intelligent video surveillance and control system that combines multispectral video data, spatiotemporal feature extraction, multi-channel linkage factor analysis and spatial path dynamic modeling to achieve early identification of abnormally high condensation areas, spatial evolution tracking and graded risk warning, so as to significantly improve the inherent safety level and emergency response capabilities of explosion-proof places. Summary of the Invention

[0005] Based on the above-mentioned shortcomings of the prior art, the purpose of the present invention is to provide an explosion-proof intelligent video surveillance control system to solve the above-mentioned technical problems.

[0006] To achieve the above objectives, the present invention provides the following technical solutions: an explosion-proof intelligent video surveillance control system, comprising: Acquisition module: used to collect multispectral video data in the explosion-proof monitoring area in real time. Based on the preset spatial grid division, multi-channel high-frequency synchronous sampling is performed on the sub-area to generate the partitioned time series raw data set; Feature extraction module: used to extract the pixel intensity changes of consecutive frames within the time window of the partitioned time series raw data set, obtain the light field disturbance characteristics and construct the original fluctuation spectrum through time domain transformation and frequency domain transformation; Linkage factor construction module: used to construct a multi-channel perturbation parameter joint matrix based on the native wave spectra of different channels in the same area, and obtain the multi-dimensional eddy current spectrum linkage factor. The multi-dimensional eddy current spectrum linkage factor includes the vortex density index, the spectrum aggregation flow index and the perturbation transition rate; Anomaly discrimination module: used to perform high-dimensional density peak clustering on the multi-dimensional eddy current spectrum linkage factors of all spatial partitions, identify areas with significant eddy current disturbance aggregation trends, calculate the spectral convergence factor, and identify and mark abnormally high-condensation areas in the space based on the spectral convergence factor; Path modeling module: used to track the time series of abnormally high cohesion areas, extract spatial diffusion speed, displacement direction and duration, and build a multidimensional diffusion path model; Risk warning module: used to set multi-level response thresholds based on the multidimensional diffusion path model to provide risk warnings.

[0007] The present invention is further configured such that the logic for generating the partitioned time series original data set includes: The explosion-proof monitoring area is spatially divided according to the preset spatial grid parameters, and the area is discretized into multiple sub-areas; Explosion-proof intelligent video surveillance is deployed in each sub-area. The explosion-proof intelligent video surveillance is equipped with a multi-spectral video acquisition unit for multi-channel synchronous acquisition control. Through a unified synchronous clock and sampling trigger mechanism, high-frequency data acquisition is performed within the same time window to obtain video frame data containing infrared, visible light, and ultraviolet light; The multi-channel video frames collected from each sub-region at all sampling moments are structured and organized according to the partition number, time series and channel type to generate a partition time series raw data set.

[0008] The present invention is further configured such that the construction logic of the original wave spectrum includes: The partitioned time series original data set is segmented and processed according to the set time window for each partition and each channel's continuous multi-frame video images; For each continuous video frame in a time window, the pixel intensity sequence in the time window is extracted according to the pixel coordinates to generate a time-series pixel intensity matrix; Perform first-order difference on the time-series pixel intensity matrix to obtain the inter-frame pixel intensity change sequence; Extract temporal statistical features of pixel intensity change sequence, including sliding mean, variance, kurtosis and skewness; Perform fast Fourier transform on each pixel intensity change sequence to extract frequency domain statistical features, including main frequency component, energy spectrum distribution, local amplitude peak and instantaneous phase parameters; The time domain statistical features and frequency domain statistical features of all pixels and all channels in each time window are summarized by spatial partition to obtain a partitioned light field disturbance feature group; Taking partition, channel and time window as the ternary index, the partition light field disturbance feature group is combined to generate the native wave spectrum data group.

[0009] The present invention is further configured such that the logic for obtaining the multi-dimensional eddy current spectrum linkage factor includes: For the native fluctuation spectrum data sets of different channels in the same partition and the same time window, they are aligned according to the main disturbance parameters to construct a multi-channel disturbance parameter joint matrix; Based on the energy spectrum distribution, the spatial concentration of disturbance energy in the partition is calculated and the vortex density index is extracted; According to the main frequency component and the local amplitude peak, the energy synchronization change rate and peak energy synthesis intensity of different channels in the main frequency band are counted to extract the spectral aggregation flow index; According to the gradient of the energy spectrum in the energy spectrum distribution over time, the rate of transfer of disturbance energy from low frequency to high frequency is calculated, and the perturbation transition rate is extracted; The vortex density index, spectral aggregation flow index and perturbation transition rate are set as multidimensional vortex spectrum linkage factors.

[0010] The present invention is further configured to align the main disturbance parameters and construct a multi-channel disturbance parameter joint matrix, including: In each channel, the main frequency component, local amplitude peak and instantaneous phase parameters are extracted from the original wave spectrum data set and set as the main disturbance parameters; For the main frequency components of all channels in the same partition and the same time window, a main frequency index set is formed; For the main frequency index set, if there is a slight offset between the main frequency components of different channels, the parameter values ​​of each channel are uniformly mapped to the frequency points of the main frequency index set through linear interpolation or nearest neighbor matching algorithm to ensure that the main disturbance parameters of each channel correspond one to one in the frequency dimension; The aligned main frequency components, local amplitude peaks and instantaneous phase parameters of each channel are arranged according to the main frequency index to form a multi-channel disturbance parameter joint matrix, where each row of the matrix corresponds to a channel and each column corresponds to a disturbance parameter at a main frequency point.

