Intelligent total station multi-station data fusion method and system

Abstract

This invention relates to the field of intelligent surveying technology, specifically to a method and system for multi-station data fusion using an intelligent total station. The method includes: constructing a dynamic edge weight model representing the spatiotemporal dynamic correlation between stations, and generating edge weights; wherein the dynamic edge weight model generates edge weights by setting adaptive adjustment parameters to adjust the proportion of distances between stations and the variance of station coordinate data; adaptively adjusting the adaptive adjustment parameters through chaotic adaptive collaborative optimization to obtain the optimal adaptive adjustment parameters; generating edge weights based on the optimal adaptive adjustment parameters; and weighting and fusing the data from multiple stations based on the edge weights. This solution, by constructing a model representing the spatiotemporal dynamic correlation between stations and using chaotic adaptive collaborative optimization, can dynamically allocate weights, adapt to dynamic environmental changes, reduce errors, and improve the accuracy and robustness of multi-station data fusion.

Description

Technical Field

[0001] This invention relates to the field of intelligent surveying and mapping technology, specifically to a method and system for fusing multi-station data from an intelligent total station. Background Technology

[0002] As a core piece of equipment in engineering surveying, the total station's multi-station collaborative measurement technology has always been a research hotspot. However, existing technologies face the following key challenges in multi-station data fusion: Limitations of static fusion models: Traditional methods, such as least squares and Kalman filtering, rely on fixed mathematical models to allocate station weights, which cannot adapt to dynamic environmental changes. For example, in occlusion scenarios, the confidence level of data from a certain station drops sharply, but traditional models still fuse data according to preset weights, leading to error accumulation. Error propagation and accumulation issues: Multi-station data fusion usually requires global alignment through coordinate transformation models, but existing models, such as the seven-parameter transformation method, are sensitive to local errors. Insufficient real-time performance and automation: Existing technologies rely on manual screening of abnormal data or adjustment of model parameters, which is inefficient.

[0003] The main reason for the above problems is that existing technologies have failed to effectively model the spatiotemporal dynamic correlation characteristics of total station data, and lack systematic solutions for non-uniform error distribution, dynamic environmental interference, and multi-objective (such as accuracy and real-time performance) collaborative optimization.

[0004] Therefore, there is an urgent need for a method and system for fusion of multi-station data from intelligent total stations. By constructing a model that represents the spatiotemporal dynamic relationship between each station and using chaotic adaptive collaborative optimization, weights can be dynamically allocated to adapt to dynamic environmental changes, reduce errors, and improve the accuracy and robustness of multi-station data fusion. Summary of the Invention

[0005] One of the objectives of this invention is to provide a method for fusing multi-station data from an intelligent total station. By constructing a model that characterizes the spatiotemporal dynamic correlation between the stations and implementing chaotic adaptive collaborative optimization, the method can dynamically allocate weights, adapt to dynamic environmental changes, reduce errors, and improve the accuracy and robustness of multi-station data fusion.

[0006] The basic solution provided by this invention is a method for fusing multi-station data from an intelligent total station, comprising the following: A dynamic edge weight model is constructed to characterize the spatiotemporal dynamic relationship between various stations, and the edge weights between various stations are generated. The dynamic edge weight model generates edge weights by setting adaptive adjustment parameters to adjust the distance between stations and the proportion of variance of station coordinate data. Through chaotic adaptive collaborative optimization, the adaptive adjustment parameters are adaptively adjusted to obtain the optimal adaptive adjustment parameters. The edge weights are generated based on the optimal adaptive adjustment parameters, and the data from multiple stations are fused in a weighted manner based on the edge weights.

[0007] The second objective of this invention is to provide an intelligent total station multi-station data fusion system. By constructing a model representing the spatiotemporal dynamic correlation between each measurement and using chaotic adaptive collaborative optimization, the system can dynamically allocate weights, adapt to dynamic environmental changes, reduce errors, and improve the accuracy and robustness of multi-station data fusion.

