Video big data analysis-based population information acquisition and early warning analysis method

Abstract

The invention relates to the technical field of population collection and early warning methods, in particular to a population information collection and early warning analysis method based on video big data analysis, and aims to solve the problems that data analysis is carried out based on single dimensions such as time or space, and a model possibly lacks self-learning and optimization capabilities. Due to the characteristics of fast urbanization process, large population mobility and the like in China, the complexity and diversity of user behaviors cannot be comprehensively reflected by a traditional general technical scheme, and on the basis of a traditional population information collection and early warning analysis method based on video big data analysis, a three-dimensional graph model construction method is combined, so that the population information collection and early warning analysis method is established. Behavior modes of a user in different time and space dimensions are represented by utilizing a three-dimensional diagram theory, user behaviors are comprehensively analyzed through x-axis, y-axis and z-axis three dimensions, and meanwhile, an evaluation model is combined with a strategy model to output data, so that the model continuously learns and evolves, and the accuracy and effectiveness of classification are improved.

Description

Technical Field

[0001] The present invention relates to the technical field of population collection and early warning methods, and in particular to a population information collection and early warning analysis method based on video big data analysis. Background Art

[0002] The population information collection and early warning analysis method based on video big data analysis is a method that uses video big data technology and three-dimensional graph theory to comprehensively analyze user behavior patterns, thereby performing population information collection and model early warning analysis.

[0003] Existing population information collection and early warning analysis methods based on video big data analysis generally perform data analysis based on a single dimension such as time or space, which cannot fully reflect the complexity and diversity of user behavior. At the same time, the model may lack the ability of self-learning and optimization, and it is difficult to adapt to changes in the data environment, which can easily lead to delayed or inaccurate analysis results.

[0004] In view of the problem that the existing population information collection and early warning analysis method based on video big data analysis cannot fully reflect the complexity and diversity of user behavior, this population information collection and early warning analysis method based on video big data analysis uses three-dimensional graph theory to represent the user's behavior patterns in different time and space dimensions, and comprehensively analyzes user behavior through the three dimensions of x-axis, y-axis, and z-axis. It can more accurately identify user behavior patterns, and at the same time, continuously optimize the classification strategy model by combining the evaluation model with the strategy model output data, so that the model can continuously learn and evolve, thereby improving the accuracy and effectiveness of classification. Summary of the invention

[0005] In order to overcome the existing population information collection and early warning analysis methods based on video big data analysis, data analysis is generally based on a single dimension such as time or space, which cannot fully reflect the complexity and diversity of user behavior. At the same time, the model may lack the ability of self-learning and optimization, and it is difficult to adapt to changes in the data environment, which can easily lead to delayed or inaccurate analysis results.

[0006] The technical solution of the present invention is: a population information collection and early warning analysis method based on video big data analysis, including the following methods:

[0007] S11: Collect user behavior data in different time and space dimensions, including user access control entry and exit records and floor visit records;

[0008] S12: Clean the collected data, remove errors and invalid information, standardize the data, and ensure the consistency and availability of the data;

[0009] S13: Apply three-dimensional graph theory to construct a three-dimensional graph model;

[0010] S14: using a clustering algorithm to preliminarily classify user behaviors in the three-dimensional graph, and classify users into different groups according to their behavior patterns;

[0011] S15: Calculate the frequency cardinality, proximity cardinality, habit cardinality and normal cardinality of each classified user, and calculate the Val1 value based on these parameters for further segmentation and classification of users;

[0012] S16: Optimizing the classification strategy model by combining the output data of the pre-selected multiple models;

[0013] S17: Based on the optimized model, real-time monitoring and early warning analysis of user behavior are performed;

[0014] S18: When abnormal user behavior is detected or meets the preset warning conditions, timely issue warning information;

[0015] S19: Evaluate and verify the model regularly, continuously optimize and adjust model parameters based on feedback results, introduce new data sources and algorithms, and maintain the advancement and adaptability of the model.

[0016] Preferably, the following methods are included when collecting and preprocessing data:

[0017] S21: Collect population behavior data through multiple sources such as video surveillance systems and access control systems;

[0018] S22: Remove duplicate, invalid or abnormal data records to ensure the accuracy and completeness of the data;

[0019] S23: Convert data of different dimensions into a unified dimension, and standardize the data to make it into a unified format for subsequent analysis and processing.

[0020] Preferably, when constructing a three-dimensional graph model, the following method is included:

[0021] S31: Using the three-dimensional graph theory, define the x-axis as the time dimension, the y-axis as the frequency of occurrence, and the z-axis as the entry and exit position;

[0022] S32: Map each user's behavior data to corresponding points x, y, z in three-dimensional space.