[0011] The present invention is further configured such that the extraction logic of the vortex density index, the spectral aggregation flow index and the perturbation transition rate includes: Based on the energy spectrum distribution of each partition in the original wave spectrum of each channel, a high-frequency energy threshold is set, and high-frequency disturbance energy points with energy greater than the high-frequency energy threshold are screened. The density of high-frequency disturbance energy points in the partition space is calculated and set as the vortex density index; Align the main frequency components of each channel in each partition with the corresponding local amplitude peak, calculate the synchronization change rate of the main frequency amplitude of each channel, and set the synchronization change rate as the spectrum aggregation flow index; According to the energy spectrum distribution of each channel, the main energy frequency trajectory sequence is established, the time change rate of the main energy frequency is calculated, the rate of transfer of disturbance energy from low frequency to high frequency is extracted, and the time change rate is set as the perturbation transition rate.

[0012] The present invention is further configured such that the logic of the abnormality discrimination module includes: Based on the multidimensional eddy spectrum linkage factor of all spatial partitions in the same time window, a multidimensional feature vector with eddy density index, spectral aggregation flow index and perturbation transition rate as components is constructed, and the feature vectors corresponding to each partition are combined into a multidimensional feature space. Clustering the feature vectors of each partition in the multidimensional feature space, calculating the local density of each partition and the distance between each partition and the partition with higher density, identifying the partition with high local density and far away from other high-density partitions as the density peak, and determining a preset number of high-density peak cluster centers; Based on the high-density cluster center, the remaining partitions are assigned to the nearest cluster center according to the feature distance to form a candidate set of high-density regions; For each high-condensation region candidate set, based on the spectral aggregation flow index in the same region, the consistency of energy distribution, peak synthesis intensity and correlation between channels in the region are calculated and set as the spectral convergence factor; The spectral convergence factor threshold is set, and the abnormal high-condensation area is determined for the high-condensation area candidate set whose spectral convergence factor is higher than the corresponding threshold. The area is spatially partitioned and marked, and the abnormal partition index is output.

[0013] The present invention is further configured such that the calculation logic of the consistency of energy distribution, peak composite intensity and correlation between channels within a region is as follows: For the energy data extracted at the main frequency component of all channels in the spatial partition, the mean and standard deviation of the main frequency energy of all channels are obtained, and the ratio of the mean and standard deviation is set as the energy distribution consistency index; Perform weighted summation on the main frequency energies of all channels to obtain the composite intensity of the peak energy within the partition, which is set as the peak composite intensity and used to measure the overall aggregation level of multi-channel energy; For the main frequency energy data of each channel, the correlation coefficient between any two channels is calculated respectively, and the absolute value mean of all correlation coefficients is set as the correlation index of the energy distribution between channels.

[0014] The present invention is further configured such that the construction logic of the multidimensional diffusion path model includes: For the identified abnormally high cohesion areas, time series trajectory tracking is performed within continuous time windows based on the spatial distribution of moment partitions and regional labeling results. Abnormally high cohesion areas with adjacent spatial positions in adjacent time windows are determined to be the same object, generating a continuous spatial trajectory of the abnormal area. For each abnormally high-condensation area, the coordinates of the spatial center point in each time window are extracted. Based on the position change of the center point in the time series, the diffusion speed and main diffusion direction of the abnormal area in space are quantified. The diffusion speed is calculated by the displacement and time interval of the center point in adjacent time windows, and the main diffusion direction is determined by the trend of time series position change. Count the number of continuous time windows that each abnormally high cohesion area experiences from the first identification to its disappearance, and define the duration corresponding to the number of continuous time windows as the duration of the abnormally high cohesion area; A multidimensional diffusion path model was established by integrating the spatial diffusion velocity, displacement direction and duration of the abnormally high condensation area.

[0015] The present invention is further configured to set a multi-level response threshold based on a multi-dimensional diffusion path model to perform risk warning, including: When the spatial diffusion speed is less than the first spatial diffusion speed threshold and the duration is less than the first duration threshold, it is determined to be a low-risk state and an abnormal state record is performed; When the spatial diffusion speed is greater than or equal to the first spatial diffusion speed threshold and less than the second spatial diffusion speed threshold, and the duration is greater than or equal to the first duration threshold and less than the second duration threshold, it is determined to be a medium risk state and a local risk warning is triggered; When the spatial diffusion speed is greater than or equal to the second spatial diffusion speed threshold and the duration is greater than or equal to the second duration threshold, it is determined to be a high-risk state and a remote alarm is issued.