[0008] This invention provides a second basic solution: a multi-station data fusion system for intelligent total stations, used to execute the above-mentioned multi-station data fusion method for intelligent total stations, including: a server; The server connects to each measuring station to construct a dynamic edge weight model that represents the spatiotemporal dynamic relationship between the measuring stations and generates edge weights between the measuring stations. The dynamic edge weight model generates edge weights by setting adaptive adjustment parameters to adjust the distance between the measuring stations and the proportion of variance of the measuring station coordinate data. Through chaotic adaptive collaborative optimization, the adaptive adjustment parameters are adaptively adjusted to obtain the optimal adaptive adjustment parameters. The edge weights are generated based on the optimal adaptive adjustment parameters, and the data from multiple stations are fused in a weighted manner based on the edge weights.

[0009] Beneficial effects: This scheme transforms the data fusion problem between various stations into a graph node fusion problem. The proportion of data between stations in the fusion process can be regarded as edge weights. Therefore, a dynamic edge weight model of each station is constructed to represent the spatiotemporal dynamic relationship between various stations. This model is used to dynamically generate edge weights to adapt to changes in the dynamic environment, reallocate weights, and reduce errors and error accumulation in the data fusion process. Specifically, the dynamic edge weight model generates edge weights by setting adaptive adjustment parameters to regulate the proportion of distances between stations and the variance of station coordinate data. The stability of the variance data is ensured by the adaptive adjustment parameters, which can adjust the proportions of different indicators (distance, variance, etc.) to accurately reflect spatiotemporal dynamic correlations. The adaptive adjustment parameters, through chaotic adaptive collaborative optimization, can perform spatial search to adaptively adjust the parameters, obtain the optimal parameters, generate edge weights based on the optimal parameters, and then weight and fuse data from multiple stations according to the edge weights. This improves the accuracy and robustness of multi-station data fusion. Especially in occlusion scenarios, chaotic adaptive collaborative optimization can significantly improve fusion accuracy. In practical applications, this solution is suitable for complex scenarios. In densely populated urban areas, it effectively addresses signal obstruction and multipath interference through dynamic edge weight allocation. For disaster monitoring, it supports the rapid deployment of monitoring stations by drones, and the chaotic adaptive collaborative optimization achieves autonomous fusion in a network-free environment. It eliminates the need for manual intervention in abnormal data processing, saving manpower costs.

[0010] In summary, this solution, by constructing a model representing the spatiotemporal dynamic correlation between measurements and employing chaotic adaptive collaborative optimization, can dynamically allocate weights, adapt to dynamic environmental changes, and reduce errors. It solves the problem of real-time fusion of multi-source data from total stations in unstructured environments (such as dynamic occlusion, signal fluctuations, and instrument drift), and is particularly suitable for high-precision measurement scenarios such as densely populated urban areas, mine monitoring, and disaster emergency response. Through dynamic modeling and intelligent optimization, it improves data accuracy and system robustness. Attached Figure Description

[0011] Figure 1 This is a flowchart illustrating an embodiment of the intelligent total station multi-station data fusion method of the present invention. Detailed Implementation

[0012] The following detailed description illustrates the specific implementation method: Example 1 This embodiment is basically as shown in the appendix. Figure 1 As shown, a method for fusing multi-station data from an intelligent total station is provided, including the following: Data and attributes of each station, as well as correlation data between stations, are collected. The data includes coordinates; the correlation data includes distances between stations; station attributes include real-time location and environmental interference factors; environmental interference factors include temperature, humidity, and vibration amplitude. In other embodiments, stations can be mapped as nodes. The data is set according to the actual data to be fused, and other data besides coordinates can be fused. In this embodiment, the data is coordinates, i.e., the coordinates of multiple stations are fused. A dynamic edge weight model representing the spatiotemporal dynamic relationship between each station is constructed, and the edge weights between each station are generated. The dynamic edge weight model generates edge weights by setting adaptive adjustment parameters to adjust the distance between stations and the proportion of variance in station coordinate data. The dynamic edge weight model is as follows: in, For the station and Distance, in meters; and For the station and The variance of coordinate data reflects data stability; unit: mm. 2 ; , , For adaptive adjustment parameters.