[0023] Preferably, three-dimensional graph theory is used to represent the user's behavior patterns in different time and space dimensions. The time dimension, frequency of occurrence and entry and exit positions are integrated in the form of three-dimensional points, thereby achieving a more detailed and comprehensive characterization of user behavior. The three-dimensional graph theory can make the characterization of user behavior patterns more detailed, thereby improving the accuracy of information collection, helping to reduce false alarms and omissions, and improving the accuracy of early warning analysis.

[0024] Preferably, the following methods are included when performing preliminary classification by cluster analysis:

[0025] S41: clustering points in the three-dimensional space using a suitable clustering algorithm;

[0026] S42: Based on the clustering results, the users are initially divided into different groups, each group representing a behavior pattern.

[0027] Preferably, when calculating the frequency base, the following method is included:

[0028] S51: Calculate the appearance frequency of each user or user group in different time periods;

[0029] S52: Determine the frequency base P of each user or user group based on the statistical results.

[0030] Preferably, when calculating the nearest cardinality, the following method is included:

[0031] S61: Analyze the data of the user's entry and exit locations, including access control points and floors;

[0032] S62: Calculate the movement distance or frequency of the user between different locations;

[0033] S63: Through S61 and S62, determine the user's proximity cardinality J, reflecting the geographical concentration of user activities.

[0034] Preferably, when calculating the customary base, the following method is included:

[0035] S71: Identify the user's habitual behavior patterns by analyzing the user's behavior data at different time periods and locations;

[0036] S72: Quantify the identified habitual behaviors and calculate the habit base X to reflect the stability and regularity of the user's behavior.

[0037] Preferably, when calculating the normal base, the following steps are included:

[0038] S81: Perform a normality test on the user behavior data to determine whether it conforms to a normal distribution;

[0039] S82: For data that conform to the normal distribution, calculate its standard deviation;

[0040] S83: Based on the standard deviation and other statistics, determine the user's normal cardinality S, which reflects the volatility and stability of user behavior.

[0041] Preferably, the comprehensive evaluation and classification optimization are performed, including the following steps:

[0042] S91: Combine the frequency base P, the proximity base J, the habit base X and the normal base S to calculate the Val1 value;

[0043] S92: Classify users or user groups according to the Val1 value, and continuously optimize the classification strategy model in combination with the evaluation model.

[0044] Preferably, the following steps are included when performing early warning analysis and output:

[0045] S1001: Based on the optimized classification strategy model, real-time monitoring of user behavior data is performed to detect abnormal behavior;

[0046] S1002: Perform early warning analysis on the detected abnormal behavior to determine its possible causes and impacts;

[0047] S1003: Output warning information to users, managers or related systems so that timely measures can be taken to deal with potential risks.

[0048] Beneficial effects of the present invention:

[0049] 1. Compared with the traditional population information collection and early warning analysis method based on video big data analysis, which generally performs data analysis based on a single dimension such as time or space, it cannot fully reflect the complexity and diversity of user behavior. At the same time, the model may lack the ability of self-learning and optimization, and it is difficult to adapt to changes in the data environment, which easily leads to delayed or inaccurate analysis results. The population information collection and early warning analysis method based on video big data analysis uses three-dimensional graph theory to represent the user's behavior patterns in different time and space dimensions, and comprehensively analyzes user behavior through the three dimensions of x-axis, y-axis, and z-axis. It can more accurately identify user behavior patterns, and at the same time, it combines the evaluation model with the strategy model to output data, continuously optimizes the classification strategy model, so that the model can continuously learn and evolve, and improve the accuracy and effectiveness of classification;

[0050] 2. Use three-dimensional graph theory to represent user behavior patterns in different time and space dimensions. In the form of three-dimensional points, the time dimension, frequency of occurrence and entry and exit positions are integrated to achieve a more detailed and comprehensive characterization of user behavior. The three-dimensional graph theory can make the characterization of user behavior patterns more detailed, thereby improving the accuracy of information collection, helping to reduce false alarms and missed alarms, and improving the accuracy of early warning analysis. BRIEF DESCRIPTION OF THE DRAWINGS

[0051] Figure 1 What is shown is a schematic diagram of the workflow of the population information collection and early warning analysis method based on video big data analysis of the present invention;

[0052] Figure 2What is shown is a schematic diagram of data collection and preprocessing workflow of the population information collection and early warning analysis method based on video big data analysis of the present invention;

[0053] Figure 3 What is shown is a schematic diagram of the process of constructing a three-dimensional graph model of the population information collection and early warning analysis method based on video big data analysis of the present invention. DETAILED DESCRIPTION

[0054] The present invention will be further described below in conjunction with the accompanying drawings and embodiments.