[0016] The present invention provides an explosion-proof intelligent video surveillance control system, comprising an acquisition module for real-time acquisition of multi-spectral video data within an explosion-proof monitoring area, performing multi-channel high-frequency synchronous sampling on sub-areas based on a preset spatial grid division, and generating a partitioned time series original data set; a feature extraction module for extracting the pixel intensity changes of consecutive frames within a time window of the partitioned time series original data set, obtaining light field disturbance features and constructing a native wave spectrum through time domain transformation and frequency domain transformation processing; a linkage factor construction module for constructing a multi-channel disturbance parameter joint matrix based on the native wave spectra of different channels in the same area, obtaining a multi-dimensional eddy current spectrum linkage factor, and a multi-dimensional eddy current spectrum linkage factor. The eddy current spectrum linkage factors include the vortex density index, the spectral aggregation flow index, and the perturbation transition rate. The anomaly discrimination module is used to perform high-dimensional density peak clustering on the multidimensional eddy current spectrum linkage factors of all spatial partitions, identify areas with significant eddy current disturbance aggregation trends, calculate the spectral convergence factor, and identify and mark abnormally high-condensation areas in space based on the spectral convergence factor. The path modeling module is used to track the time series of abnormally high-condensation areas, extract the spatial diffusion speed, displacement direction, and duration, and construct a multidimensional diffusion path model. The risk warning module is used to set multi-level response thresholds based on the multidimensional diffusion path model to issue risk warnings. The beneficial effects produced include: Pixel-level early anomaly identification: Utilizing pixel intensity change analysis within a time window and native fluctuation spectrum construction, it can keenly capture tiny disturbances and early abnormal signals, effectively enhancing the ability to identify early risks in complex scenarios such as smoke, flames, and gas leaks. Multi-channel disturbance coordination modeling: By extracting multi-dimensional linkage factors such as vortex density index, spectral aggregation flow index, and perturbation transition rate, it can quantitatively reflect multi-channel, cross-regional disturbance coordination and spatial energy aggregation phenomena, making up for the shortcomings of traditional systems in multi-source linkage anomaly analysis; Intelligent abnormal clustering area identification: Utilizing high-dimensional density peak clustering and spectral convergence factor statistical methods, it automatically identifies and labels highly concentrated areas of spatial abnormalities, significantly reducing the risk of missed detection and false alarms, and improving the accuracy and intelligence of anomaly detection. Dynamic path and diffusion process tracking: Through multi-dimensional parameter modeling such as trajectory correlation, diffusion speed, displacement direction and duration of abnormally high-condensation areas, the dynamic evolution of abnormal areas can be tracked in real time, providing data support for subsequent risk analysis and event tracing.

[0017] The above description is only an overview of the technical solution of the present application. In order to more clearly understand the technical means of the present application, it can be implemented in accordance with the contents of the specification. In order to make the above and other purposes, features and advantages of the present application more obvious and easy to understand, the specific implementation methods of the present application are listed below. BRIEF DESCRIPTION OF THE DRAWINGS

[0018] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the following briefly introduces the drawings required for describing the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without inventive efforts. In the drawings: Figure 1 The flowchart of an explosion-proof intelligent video surveillance control system is shown as an exemplary embodiment of the present invention. DETAILED DESCRIPTION

[0019] The following describes the embodiments of the present invention with reference to the accompanying drawings and preferred embodiments. Those skilled in the art will readily appreciate the other advantages and benefits of the present invention from the disclosure herein. The present invention may also be implemented or applied through various other specific embodiments, and the various details in this specification may be modified or altered based on different viewpoints and applications without departing from the spirit of the present invention. It should be understood that the preferred embodiments are intended only to illustrate the present invention and are not intended to limit the scope of protection of the present invention.

[0020] It should be noted that the illustrations provided in the following embodiments are merely schematic illustrations of the basic concept of the present invention. Therefore, the illustrations only show components related to the present invention and are not drawn according to the number, shape, and size of components in actual implementation. In actual implementation, the type, quantity, and proportion of each component may be changed arbitrarily, and the component layout may also be more complex.

[0021] In the following description, numerous details are discussed to provide a more thorough explanation of the embodiments of the present invention. However, it will be apparent to those skilled in the art that the embodiments of the present invention may be practiced without these specific details. In other embodiments, well-known structures and devices are shown in block diagram form rather than in detail to avoid obscuring the embodiments of the present invention.

[0022] An explosion-proof intelligent video surveillance control system, such as Figure 1 Shown, including: Acquisition module: used to collect multispectral video data in the explosion-proof monitoring area in real time. Based on the preset spatial grid division, multi-channel high-frequency synchronous sampling is performed on the sub-area to generate the partitioned time series raw data set; Feature extraction module: used to extract the pixel intensity changes of consecutive frames within the time window of the partitioned time series raw data set, obtain the light field disturbance characteristics and construct the original fluctuation spectrum through time domain transformation and frequency domain transformation; Linkage factor construction module: used to construct a multi-channel perturbation parameter joint matrix based on the native wave spectra of different channels in the same area, and obtain the multi-dimensional eddy current spectrum linkage factor. The multi-dimensional eddy current spectrum linkage factor includes the vortex density index, the spectrum aggregation flow index and the perturbation transition rate; Anomaly discrimination module: used to perform high-dimensional density peak clustering on the multi-dimensional eddy current spectrum linkage factors of all spatial partitions, identify areas with significant eddy current disturbance aggregation trends, calculate the spectral convergence factor, and identify and mark abnormally high-condensation areas in the space based on the spectral convergence factor; Path modeling module: used to track the time series of abnormally high cohesion areas, extract spatial diffusion speed, displacement direction and duration, and build a multidimensional diffusion path model; Risk warning module: used to set multi-level response thresholds based on the multidimensional diffusion path model to provide risk warnings.

[0023] The present invention is further configured such that the logic for generating the partitioned time series original data set includes: The explosion-proof monitoring area is spatially divided according to preset spatial grid parameters, and the area is discretized into multiple sub-areas. Specifically, based on the actual scope of the explosion-proof monitoring area and monitoring requirements, the spatial grid parameters are preset to divide the entire area evenly or as needed into several sub-areas, and each sub-area serves as an independent video monitoring acquisition unit. Explosion-proof intelligent video surveillance is deployed in each sub-area. The explosion-proof intelligent video surveillance is equipped with a multi-spectral video acquisition unit to perform multi-channel synchronous acquisition control. Specifically, explosion-proof intelligent video surveillance equipment is deployed in each sub-area and equipped with a multi-spectral video acquisition unit to simultaneously acquire infrared, visible light, and ultraviolet channel data. Through a unified synchronous clock and sampling trigger mechanism, high-frequency data acquisition is performed within the same time window to obtain video frame data containing infrared, visible light, and ultraviolet light. Specifically, through a global synchronous clock and a unified sampling trigger mechanism, it is ensured that the multispectral video acquisition units of all sub-areas perform high-frequency and synchronous data acquisition within the same time window, ensuring the timing alignment between different areas and different channels. At each sampling moment, infrared, visible light, ultraviolet and other multi-channel video frames of each sub-area are collected to form a data stream classified by channel. The multi-channel video frames collected from each sub-area at all sampling moments are structured and organized according to the partition number, time series and channel type to generate a partition time series original data set; specifically, the multi-channel video frames from all sub-areas and all sampling moments are structured and organized according to the partition number, time series and channel type to generate a partition time series original data set and upload it to the data processing center.