[0013] Through chaotic adaptive collaborative optimization, the adaptive adjustment parameters are adaptively adjusted to obtain the optimal adaptive adjustment parameters. The edge weights are generated based on the optimal adaptive adjustment parameters, and the data from multiple stations are fused in a weighted manner based on the edge weights. The specific process is as follows: Chaos initialization steps: Set up chaotic mapping relationships and initialize the number of iterations. This generates initial adaptive adjustment parameters, and then generates the initial weight distribution; The chaotic mapping relationship is as follows: in For the first The chaotic state value of the sub-mapping has a range of (0,1). In chaotic mappings, the number of mappings is represented. This represents the initial value, with a range of (0, 1), and is used for initial assignment. =0.4, at which point the data discreteness is maximized, which maximizes the requirements of the optimization algorithm. Based on the mathematical properties of the new chaotic mapping, this point can maximize the chaotic discreteness. However, the existing Sine mapping has low sensitivity to initial values, which may lead to insufficient initialization diversity. The initial adaptive adjustment parameters are: ; The novel chaotic mapping relationship set up in this scheme generates an initial weight distribution, enhancing the diversity of the search space.

[0014] Real-time weight calculation and data fusion steps: The current adaptive adjustment parameters... Substitute the edge weights into the dynamic edge weight model to calculate the edge weights. The data is then weighted and fused according to the edge weights to generate fused data. The specific process is as follows: Global data fusion in fused data That is, globally fused coordinate data, calculated by weighted average of the coordinates of all stations: ; Among them, the measuring station Its fusion weight For it and all adjacent stations The sum of edge weights; , station The original coordinates, Indicates the station The set of neighboring nodes.

[0015] The adaptive adjustment parameter optimization steps are as follows: The chaotic control parameters are updated according to the environmental disturbance factor, and the adaptive adjustment parameters are iteratively optimized. If the termination condition is met, the adaptive adjustment parameters of the current iteration are taken as the optimal adaptive adjustment parameters. Edge weights are generated according to the optimal adaptive adjustment parameters, and the data from multiple stations are fused in a weighted manner according to the edge weights. The termination condition is that the number of iterations is met, or the fusion difference between the fused data and the measured data is less than a preset threshold. The specific process is as follows: Calculate the fusion difference : ; in This represents the average of the measured values. Determine if the termination condition is met; if so, adjust the current adaptive parameters. The optimal adaptive adjustment parameter is selected and output; otherwise, it is updated. Update the chaotic control parameters, generate new adaptive adjustment parameters, and then perform real-time weight calculation and data fusion steps. The termination condition is as follows: ; in To preset the threshold for the difference between the before and after fusion, In this embodiment, the threshold number of iterations is used. , .

[0016] The current adaptive adjustment parameter is output. That is, the entire optimization process The minimum group parameter (adaptive adjustment parameter), and the global fused coordinates corresponding to the optimal adaptive adjustment parameter. The final fused coordinates are obtained by weighting and fusing data from multiple stations based on edge weights.

[0017] This involves updating the chaotic control parameters and generating new adaptive adjustment parameters: Update chaos control parameters: ; Chaos control parameters Used to control the convergence speed, with an initial value of 0.1. Decreasing this value refines the local development. The environmental stability index is calculated from vibration amplitude, temperature, and humidity data, ranging from [0,1], with larger values ​​indicating greater stability; specifically: ; in These are weighting coefficients, which are set to 0.2, 0.5, and 0.3 in this embodiment. The amplitude of the vibration is expressed in grams (g). For the rate of temperature change, The rate of change of humidity; The learning rate is set to 0.1 in this embodiment. This approach couples environmental stability with chaotic parameters, which improves the convergence speed compared to traditional methods. New adaptive adjustment parameters are generated through chaotic perturbation. : ; The chaotic mapping relationship set in this scheme is a novel chaotic mapping relationship, which is different from the traditional Sine chaotic mapping. (in As an external control parameter, typically 4), several improvements have been made, as detailed below: Nonlinear Enhancement: The novel chaotic mapping relationship introduces an additional nonlinear term within the sine function. This nonlinear term is similar to the core structure of the Logistic Map, and this combination creates more complex dynamic behavior, while existing Sine chaotic maps rely only on simple sine functions; Parameter simplification: Existing Sine chaotic mappings require external control parameters. The new mapping requires no external parameters, reducing the complexity of parameter tuning and improving the adaptability of chaotic mapping. Initial value optimization: Based on the mathematical characteristics of the new mapping, this scheme can set a more suitable initial value for the current situation, such as x(1)=0.4, which can maximize the chaotic discreteness. However, the existing Sine mapping has low sensitivity to initial values, which may lead to insufficient initialization diversity.

