CN-122018738-A - Method and system for generating dynamic visualized interactive display interface of people flow

CN122018738ACN 122018738 ACN122018738 ACN 122018738ACN-122018738-A

Abstract

The invention discloses a method and a system for generating a dynamic visualized interactive display interface of people flow, relates to the technical field of data visualization, and is used for solving the problem that macroscopic situation and microscopic detail cannot be considered in large-scale people flow monitoring. The method comprises the steps of analyzing track vector characteristics, constructing an interactive entropy field reflecting microscopic motion disorder degree to generate global continuous data, then utilizing Gaussian differential convolution to extract characteristics, constructing a visual saliency weight matrix, executing probability sampling and weighted Thiessen polygon subdivision based on weights to generate a space geometric constraint type self-adaptive viewport topology grid, finally utilizing the grid as a projection container to execute differential rendering, reconstructing microscopic track dynamics at a small viewport, generating pressure tone filling at a large viewport, realizing fusion interactive display of multi-granularity data, and improving decision efficiency of public safety monitoring.

Inventors

CHEN MAO
ZHANG JUNWEI
Chen Xiumizi

Assignees

深圳市卖点科技股份有限公司

Dates

Publication Date: 20260512
Application Date: 20260204

Claims (9)

1. The method for generating the dynamic visualized interactive display interface of the people flow is characterized by comprising the following steps of: S1, carrying out vectorization analysis on acquired personnel track time sequence data, intercepting track fragments through a sliding time window, calculating included angle dispersion and speed modular length attenuation rate between adjacent track vectors, constructing a local interaction entropy value reflecting the disorder degree of group microscopic motion, and carrying out inverse distance weighted interpolation operation to map the discrete local interaction entropy value to a physical space grid unit so as to generate global people stream interaction entropy field data; S2, performing multi-scale Gaussian difference convolution operation on the global people stream interaction entropy field data, extracting spatial extreme points and edge gradient features of local entropy values, and constructing a visual saliency weight distribution matrix corresponding to a screen pixel coordinate system through normalization mapping; S3, taking a visual saliency weight distribution matrix as a probability density function, executing Monte Carlo importance sampling in a visible area to generate a non-uniform seed point set with density changing along with weight, executing weighted Thiessen polygon subdivision based on the seed point set to generate a space geometry constraint type self-adaptive viewport topology grid, wherein a high-weight area generates a small-area high-frequency sampling viewport due to high-density distribution of seed points, and a low-weight area generates a large-area low-frequency sampling viewport due to low-density distribution of seed points; s4, taking a space geometry constraint type self-adaptive view port topological grid as a data projection container, resampling and mapping the global human flow interaction entropy field data into each view port unit, namely, in a small-area high-frequency sampling view port, reserving vector characteristics of the entropy field data, executing flow field line integral convolution rendering to reconstruct microscopic track dynamics, and in a large-area low-frequency sampling view port, executing scalar homogenization processing of the entropy field data, generating a tone filling layer reflecting an area average pressure value, and combining to generate a human flow dynamic visualization interaction display interface with multi-granularity data fusion.
2. The method for generating the dynamic visualized interactive display interface of the traffic of people according to claim 1, wherein the method is characterized in that collected time series data of the trajectories of people are subjected to vectorization analysis, trajectory fragments are intercepted through a sliding time window, the included angle dispersion and the speed module length attenuation rate between adjacent trajectory vectors are calculated, and the specific process for constructing the local interaction entropy value reflecting the disorder degree of microscopic movements of groups is as follows: Traversing the track segment in the sliding time window, extracting an instantaneous speed vector of the track segment at the current moment and an instantaneous speed vector of the previous moment, carrying out vector normalization processing on the instantaneous speed vector and the instantaneous speed vector of the previous moment, calculating dot products of the two normalized vectors to obtain a cosine value of a direction included angle, and counting the variance of the cosine value of the direction included angle in the sliding time window as an included angle dispersion; Meanwhile, calculating a first-order differential sequence of the velocity modular length of the track segment in the sliding time window, and counting the proportion of the accumulated amplitude of the negative elements in the first-order differential sequence to the total sequence amplitude as the velocity modular length attenuation rate; And carrying out weighted summation on the included angle dispersion and the speed module length attenuation rate to generate a disorder index of a single track, counting the arithmetic average value of the disorder index of the track in the unit physical space grid, and taking the arithmetic average value as the local interaction entropy value of the grid center.