[0024] The present invention is further configured such that the construction logic of the original wave spectrum includes: The partitioned time series original data set is segmented according to the set time window, and the continuous multiple frames of video images of each partition and each channel are segmented; specifically, according to the set time window, the continuous multiple frames of video images of each partition and each channel in the partitioned time series original data set are time segmented, and the image frames in each time window are treated as a group for subsequent processing; For each continuous video frame in a time window, the pixel intensity sequence in the time window is extracted according to the pixel coordinates to generate a time-series pixel intensity matrix. Specifically, for each video frame sequence in a time window, all pixel coordinates are traversed to extract the pixel intensity value of the pixel point in the time window to form a time-series pixel intensity matrix arranged in time. Performing a first-order difference operation on the temporal pixel intensity matrix to obtain a sequence of pixel intensity changes between frames. Specifically, performing a first-order difference operation on the temporal pixel intensity matrix to obtain a sequence of intensity changes of the pixel between consecutive frames, reflecting a small disturbance of the local light field. Extract temporal statistical features of the pixel intensity change sequence, including sliding mean, variance, kurtosis, and skewness. Specifically, for each pixel intensity change sequence, calculate temporal statistical features such as sliding mean, variance, kurtosis, and skewness to reflect the change trend, fluctuation amplitude, and non-Gaussian characteristics. Perform a fast Fourier transform on each pixel intensity change sequence to extract frequency domain statistical features, including the main frequency component, energy spectrum distribution, local amplitude peak value, and instantaneous phase parameters. Specifically, perform a fast Fourier transform (FFT) on each pixel intensity change sequence to extract frequency domain characteristic parameters such as the main frequency component, energy spectrum distribution, local amplitude peak value, and instantaneous phase, which are used to reflect the dominant period and energy distribution characteristics of the disturbance. The time domain statistical features and frequency domain statistical features of all pixels and all channels in each time window are summarized according to spatial partitions to obtain a partitioned light field disturbance feature group; specifically, the time domain and frequency domain feature parameters of all pixels and all channels in each time window are statistically reduced according to spatial partitions to form a partitioned light field disturbance feature group; Taking partition, channel and time window as ternary indexes, the partition light field disturbance feature group is combined to generate a native wave spectrum data group. Specifically, the disturbance features under each partition, each channel and each time window are combined and stored in a structured manner according to the ternary index to generate a native wave spectrum data group.

[0025] The present invention is further configured such that the logic for obtaining the multi-dimensional eddy current spectrum linkage factor includes: For the native wave spectrum data groups of different channels in the same partition and the same time window, they are aligned according to the main disturbance parameters to construct a multi-channel disturbance parameter joint matrix. Specifically, for the native wave spectrum data of different channels in the same partition and the same time window, the main frequency components of each channel and the corresponding local amplitude peaks are extracted. Through interpolation or frequency matching, the main disturbance parameters of different channels are aligned according to the unified main frequency index to generate a multi-channel disturbance parameter joint matrix. Based on the energy spectrum distribution, the spatial concentration of disturbance energy in the partition is calculated and the vortex density index is extracted. Specifically, based on the energy spectrum distribution in the original wave spectrum, a high-frequency energy threshold is set, and high-frequency disturbance energy points with energy exceeding the threshold are screened. The density of these high-frequency energy points is then counted in the partition space. The density value is used as the vortex density index to reflect the energy agglomeration characteristics. Based on the main frequency components and local amplitude peaks, the energy synchronization change rates and peak energy synthesis strengths of different channels in the main frequency band are counted to extract the spectral aggregation flow index. Specifically, for the main frequency components and local amplitude peaks, the energy synchronization change rates of each channel in the main frequency band and the weighted synthesis strength of the main frequency amplitudes of each channel are counted in the aligned multi-channel parameter matrix. The synthesis result is defined as the spectral aggregation flow index, which reflects the degree of energy aggregation and synchronization change. Based on the gradient of the energy spectrum over time in the energy spectrum distribution, the rate of transfer of perturbation energy from low frequency to high frequency is calculated, and the perturbation transition rate is extracted. Specifically, the trajectory of the energy spectrum changing over time in the native wave spectrum is analyzed, the rate of change of the main energy frequency over time is tracked, and the speed of energy migration from low frequency to high frequency in the spectrum is calculated. This rate is used as the perturbation transition rate to measure energy migration and dynamic activity. The vortex density index, spectral aggregation flow index and perturbation transition rate are set as multidimensional vortex spectrum linkage factors; specifically, the vortex density index, spectral aggregation flow index and perturbation transition rate obtained above are jointly used as multidimensional vortex spectrum linkage factors, and as input features for subsequent high-dimensional clustering, anomaly discrimination and regional labeling.