[0055] See also Figure 1-2 The present invention provides an embodiment: a method for collecting and analyzing population information based on video big data analysis, including the following methods:

[0056] S11: Collect user behavior data in different time and space dimensions, including user access control entry and exit records and floor visit records;

[0057] S12: Clean the collected data, remove errors and invalid information, standardize the data, and ensure the consistency and availability of the data;

[0058] S13: Apply three-dimensional graph theory to construct a three-dimensional graph model;

[0059] S14: using a clustering algorithm to preliminarily classify user behaviors in the three-dimensional graph, and classify users into different groups according to their behavior patterns;

[0060] S15: Calculate the frequency cardinality, proximity cardinality, habit cardinality and normal cardinality of each classified user, and calculate the Val1 value based on these parameters for further segmentation and classification of users;

[0061] S16: Optimizing the classification strategy model by combining the output data of the pre-selected multiple models;

[0062] S17: Based on the optimized model, real-time monitoring and early warning analysis of user behavior are performed;

[0063] S18: When abnormal user behavior is detected or meets the preset warning conditions, timely issue warning information;

[0064] S19: Evaluate and verify the model regularly, continuously optimize and adjust model parameters based on feedback results, introduce new data sources and algorithms, and maintain the advancement and adaptability of the model.

[0065] Preferably, the following methods are included when collecting and preprocessing data:

[0066] S21: Collect population behavior data through multiple sources such as video surveillance systems and access control systems;

[0067] S22: Remove duplicate, invalid or abnormal data records to ensure the accuracy and completeness of the data;

[0068] S23: Convert data of different dimensions into a unified dimension, and standardize the data to make it into a unified format for subsequent analysis and processing.

[0069] Preferably, when constructing a three-dimensional graph model, the following method is included:

[0070] S31: Using the three-dimensional graph theory, define the x-axis as the time dimension, the y-axis as the frequency of occurrence, and the z-axis as the entry and exit position;

[0071] S32: Map each user's behavior data to corresponding points x, y, z in three-dimensional space.

[0072] Preferably, three-dimensional graph theory is used to represent the user's behavior patterns in different time and space dimensions. The time dimension, frequency of occurrence and entry and exit positions are integrated in the form of three-dimensional points, thereby achieving a more detailed and comprehensive characterization of user behavior. The three-dimensional graph theory can make the characterization of user behavior patterns more detailed, thereby improving the accuracy of information collection, helping to reduce false alarms and omissions, and improving the accuracy of early warning analysis.

[0073] Preferably, the following methods are included when performing preliminary classification by cluster analysis:

[0074] S41: clustering points in the three-dimensional space using a suitable clustering algorithm;

[0075] S42: Based on the clustering results, the users are initially divided into different groups, each group representing a behavior pattern.

[0076] Preferably, when calculating the frequency base, the following method is included:

[0077] S51: Calculate the appearance frequency of each user or user group in different time periods;

[0078] S52: Determine the frequency base P of each user or user group based on the statistical results.

[0079] Preferably, when calculating the nearest cardinality, the following method is included:

[0080] S61: Analyze the data of the user's entry and exit locations, including access control points and floors;

[0081] S62: Calculate the movement distance or frequency of the user between different locations;

[0082] S63: Through S61 and S62, determine the user's proximity cardinality J, reflecting the geographical concentration of user activities.

[0083] Preferably, when calculating the customary base, the following method is included:

[0084] S71: Identify the user's habitual behavior patterns by analyzing the user's behavior data at different time periods and locations;

[0085] S72: Quantify the identified habitual behaviors and calculate the habit base X to reflect the stability and regularity of the user's behavior.

[0086] Preferably, when calculating the normal base, the following steps are included:

[0087] S81: Perform a normality test on the user behavior data to determine whether it conforms to a normal distribution;

[0088] S82: For data that conform to the normal distribution, calculate its standard deviation;

[0089] S83: Based on the standard deviation and other statistics, determine the user's normal cardinality S, which reflects the volatility and stability of user behavior.

[0090] Preferably, the comprehensive evaluation and classification optimization are performed, including the following steps:

[0091] S91: Combine the frequency base P, the proximity base J, the habit base X and the normal base S to calculate the Val1 value;

[0092] S92: Classify users or user groups according to the Val1 value, and continuously optimize the classification strategy model in combination with the evaluation model.

[0093] Preferably, the following steps are included when performing early warning analysis and output:

[0094] S1001: Based on the optimized classification strategy model, real-time monitoring of user behavior data is performed to detect abnormal behavior;

[0095] S1002: Perform early warning analysis on the detected abnormal behavior to determine its possible causes and impacts;

[0096] S1003: Output warning information to users, managers or related systems so that timely measures can be taken to deal with potential risks.