3. The method for generating a dynamic visualized interactive display interface of people flow according to claim 2, wherein the specific process of mapping discrete local interaction entropy values to a physical space grid unit and generating global people flow interaction entropy field data by performing inverse distance weighted interpolation operation is as follows: Establishing a coordinate index matrix of the physical space grid units, determining the geometric center coordinates of each grid unit to be interpolated, setting a search radius by taking the geometric center coordinates as the circle center, and searching reference grid points with local interaction entropy values existing in the search radius; calculating Euclidean distance between geometric center coordinates of grid units to be interpolated and reference grid point coordinates, constructing a weight attenuation function based on the inverse of the Euclidean distance, and calculating an interpolation weight coefficient of each reference grid point by using the weight attenuation function; And carrying out weighted linear combination on the local interaction entropy value of the reference grid point in the search radius and the corresponding interpolation weight coefficient, assigning a calculation result to a grid unit to be interpolated, and outputting global people stream interaction entropy field data after traversing calculation is completed on the grid unit.
4. The method for generating the dynamic visualized interactive display interface of pedestrian traffic according to claim 1, wherein step S2 comprises the following steps: Carrying out multi-level Gaussian blur processing on the global people stream interaction entropy field data, respectively generating a first Gaussian blur image and a second Gaussian blur image by using Gaussian kernel functions of different scales, and carrying out pixel-level difference operation on the first Gaussian blur image and the second Gaussian blur image to generate a Gaussian difference response diagram; traversing the pixel points of the Gaussian differential response graph, comparing the response values of the current pixel point and the eight neighborhood pixel points, and marking the points with the response values larger than the eight neighborhood pixel point response values as spatial extreme points; Calculating horizontal gradient components and vertical gradient components of each pixel point in the Gaussian differential response diagram, synthesizing a gradient amplitude matrix, performing spatial superposition on position information of spatial extreme points and the gradient amplitude matrix, mapping a superposition result to a numerical interval from 0 to 1 through linear normalization, and generating a visual saliency weight distribution matrix.
5. The method for generating a dynamic visualized interactive display interface of traffic according to claim 1, wherein the specific process of generating a non-uniform seed point set with density varying with weight by performing Monte Carlo importance sampling in a visible area by using a visual saliency weight distribution matrix as a probability density function is as follows: Performing global value accumulation on the visual saliency weight distribution matrix, taking an accumulation result as a normalization factor, and performing division normalization operation on the weight values of elements in the matrix to construct a two-dimensional accumulation probability distribution function; Generating a group of pseudo-random number sequences subject to uniform distribution, taking each random number in the pseudo-random number sequences as an index key, executing reverse search operation in a two-dimensional cumulative probability distribution function, and positioning the corresponding matrix coordinate position; And extracting matrix coordinate positions locked by reverse search operation, removing repeated coordinates and boundary overflow coordinates, and recording the rest coordinate position set as a non-uniform seed point set with density changing along with weight.
6. The method for generating a dynamic visualized interactive display interface of traffic according to claim 5, wherein the specific process of generating a spatial geometry constrained adaptive viewport topology grid by performing weighted Thiessen polygon subdivision based on the seed point set is as follows: reading the coordinate position of each seed point in the non-uniform seed point set, and reading the weight value corresponding to the coordinate position in the visual saliency weight distribution matrix as the growth potential energy parameter of the seed point; Traversing screen pixel points in a visible area, quantifying Euclidean space adjacency between the screen pixel points and seed points, introducing corresponding growth potential energy parameters, performing potential field intensity modulation and bias correction operation on the Euclidean space adjacency, and constructing a power distance metric value representing expansion competitiveness of a viewport area; Comparing the power distance measurement values of the seed points to the current screen pixel points, classifying the current screen pixel points to the index area to which the seed points with the minimum power distance measurement values belong, extracting pixel boundary lines between different index areas after the area classification of the full-screen pixel points is completed, and constructing a space geometry constraint type self-adaptive viewport topology grid surrounded by polygon boundaries.