[0026] The present invention is further configured to align the main disturbance parameters and construct a multi-channel disturbance parameter joint matrix, including: In each channel, the main frequency component, local amplitude peak value and instantaneous phase parameters are extracted from the native wave spectrum data set and set as the main disturbance parameters. Specifically, in each partition and each time window, the native wave spectrum data set of each channel (including infrared, visible light, and ultraviolet) is analyzed respectively, and the main frequency component, local amplitude peak value and instantaneous phase parameters of the channel in the time window are extracted and used as the main disturbance parameter set of the channel. For all channels in the same partition and time window, the main frequency components are combined to form a main frequency index set. Specifically, the main frequency components extracted from all channels in the current partition and time window are summarized to form a main frequency index set. This index set provides a unified frequency reference for all channel parameter alignment. For the main frequency index set, if there is a slight offset between the main frequency components of different channels, the parameter values ​​of each channel are uniformly mapped to the frequency points of the main frequency index set through linear interpolation or nearest neighbor matching algorithm, ensuring that the main disturbance parameters of each channel correspond one-to-one in the frequency dimension; specifically, for the slight offset that may occur between the main frequency components of different channels, the main disturbance parameter values ​​of each channel are mapped to the frequency points of the main frequency index set using linear interpolation or nearest neighbor frequency matching algorithm, achieving one-to-one correspondence in the frequency dimension of the main disturbance parameters of each channel; The aligned main frequency components, local amplitude peaks, and instantaneous phase parameters of each channel are arranged according to the main frequency index to form a multi-channel perturbation parameter joint matrix, where each row of the matrix corresponds to a channel, and each column corresponds to the perturbation parameter at a main frequency point. Specifically, the aligned main frequency components, local amplitude peaks, and instantaneous phase parameters of each channel are combined into a multi-channel perturbation parameter joint matrix according to the arrangement order of the main frequency index set. Each row of the matrix represents a channel, each column represents a main frequency point in the main frequency index set, and each element is the perturbation parameter of the corresponding channel at that main frequency point.

[0027] The present invention is further configured such that the extraction logic of the vortex density index, the spectral aggregation flow index and the perturbation transition rate includes: Based on the energy spectrum distribution of each partition in the native wave spectrum of each channel, a high-frequency energy threshold is set, and high-frequency perturbation energy points with energy greater than the high-frequency energy threshold are screened. The density of high-frequency perturbation energy points in the partition space is calculated, and the density is set as the vortex density index. Specifically, in each partition, based on the native wave spectrum of each channel, a high-frequency energy threshold is set (for example, the mean of the energy spectrum distribution of each channel or the mean doubled standard deviation is selected), and high-frequency perturbation energy points with energy higher than the threshold are screened. The spatial distribution of all high-frequency perturbation energy points is counted, and their spatial density in the partition is calculated (that is, the ratio of high-frequency points to the total number of pixels). This density is used as the vortex density index of the partition to characterize the agglomeration phenomenon of high-frequency perturbation energy. The main frequency components and corresponding local amplitude peaks of each channel in each partition are aligned, the synchronous change rates of the main frequency amplitudes of each channel are counted, and the synchronous change rates are set as the spectral aggregation flow index. Specifically, the main frequency indexes of the main frequency components and local amplitude peaks of different channels in each partition are aligned, the amplitude sequences of all channels at each main frequency point are counted, and the synchronous change rates of the main frequency amplitudes of each channel at the same main frequency point are calculated (for example, using the correlation coefficient, covariance, or the inverse of the standard deviation). After obtaining the synchronous change rates of all main frequency points, the average or weighted synthesis is taken as the spectral aggregation flow index of the partition, which is used to characterize the synergistic strength of energy perturbations of different channels. According to the energy spectrum distribution of each channel, a main energy frequency trajectory sequence is established, the time change rate of the main energy frequency is calculated, the rate of transfer of disturbance energy from low frequency to high frequency is extracted, and the time change rate is set as the perturbation transition rate; specifically, for each partition and each channel, the evolution process of the energy spectrum distribution on the time axis is analyzed, the main energy frequency position in each time window (that is, the frequency point of the maximum value of the energy spectrum) is recorded, the time series trajectory of the main energy frequency is established, and the rate of change of the frequency over time (such as the ratio of the difference between the main energy frequencies of adjacent time windows to the time interval) is calculated. The average change rate is used as the perturbation transition rate, which is used to reflect the activeness of the migration of disturbance energy from low frequency to high frequency.

[0028] The present invention is further configured such that the logic of the abnormality discrimination module includes: Based on the multidimensional eddy current spectrum linkage factor of all spatial partitions in the same time window, a multidimensional feature vector with vortex density index, spectral aggregation flow index and perturbation transition rate as components is constructed, and the feature vectors corresponding to each partition are combined into a multidimensional feature space; specifically, for all spatial partitions, the vortex density index, spectral aggregation flow index and perturbation transition rate of each partition are collected in the same time window to form a multidimensional feature vector of the three-dimensional coordinates, which are summarized into a high-dimensional feature space; Cluster the feature vectors of each partition in the multidimensional feature space, calculate the local density of each partition and the distance between it and the partition with higher density, identify the partition with high local density and far away from other high-density partitions as density peaks, and determine a preset number of high-density peak cluster centers; specifically, in the high-dimensional feature space, perform density peak clustering on all partition feature vectors, and calculate the local density of each partition in the feature space (such as the number of partitions within a certain distance radius). Then, calculate the distance between each partition and all higher-density partitions, and select points with high local density and far away from other high-density partitions as density peaks (cluster centers). Based on the preset number of cluster centers, the high-density peak is finally selected as the center of the high-condensation area; Based on the high-density cluster center, the remaining partitions are assigned to the nearest cluster center according to the characteristic distance to form a high-cohesion area candidate set; specifically, for the remaining partitions, the distance between their characteristic vectors and each cluster center is calculated, and they are assigned to the nearest center, thereby forming several high-cohesion area candidate sets; For each candidate set of highly cohesive regions, the consistency, peak synthesis intensity, and correlation of energy distribution among channels within the region are calculated based on the spectral aggregation flow index within the same region, and this is set as the spectral convergence factor. Specifically, for each candidate set of highly cohesive regions, the consistency, peak synthesis intensity, and correlation (such as mean, standard deviation, correlation coefficient, and other statistics) of the main frequency energy distribution of all channels within the region are calculated based on the spectral aggregation flow index of each partition within the region. The spectral convergence factor of the region is synthesized using these statistics to reflect the synchronization and coordination characteristics of energy distribution among each partition. A threshold value of the spectral convergence factor is set, and the candidate set of high cohesion areas with a spectral convergence factor higher than the corresponding threshold is judged as an abnormally high cohesion area, the area is spatially partitioned and marked, and the abnormal partition index is output; specifically, a discrimination threshold of the spectral convergence factor is set, and the candidate set of high cohesion areas with a spectral convergence factor higher than the threshold is judged as an abnormally high cohesion area, and the spatial partition marking and abnormal area index output are completed.