[0018] The novel chaotic mapping relationship brings significantly better results in the process of chaotic initialization and optimization, specifically including: Greater chaos and diversity: The novel chaotic mapping relationship has a wider range of chaos and a higher Lyapunov exponent, which makes the generated sequence more random and uniformly cover the search space, directly enhancing the diversity of the initial weight distribution and avoiding premature convergence of the optimization algorithm and getting stuck in local optima.

[0019] Faster convergence speed: In chaotic adaptive cooperative optimization, novel chaotic mapping relationships can generate effective parameter candidate values ​​more quickly. This reduces global fusion differences. The number of iterations; the efficient chaotic perturbation of the novel chaotic mapping relationship improves the convergence speed compared with traditional methods; Better stability: The novel chaotic mapping relationship has a smoother dependence on initial values, resulting in improved environmental stability metrics. When changes occur, it can generate chaotic sequences more robustly, reducing the risk of the algorithm diverging due to environmental interference.

[0020] The superior effects described above are due to the mathematical properties and structural design of the novel chaotic mapping relationship. Specifically, the nonlinear superposition effect occurs: the sine function provides periodic ergodicity, while the logistic mapping... Introducing nonlinear feedback, the superposition of the two enhances the hybridity and ergodicity of the chaotic map, which makes the sequence more difficult to predict and less likely to have periodic windows, thus avoiding getting trapped in a periodic orbit. Complexity of dynamical systems: By analyzing the Lyapunov exponent, the new chaotic mapping relationship usually maintains a positive value (indicating chaos) in the range x(t)∈(0,1), and the value is higher, while the existing Sine mapping is chaotic when the parameter a=4, but the chaotic intensity is weaker; the derivative of the new chaotic mapping is more complex and the sensitivity to dependence is stronger.

[0021] Initial value optimization design: The choice of x(1)=0.4 is based on bifurcation analysis of a new type of mapping. Near this point, the mapping can quickly enter a chaotic state, maximizing the dispersion of the initial population.

[0022] The novel chaotic mapping relationship, as a core component of chaotic adaptive cooperative optimization, has had a profound impact on this scheme, including: Enhanced global optimization capabilities: In chaotic initialization, the new chaotic mapping relationship generates a more diverse set of initial parameters, which makes subsequent weight calculation and data fusion more accurate and reduces the impact of initial deviation on global fusion coordinates; Enhancing Dynamic Adaptability: Chaotic Perturbation and Environmental Stability Indicators of Novel Chaotic Mapping Relationships in Chaotic Adaptive Co-optimization Coupling enables more intelligent parameter updates, allowing parameters to be quickly adjusted and fusion accuracy maintained when environmental disturbances (such as temperature, vibration, and humidity) change. Improved termination efficiency: Due to the increased convergence speed, the iteration meets the termination condition earlier, saving computational resources and improving the real-time performance of the system.

[0023] The novel chaotic mapping relationship interacts closely with other steps in this scheme, creating a synergistic effect: The adaptive adjustment parameters generated during the initialization of the chaotic mapping relationship are substituted into the dynamic edge weight model to calculate the edge weights. ,in Depends on , , Therefore, the initialization is more diverse, and the edge weight calculation can better reflect the real spatiotemporal relationship between the stations, thereby improving the foundation of data fusion; Chaotic mappings indirectly affect edge weights by optimizing chaotic control parameters. and global fusion coordinates Better adaptive adjustment parameters can lead to more accurate weighted fusion and reduce fusion error. ; The novel chaotic mapping relationship can be directly used to generate new candidate values ​​for adaptive adjustment parameters. Furthermore, by coupling chaotic control parameters with environmental stability, a feedback loop is formed: environmental changes affect... , Adjusting the chaotic perturbation updates the adaptive adjustment parameters, which in turn optimize the edge weights and data fusion results. The novel chaotic mapping relationship leads to faster convergence, which means reaching the termination condition earlier and outputting the optimal parameters. The final fusion coordinates improved overall efficiency.

[0024] In summary, the novel chaotic mapping relationship enhances the chaotic characteristics by introducing nonlinear superposition, thereby improving the diversity, convergence speed, and stability of optimization. It not only directly improves chaotic initialization but also affects the performance of the dynamic edge weight model through parameter optimization chain. Compared with the existing Sine mapping, the novel chaotic mapping relationship is more suitable for the real-time data fusion needs in complex environments.