7. The method for generating the dynamic visualized interactive display interface of the people flow according to claim 1 is characterized by taking a space geometry constraint type self-adaptive viewport topological grid as a data projection container, and resampling and mapping global people flow interactive entropy field data into each viewport unit, wherein in a small-area high-frequency sampling viewport, vector features of the entropy field data are reserved, and a specific process of performing flow field line integral convolution rendering to reconstruct microscopic track dynamics is as follows: generating a Gaussian white noise texture image with the same size as a small-area high-frequency sampling viewport; Traversing each pixel point in the high-frequency sampling viewport, taking the vector direction in the global people stream interaction entropy field data as a guide, and executing bidirectional streamline integration from the current pixel point to obtain a streamline track coordinate sequence passing through the pixel point; And reading the corresponding gray value of the streamline track coordinate sequence in the white noise texture image, calculating a convolution weighted average value of the corresponding gray value, and assigning the convolution weighted average value to the current pixel point to generate a microscopic track dynamic rendering layer with fluid texture characteristics.
8. The method for generating the dynamic visualized interactive display interface of the traffic of people according to claim 7, wherein the specific process of generating the dynamic visualized interactive display interface of the traffic of people with multiple granularities data fusion by executing scalar homogenization processing of entropy field data in a large-area low-frequency sampling viewport and generating a tone filling layer reflecting an area average pressure value is as follows: Counting scalar magnitudes of global people flow interaction entropy field data in a physical space grid unit covered by a large-area low-frequency sampling viewport, and calculating an arithmetic average of the scalar magnitudes as a unified pressure value of the viewport unit; inputting the unified pressure value into a preset color lookup table for linear interpolation mapping, obtaining corresponding RGBA color channel parameters, and filling the low-frequency sampling viewport by utilizing the RGBA color channel parameters to generate a tone filling layer; An image synthesis buffer area is established, a tone filling layer is used as a background layer, a micro-track dynamic rendering layer is used as a foreground layer, an area ID of a space geometric constraint type self-adaptive viewport topological grid is read and used as a mask channel, the foreground layer is added to the background layer through an Alpha mixing algorithm, and a human flow dynamic visual interactive display interface with multi-granularity data fusion is output.
9. The system for generating the dynamic visualized interactive display interface of the traffic is applied to the method for generating the dynamic visualized interactive display interface of the traffic according to any one of claims 1 to 8, and is characterized by comprising the following modules: The entropy field construction module is used for carrying out vectorization analysis on the acquired personnel track time sequence data, intercepting track fragments through a sliding time window, calculating the included angle dispersion and the speed modulus attenuation rate between adjacent track vectors, constructing a local interaction entropy value reflecting the microscopic motion disorder degree of the group, carrying out inverse distance weighted interpolation operation, mapping the discrete local interaction entropy value to a physical space grid unit, and generating global people stream interaction entropy field data; The salient weight calculation module is used for executing multi-scale Gaussian difference convolution operation on the global people stream interaction entropy field data, extracting spatial extreme points and edge gradient characteristics of local entropy values, and constructing a visual salient weight distribution matrix corresponding to a screen pixel coordinate system through normalized mapping; The visual port topology generation module is used for taking a visual saliency weight distribution matrix as a probability density function, executing Monte Carlo importance sampling in a visible area to generate a non-uniform seed point set with density changing along with weight, executing weighted Thiessen polygon subdivision based on the seed point set to generate a space geometry constraint type self-adaptive visual port topology grid, wherein a high-weight area generates a small-area high-frequency sampling visual port due to high-density distribution of seed points, and a low-weight area generates a large-area low-frequency sampling visual port due to low-density distribution of seed points; The interface rendering generation module is used for taking a space geometry constraint type self-adaptive view port topological grid as a data projection container, re-sampling and mapping the global human flow interaction entropy field data into each view port unit, wherein vector characteristics of the entropy field data are reserved in a small-area high-frequency sampling view port, flow field line integral convolution rendering is executed to reconstruct microscopic track dynamics, scalar homogenization processing of the entropy field data is executed in a large-area low-frequency sampling view port, a tone filling layer reflecting an area average pressure value is generated, and a human flow dynamic visualization interaction display interface with multi-granularity data fusion is generated by combination.