[0029] The present invention is further configured such that the calculation logic of the consistency of energy distribution, peak composite intensity and correlation between channels within a region is as follows: For the energy data extracted at the main frequency component of all channels within the spatial partition, the mean and standard deviation of the main frequency energy of all channels are obtained, and the ratio of the mean and standard deviation is set as the energy distribution consistency index. Specifically, for all channels in the same area, energy data is extracted at the main frequency component respectively, and the mean and standard deviation of the main frequency energy of all channels are calculated. The ratio of the main frequency energy standard deviation to the mean is defined as the energy distribution consistency index of the area. The smaller the ratio, the more uniform the main frequency energy distribution of each channel and the higher the consistency. The main frequency energies of all channels are weighted to obtain the composite intensity of the peak energy within the partition, which is set as the peak composite intensity and used to measure the overall aggregation level of multi-channel energy. Specifically, the main frequency energies of all channels in the area are weighted and summed according to the set weights (such as equal weight or weighted by channel importance) to obtain the total energy, which is used as the peak energy composite intensity. This indicator reflects the overall aggregation effect of multi-channel energy. The larger the value, the higher the overall energy level between channels. For each channel's main frequency energy data, the correlation coefficient between any two channels is calculated. The absolute mean of all correlation coefficients is used as the correlation index for the energy distribution between channels. Specifically, the main frequency energy data for each channel is combined with each other and the Pearson correlation coefficient of the main frequency energy between channels is calculated. The absolute values ​​of all correlation coefficients are taken and averaged to obtain the correlation index for the energy distribution between channels. A larger value indicates more synchronized energy changes between channels and a higher degree of coordination.

[0030] The present invention is further configured such that the construction logic of the multidimensional diffusion path model includes: For the identified abnormally high cohesion areas, time-series trajectory tracking is performed within continuous time windows based on the spatial distribution of moment partitions and regional labeling results. Abnormally high cohesion areas with adjacent spatial positions in adjacent time windows are determined to be the same object, and a continuous spatial trajectory of the abnormal area is generated. Specifically, within the continuous time window, for each spatial partition identified in each frame, it is determined whether it belongs to an abnormally high cohesion area. Based on the spatial position (such as partition number, regional centroid coordinates, etc.) and regional labeling, the trajectories of abnormally high cohesion areas with adjacent or overlapping spatial positions in the previous and next adjacent time windows are associated and determined to be the same object, and a continuous spatial trajectory of its change over time is generated. For each abnormally high cohesion area, the coordinates of the spatial center point in each time window are extracted, and based on the position change of the center point in the time series, the diffusion speed and main diffusion direction of the abnormal area in space are quantified. The diffusion speed is calculated by the displacement of the center point and the time interval in adjacent time windows, and the main diffusion direction is determined by the trend of time series position change. Specifically, for each abnormally high cohesion area, the coordinates of its spatial center point (such as the position of the maximum disturbance point) are extracted in each time window; the spatial displacement between the center points of two consecutive time windows is calculated and divided by the time interval to obtain the diffusion speed at each moment; according to the trend of the time series position change of the center point, the main diffusion direction is determined using the principal component direction, vector accumulation or fitting method; The number of continuous time windows that each abnormally high cohesion area experienced from the first recognition to its disappearance was counted, and the duration corresponding to the number of continuous time windows was defined as the duration of the abnormally high cohesion area. Specifically, the number of continuous time windows from the first appearance to the last recognition of each abnormally high cohesion area was recorded, and the number of windows was multiplied by the window duration to obtain the actual duration of the abnormally high cohesion area. The spatial diffusion velocity, displacement direction and duration of abnormally high condensation areas are integrated to establish a multidimensional diffusion path model. Specifically, based on the above parameters, the spatial trajectory, diffusion velocity sequence, main diffusion direction and duration of each abnormally high condensation area are jointly modeled into a multidimensional diffusion path model as the input for subsequent risk warning and response control.

[0031] The present invention is further configured to set a multi-level response threshold based on a multi-dimensional diffusion path model to perform risk warning, including: When the spatial diffusion speed is less than the first spatial diffusion speed threshold and the duration is less than the first duration threshold, it is determined to be a low-risk state and an abnormal state record is performed; When the spatial diffusion speed is greater than or equal to the first spatial diffusion speed threshold and less than the second spatial diffusion speed threshold, and the duration is greater than or equal to the first duration threshold and less than the second duration threshold, it is determined to be a medium risk state and a local risk warning is triggered; When the spatial diffusion velocity is greater than or equal to the second spatial diffusion velocity threshold and the duration is greater than or equal to the second duration threshold, a high-risk state is determined and a remote alarm is issued. Specifically, for each abnormally high-condensation area, the spatial diffusion velocity and duration of the area are continuously collected during the continuous monitoring cycle. Based on actual application needs, the system pre-sets the first and second thresholds for spatial diffusion velocity and duration, which serve as the basis for grading low, medium, and high risks.