[0097] The embodiments of the present invention are described in detail above with reference to the accompanying drawings, but the present invention is not limited to the above embodiments, and various changes can be made within the knowledge scope of those skilled in the art without departing from the spirit of the present invention.

Claims

1. Population information collection and early warning analysis method based on video big data analysis; characterized by: The following methods are included: S11: Collect user behavior data in different time and space dimensions, including user access control entry and exit records and floor visit records; S12: Clean the collected data, remove errors and invalid information, standardize the data, and ensure the consistency and availability of the data; S13: Apply three-dimensional graph theory to construct a three-dimensional graph model; S14: using a clustering algorithm to preliminarily classify user behaviors in the three-dimensional graph, and classify users into different groups according to their behavior patterns; S15: Calculate the frequency cardinality, proximity cardinality, habit cardinality and normal cardinality of each classified user, and calculate the Val1 value based on these parameters for further segmentation and classification of users; S16: Optimizing the classification strategy model by combining the output data of the pre-selected multiple models; S17: Based on the optimized model, real-time monitoring and early warning analysis of user behavior are performed; S18: When abnormal user behavior is detected or meets the preset warning conditions, timely issue warning information; S19: Evaluate and verify the model regularly, continuously optimize and adjust model parameters based on feedback results, introduce new data sources and algorithms, and maintain the advancement and adaptability of the model.

2. The method for collecting and analyzing population information based on video big data analysis according to claim 1 is characterized in that: When collecting and preprocessing data, the following methods are included: S21: Collect population behavior data through multiple sources such as video surveillance systems and access control systems; S22: Remove duplicate, invalid or abnormal data records to ensure the accuracy and completeness of the data; S23: Convert data of different dimensions into a unified dimension, and standardize the data to make it into a unified format for subsequent analysis and processing.

3. The method for population information collection and early warning analysis based on video big data analysis according to claim 1 is characterized in that: When building a three-dimensional graph model, the following methods are included: S31: Using the three-dimensional graph theory, define the x-axis as the time dimension, the y-axis as the frequency of occurrence, and the z-axis as the entry and exit position; S32: Map each user's behavior data to corresponding points x, y, z in three-dimensional space.

4. The method for collecting population information and analyzing early warning information based on video big data analysis according to claim 1 is characterized in that: When conducting a preliminary classification of cluster analysis, the following methods are included: S41: clustering points in the three-dimensional space using a suitable clustering algorithm; S42: Based on the clustering results, the users are initially divided into different groups, each group representing a behavior pattern.

5. The method for population information collection and early warning analysis based on video big data analysis according to claim 1 is characterized in that: When calculating the frequency base, the following methods are included: S51: Calculate the appearance frequency of each user or user group in different time periods; S52: Determine the frequency base P of each user or user group based on the statistical results.

6. The method for population information collection and early warning analysis based on video big data analysis according to claim 1 is characterized in that: When calculating the nearest cardinality, the following methods are included: S61: Analyze the data of the user's entry and exit locations, including access control points and floors; S62: Calculate the movement distance or frequency of the user between different locations; S63: Through S61 and S62, determine the user's proximity cardinality J, reflecting the geographical concentration of user activities.

7. The method for population information collection and early warning analysis based on video big data analysis according to claim 1 is characterized in that: When calculating the habit base, the following methods are included: S71: Identify the user's habitual behavior patterns by analyzing the user's behavior data at different time periods and locations; S72: Quantify the identified habitual behaviors and calculate the habit base X to reflect the stability and regularity of the user's behavior.

8. The method for population information collection and early warning analysis based on video big data analysis according to claim 1 is characterized in that: When calculating the normal base, the following steps are involved: S81: Perform a normality test on the user behavior data to determine whether it conforms to a normal distribution; S82: For data that conform to the normal distribution, calculate its standard deviation; S83: Based on the standard deviation and other statistics, determine the user's normal cardinality S, which reflects the volatility and stability of user behavior.

9. The method for collecting population information and analyzing early warning information based on video big data analysis according to claim 1 is characterized in that: When conducting comprehensive evaluation and classification optimization, the following steps are included: S91: Combine the frequency base P, the proximity base J, the habit base X and the normal base S to calculate the Val1 value; S92: Classify users or user groups according to the Val1 value, and continuously optimize the classification strategy model in combination with the evaluation model.

10. The method for population information collection and early warning analysis based on video big data analysis according to claim 1 is characterized in that: When conducting early warning analysis and output, the following steps are included: S1001: Based on the optimized classification strategy model, real-time monitoring of user behavior data is performed to detect abnormal behavior; S1002: Perform early warning analysis on the detected abnormal behavior to determine its possible causes and impacts; S1003: Output warning information to users, managers or related systems so that timely measures can be taken to deal with potential risks.

Images