Description

Method and system for generating dynamic visualized interactive display interface of people flow Technical Field The invention relates to the technical field of data visualization, in particular to a method and a system for generating a dynamic visual interactive display interface for human flow. Background With the acceleration of the urban process, the people's mobile density in large public places such as transportation hubs, commercial complexes and stadiums has increased significantly. Under the complex scenes, the crowd flowing state is monitored and analyzed in real time, and the method has important significance in guaranteeing public safety, optimizing space layout and improving emergency response efficiency. The manager needs to quickly perceive the scene situation through a visual interface, and extracts potential congestion risks and abnormal behavior patterns from massive track data, so that the transition from data passive monitoring to active decision making is realized. However, existing methods of visualizing traffic typically employ fixed resolution grid thermodynamic diagrams or globally unified vector flow field rendering. The uniform rendering mode has the obvious and substantial defects when processing non-uniformly distributed people stream data, namely, on one hand, in a high-density aggregation area, a grid with fixed resolution can cause pixel aliasing and visual shielding of motion details of microscopic individuals, so that a manager is difficult to distinguish whether people are in an orderly queuing or unordered pushing state, and on the other hand, in a low-density open area, a large amount of graphics processor computing resources are occupied by overall uniform high-precision rendering, and the system frame rate is reduced, so that rendering delay is generated. In addition, the existing interface generation logic is usually preset statically, and the allocation strategy of the screen space cannot be dynamically adjusted according to the chaotic degree of the real-time data, so that key risk information is easily submerged in a large amount of redundant background data, and the accurate interaction requirement under a complex dynamic scene cannot be met. Disclosure of Invention Aiming at the defects of the prior art, the invention provides a method and a system for generating a dynamic visualized interactive display interface of the traffic flow, which solve the problems of the background art. S1, carrying out vectorization analysis on collected personnel track time sequence data, intercepting track fragments through a sliding time window, calculating included angle dispersion and speed module length attenuation rate between adjacent track vectors, constructing a local interaction entropy value reflecting the disorder degree of group microscopic movement, carrying out inverse distance weighted interpolation operation to map the discrete local interaction entropy value to a physical space grid unit, and generating global personnel flow interaction entropy field data; S2, performing multi-scale Gaussian differential convolution operation on global human flow interaction entropy field data, extracting spatial extreme points and edge gradient characteristics of local entropy values, constructing a visual saliency weight distribution matrix corresponding to a screen pixel coordinate system through normalization mapping, S3, taking the visual saliency weight distribution matrix as a probability density function, performing Monte Carlo importance sampling in a visible area to generate a non-uniform seed point set with density changing along with weight, performing weighted Thiessen polygon subdivision based on the seed point set to generate a spatial geometry constraint type self-adaptive view port topological grid, wherein a high-weight area generates a small-area high-frequency sampling view port due to high density distribution of seed points, a low-weight area generates a large-area low-frequency sampling view port due to low density distribution of seed points, S4, re-mapping global human flow interaction entropy field data into each view port unit by taking the spatial geometry constraint type self-adaptive view port topological grid as a data projection container, reserving vector characteristics of entropy field data in the small-area high-frequency sampling view port, performing flow field line integral convolution to reconstruct dynamic tracks in the large-area low-frequency sampling view port, and executing scalar homogenization processing of the entropy field data, generating a tone scale filling layer reflecting the average pressure value of the region, and combining to generate the multi-granularity data fusion human flow dynamic visual interactive display interface. The method comprises the steps of collecting personnel track time sequence data, carrying out vectorization analysis on the collected personnel track time sequence data, intercepting track fragments throug