[0032] When the spatial diffusion speed of the abnormally high condensation area is lower than the first threshold and the duration is lower than the first threshold, it is determined to be a low-risk state and only the abnormal state is recorded; When the spatial diffusion speed is between the first and second thresholds, and the duration is between the first and second thresholds, it is determined to be a medium-risk state, and the system automatically triggers a local risk warning; When the spatial diffusion speed is higher than or equal to the second threshold, and the duration is higher than or equal to the second threshold, it is determined to be a high-risk state, and remote alarm and emergency response measures are immediately initiated.

[0033] Corresponding to the risk level, the system adopts graded disposal measures such as exception archiving, local reminders, and remote alarms to achieve automated dynamic risk management.

[0034] The above description is merely a specific embodiment of the present application, but the scope of protection of the present application is not limited thereto. Any changes or substitutions that can be easily conceived by a person skilled in the art within the technical scope disclosed in this application should be included in the scope of protection of this application. Therefore, the scope of protection of this application should be based on the scope of protection of the claims.

Claims

1. An explosion-proof intelligent video surveillance control system, characterized in that: include: Acquisition module: used to collect multispectral video data in the explosion-proof monitoring area in real time. Based on the preset spatial grid division, multi-channel high-frequency synchronous sampling is performed on the sub-area to generate the partitioned time series raw data set; Feature extraction module: used to extract the pixel intensity changes of consecutive frames within the time window of the partitioned time series raw data set, obtain the light field disturbance characteristics and construct the original fluctuation spectrum through time domain transformation and frequency domain transformation; Linkage factor construction module: used to construct a multi-channel perturbation parameter joint matrix based on the native wave spectra of different channels in the same area, and obtain the multi-dimensional eddy current spectrum linkage factor. The multi-dimensional eddy current spectrum linkage factor includes the vortex density index, the spectrum aggregation flow index and the perturbation transition rate; Anomaly discrimination module: used to perform high-dimensional density peak clustering on the multi-dimensional eddy current spectrum linkage factors of all spatial partitions, identify areas with significant eddy current disturbance aggregation trends, calculate the spectral convergence factor, and identify and mark abnormally high-condensation areas in the space based on the spectral convergence factor; Path modeling module: used to track the time series of abnormally high cohesion areas, extract spatial diffusion speed, displacement direction and duration, and build a multidimensional diffusion path model; Risk warning module: used to set multi-level response thresholds based on the multidimensional diffusion path model to provide risk warnings.

2. The explosion-proof intelligent video surveillance control system according to claim 1, characterized in that: The logic for generating partitioned time series raw datasets includes: The explosion-proof monitoring area is spatially divided according to the preset spatial grid parameters, and the area is discretized into multiple sub-areas; Explosion-proof intelligent video surveillance is deployed in each sub-area. The explosion-proof intelligent video surveillance is equipped with a multi-spectral video acquisition unit for multi-channel synchronous acquisition control. Through a unified synchronous clock and sampling trigger mechanism, high-frequency data acquisition is performed within the same time window to obtain video frame data containing infrared, visible light, and ultraviolet light; The multi-channel video frames collected from each sub-region at all sampling moments are structured and organized according to the partition number, time series and channel type to generate a partition time series raw data set.

3. The explosion-proof intelligent video surveillance control system according to claim 1, characterized in that: The construction logic of the native wave spectrum includes: The partitioned time series original data set is segmented and processed according to the set time window for each partition and each channel's continuous multi-frame video images; For each continuous video frame in a time window, the pixel intensity sequence in the time window is extracted according to the pixel coordinates to generate a time-series pixel intensity matrix; Perform first-order difference on the time-series pixel intensity matrix to obtain the inter-frame pixel intensity change sequence; Extract temporal statistical features of pixel intensity change sequence, including sliding mean, variance, kurtosis and skewness; Perform fast Fourier transform on each pixel intensity change sequence to extract frequency domain statistical features, including main frequency component, energy spectrum distribution, local amplitude peak and instantaneous phase parameters; The time domain statistical features and frequency domain statistical features of all pixels and all channels in each time window are summarized by spatial partition to obtain a partitioned light field disturbance feature group; Taking partition, channel and time window as the ternary index, the partition light field disturbance feature group is combined to generate the native wave spectrum data group.

4. The explosion-proof intelligent video surveillance control system according to claim 3, characterized in that: The logic for obtaining the multi-dimensional eddy current spectrum linkage factor includes: For the native fluctuation spectrum data sets of different channels in the same partition and the same time window, they are aligned according to the main disturbance parameters to construct a multi-channel disturbance parameter joint matrix; Based on the energy spectrum distribution, the spatial concentration of disturbance energy in the partition is calculated and the vortex density index is extracted; According to the main frequency component and the local amplitude peak, the energy synchronization change rate and peak energy synthesis intensity of different channels in the main frequency band are counted to extract the spectral aggregation flow index; According to the gradient of the energy spectrum in the energy spectrum distribution over time, the rate of transfer of disturbance energy from low frequency to high frequency is calculated, and the perturbation transition rate is extracted; The vortex density index, spectral aggregation flow index and perturbation transition rate are set as multidimensional vortex spectrum linkage factors.