[0025] Taking emergency monitoring of dangerous rockfalls as an example: First, the monitoring stations are deployed: based on the terrain and visibility, five intelligent total stations (with built-in temperature, humidity, and vibration sensors) are set up around the dangerous rock as monitoring stations; for each monitoring point, the five intelligent total stations simultaneously measure its coordinates, once every 0.5 hours. The power supply for the monitoring stations adopts a hybrid mode of solar energy and lithium batteries to ensure long-term continuous operation, and the data is automatically transmitted.

[0026] Next, multi-station data fusion of the intelligent total station is performed: the data collected by each station is mapped to nodes; a dynamic edge weight model between nodes is constructed to represent the spatiotemporal dynamic relationship between each station; through chaotic adaptive collaborative optimization, the adaptive adjustment parameters are adaptively adjusted to obtain the optimal adaptive adjustment parameters, the edge weights generated based on the optimal adaptive adjustment parameters, and the data from multiple stations are weighted and fused based on the edge weights. Chaos initialization: =0.3, =0.03, =0.1; Real-time weight calculation and data fusion: For each station Calculate its relationship with neighboring stations edge weight For example, the edge weights between station 3 and its neighbor station 5: ; Total edge weight of aggregated station 3: When station 3 is blocked ( When the weight increases from 0.4 to 2.1, its total weight decreases to =0.483 (a decrease of 60%); The updated edge weights are used to calculate the global fused coordinates.

[0027] Adaptive parameter optimization: The optimization process is shown in Table 1: Table 1: Iteration Diagram Edge weight application and data fusion output: Using the optimal adaptive adjustment parameters, the final edge weights are calculated, the fusion weights of each station are aggregated, and the final fused coordinate results are output.

[0028] This embodiment also provides an intelligent total station multi-station data fusion system for executing the above-described intelligent total station multi-station data fusion method, including: a server; The server connects to each measuring station and is used to map the data collected by each measuring station into nodes. A dynamic edge weight model for each node is constructed to represent the spatiotemporal dynamic relationship between each station, and edge weights are generated. The dynamic edge weight model generates edge weights by setting adaptive adjustment parameters to adjust the distance between stations and the proportion of variance of station coordinate data. Through chaotic adaptive collaborative optimization, the adaptive adjustment parameters are adaptively adjusted to obtain the optimal adaptive adjustment parameters, the edge weights generated based on the optimal adaptive adjustment parameters, and the data from multiple stations are weighted and fused based on the edge weights.

[0029] The above descriptions are merely embodiments of the present invention. Commonly known structures and characteristics are not described in detail here. Those skilled in the art are aware of all common technical knowledge in the field prior to the application date or priority date, are aware of all existing technologies in that field, and have the ability to apply conventional experimental methods prior to that date. Those skilled in the art can, under the guidance of this application, improve and implement this solution in combination with their own capabilities. Some typical known structures or methods should not be obstacles for those skilled in the art to implement this application. It should be noted that those skilled in the art can make several modifications and improvements without departing from the structure of the present invention. These should also be considered within the scope of protection of the present invention, and will not affect the effectiveness of the implementation of the present invention or the practicality of the patent. The scope of protection claimed in this application should be determined by the content of its claims, and the specific embodiments described in the specification can be used to interpret the content of the claims.

Claims

1. A method for fusing multi-station data from an intelligent total station, characterized in that, Includes the following: A dynamic edge weight model representing the spatiotemporal dynamic relationship between each station is constructed, and the edge weights between each station are generated. The dynamic edge weight model generates edge weights by setting adaptive adjustment parameters to adjust the distance between stations and the proportion of variance in station coordinate data. Through chaotic adaptive collaborative optimization, the adaptive adjustment parameters are adaptively adjusted to obtain the optimal adaptive adjustment parameters. The edge weights are generated based on the optimal adaptive adjustment parameters, and the data from multiple stations are fused in a weighted manner based on the edge weights. The dynamic edge weight model is as follows: in, For the station and The distance; and For the station and The variance of the coordinate data; , , For adaptive adjustment parameters; The adaptive adjustment through chaotic adaptive collaborative optimization includes: Chaos initialization steps: Set up chaotic mapping relationships, initialize the number of iterations, and generate initial adaptive adjustment parameters; Real-time weight calculation and data fusion steps: Input the current adaptive adjustment parameters into the dynamic edge weight model, calculate the edge weights, and perform weighted fusion of the data according to the edge weights to generate fused data; The adaptive adjustment parameter optimization steps are as follows: Based on the obtained environmental disturbance factors, the chaotic control parameters are updated, and the adaptive adjustment parameters are iteratively optimized. If the termination condition is met, the adaptive adjustment parameters of the current iteration are taken as the optimal adaptive adjustment parameters, and the edge weights generated based on the optimal adaptive adjustment parameters and the data from multiple stations are obtained by weighted fusion based on the edge weights. The chaotic mapping relationship is as follows: in For the first The chaotic state value of the sub-mapping.