5. The explosion-proof intelligent video surveillance control system according to claim 1, characterized in that: Align the main disturbance parameters and construct a multi-channel disturbance parameter joint matrix, including: In each channel, the main frequency component, local amplitude peak and instantaneous phase parameters are extracted from the original wave spectrum data set and set as the main disturbance parameters; For the main frequency components of all channels in the same partition and the same time window, a main frequency index set is formed; For the main frequency index set, if there is a slight offset between the main frequency components of different channels, the parameter values ​​of each channel are uniformly mapped to the frequency points of the main frequency index set through linear interpolation or nearest neighbor matching algorithm to ensure that the main disturbance parameters of each channel correspond one to one in the frequency dimension; The aligned main frequency components, local amplitude peaks and instantaneous phase parameters of each channel are arranged according to the main frequency index to form a multi-channel disturbance parameter joint matrix, where each row of the matrix corresponds to a channel and each column corresponds to a disturbance parameter at a main frequency point.

6. The explosion-proof intelligent video surveillance control system according to claim 4, characterized in that: The extraction logic of vortex density index, spectral aggregation flow index and perturbation transition rate includes: Based on the energy spectrum distribution of each partition in the original wave spectrum of each channel, a high-frequency energy threshold is set, and high-frequency disturbance energy points with energy greater than the high-frequency energy threshold are screened. The density of high-frequency disturbance energy points in the partition space is calculated and set as the vortex density index; Align the main frequency components of each channel in each partition with the corresponding local amplitude peak, calculate the synchronization change rate of the main frequency amplitude of each channel, and set the synchronization change rate as the spectrum aggregation flow index; According to the energy spectrum distribution of each channel, the main energy frequency trajectory sequence is established, the time change rate of the main energy frequency is calculated, the rate of transfer of disturbance energy from low frequency to high frequency is extracted, and the time change rate is set as the perturbation transition rate.

7. The explosion-proof intelligent video surveillance control system according to claim 1, characterized in that: The logic of the abnormality identification module includes: Based on the multidimensional eddy spectrum linkage factor of all spatial partitions in the same time window, a multidimensional feature vector with eddy density index, spectral aggregation flow index and perturbation transition rate as components is constructed, and the feature vectors corresponding to each partition are combined into a multidimensional feature space. Clustering the feature vectors of each partition in the multidimensional feature space, calculating the local density of each partition and the distance between each partition and the partition with higher density, identifying the partition with high local density and far away from other high-density partitions as the density peak, and determining a preset number of high-density peak cluster centers; Based on the high-density cluster center, the remaining partitions are assigned to the nearest cluster center according to the feature distance to form a candidate set of high-density regions; For each high-condensation region candidate set, based on the spectral aggregation flow index in the same region, the consistency of energy distribution, peak synthesis intensity and correlation between channels in the region are calculated and set as the spectral convergence factor; The spectral convergence factor threshold is set, and the abnormal high-condensation area is determined for the high-condensation area candidate set whose spectral convergence factor is higher than the corresponding threshold. The area is spatially partitioned and marked, and the abnormal partition index is output.

8. The explosion-proof intelligent video surveillance control system according to claim 1, characterized in that: The calculation logic of the consistency of energy distribution, peak composite intensity and correlation between channels in the region is: For the energy data extracted at the main frequency component of all channels in the spatial partition, the mean and standard deviation of the main frequency energy of all channels are obtained, and the ratio of the mean and standard deviation is set as the energy distribution consistency index; Perform weighted summation on the main frequency energies of all channels to obtain the composite intensity of the peak energy within the partition, which is set as the peak composite intensity and used to measure the overall aggregation level of multi-channel energy; For the main frequency energy data of each channel, the correlation coefficient between any two channels is calculated respectively, and the absolute value mean of all correlation coefficients is set as the correlation index of the energy distribution between channels.

9. The explosion-proof intelligent video surveillance control system according to claim 1, characterized in that: The construction logic of the multidimensional diffusion path model includes: For the identified abnormally high cohesion areas, time series trajectory tracking is performed within continuous time windows based on the spatial distribution of moment partitions and regional labeling results. Abnormally high cohesion areas with adjacent spatial positions in adjacent time windows are determined to be the same object, generating a continuous spatial trajectory of the abnormal area. For each abnormally high-condensation area, the coordinates of the spatial center point in each time window are extracted. Based on the position change of the center point in the time series, the diffusion speed and main diffusion direction of the abnormal area in space are quantified. The diffusion speed is calculated by the displacement and time interval of the center point in adjacent time windows, and the main diffusion direction is determined by the trend of time series position change. Count the number of continuous time windows that each abnormally high cohesion area experiences from the first identification to its disappearance, and define the duration corresponding to the number of continuous time windows as the duration of the abnormally high cohesion area; A multidimensional diffusion path model was established by integrating the spatial diffusion velocity, displacement direction and duration of the abnormally high condensation area.

10. The explosion-proof intelligent video surveillance control system according to claim 1, characterized in that: Based on the multi-dimensional diffusion path model, multi-level response thresholds are set to provide risk warnings, including: When the spatial diffusion speed is less than the first spatial diffusion speed threshold and the duration is less than the first duration threshold, it is determined to be a low-risk state and an abnormal state record is performed; When the spatial diffusion speed is greater than or equal to the first spatial diffusion speed threshold and less than the second spatial diffusion speed threshold, and the duration is greater than or equal to the first duration threshold and less than the second duration threshold, it is determined to be a medium risk state and a local risk warning is triggered; When the spatial diffusion speed is greater than or equal to the second spatial diffusion speed threshold and the duration is greater than or equal to the second duration threshold, it is determined to be a high-risk state and a remote alarm is issued.

Images