2. The method for fusing multi-station data from an intelligent total station according to claim 1, characterized in that: The termination condition is that the iteration number threshold is met, or the absolute value of the difference between the fusion difference of the two consecutive fused data and the measured data is less than the preset fusion difference threshold.

3. The method for fusing multi-station data from an intelligent total station according to claim 1, characterized in that: The number of initialization iterations ; The initial adaptive adjustment parameter is: 。 4. The method for fusing multi-station data from an intelligent total station according to claim 3, characterized in that: The real-time weight calculation and data fusion steps include: Global fused data is calculated by weighted averaging of coordinates from all stations. : ； Among them, the measuring station Its fusion weight For it and all adjacent stations The sum of edge weights; For the station The original data, Indicates the station The set of neighboring stations.

5. The method for fusing multi-station data from an intelligent total station according to claim 4, characterized in that: The adaptive adjustment parameter optimization step includes: Calculate the fusion difference : ； in This represents the average of the measured values. Determine if the termination condition is met; if so, adjust the parameters adaptively according to the current iteration. As the optimal adaptive adjustment parameter, obtain the edge weights generated based on the optimal adaptive adjustment parameter, and the data from multiple stations weighted and fused based on the edge weights; otherwise, update... The chaotic control parameters are updated, new adaptive adjustment parameters are generated, and then real-time weight calculation and data fusion steps are performed.

6. The method for fusing multi-station data from an intelligent total station according to claim 5, characterized in that: The chaotic control parameters are updated to generate new adaptive adjustment parameters: Update chaos control parameters: ； Chaos control parameters Used to control the convergence rate; It is an environmental stability indicator, calculated from vibration amplitude, temperature, and humidity. New adaptive adjustment parameters are generated through chaotic perturbation. : 。 7. A multi-station data fusion system for intelligent total stations, characterized in that, For executing the intelligent total station multi-station data fusion method as described in any one of claims 1-6, the method includes: a server; The server connects to each measuring station to construct a dynamic edge weight model that represents the spatiotemporal dynamic relationship between the measuring stations and generates edge weights. The dynamic edge weight model generates edge weights by setting adaptive adjustment parameters to adjust the distance between measuring stations and the proportion of variance of the measuring station coordinate data. Through chaotic adaptive collaborative optimization, the adaptive adjustment parameters are adaptively adjusted to obtain the optimal adaptive adjustment parameters, the edge weights generated based on the optimal adaptive adjustment parameters, and the data from multiple stations are weighted and fused based on the edge weights. The dynamic edge weight model is as follows: in, For the station and The distance; and For the station and The variance of the coordinate data; , , For adaptive adjustment parameters; The adaptive adjustment through chaotic adaptive collaborative optimization includes: Chaos initialization steps: Set up chaotic mapping relationships, initialize the number of iterations, and generate initial adaptive adjustment parameters; Real-time weight calculation and data fusion steps: Input the current adaptive adjustment parameters into the dynamic edge weight model, calculate the edge weights, and perform weighted fusion of the data according to the edge weights to generate fused data; The adaptive adjustment parameter optimization steps are as follows: Based on the obtained environmental disturbance factors, the chaotic control parameters are updated, and the adaptive adjustment parameters are iteratively optimized. If the termination condition is met, the adaptive adjustment parameters of the current iteration are taken as the optimal adaptive adjustment parameters, and the edge weights generated based on the optimal adaptive adjustment parameters and the data from multiple stations are obtained by weighted fusion based on the edge weights. The chaotic mapping relationship is as follows: in For the first The chaotic state value of the sub-mapping.

Images