Overhead line inspection robot tower-crossing control method, device, equipment and medium

Abstract

The invention belongs to the technical field of overhead line operation and maintenance, and particularly discloses a tower-crossing control method, device and equipment for an overhead line inspection robot and a medium. A tower-crossing control method for an overhead line inspection robot comprises the following steps: acquiring image data of an overhead line in a preset area; inputting the image data into a preset image processing model to obtain a topographic map in a preset area; acquiring position data of the inspection robot, and generating a planning path according to the position data of the inspection robot and the topographic map; and controlling the inspection robot to cross the tower according to the planned path. The topographic map is generated by acquiring the image data for image recognition, and the planned path is generated according to the topographic map, so that the inspection robot is controlled to cross the tower, and compared with the prior art, mechanical arms and manual interference are not needed, and the cost is saved.

Description

Technical Field

[0001] The present invention belongs to the technical field of overhead line operation and maintenance, and in particular relates to a method, device, equipment and medium for controlling an overhead line inspection robot across towers. Background Art

[0002] With the rapid development of the power industry, the safe and stable operation of overhead power lines is essential for economic development and daily life. In order to ensure the stable and safe operation of overhead power lines, regular inspections have become an indispensable link. Traditional inspection methods mainly rely on manual work, but this method not only has a large workload and low efficiency, but also poses great safety risks to inspectors when working at high altitudes.

[0003] In recent years, with the continuous advancement of robot technology, power overhead line inspection robots have emerged, and to a certain extent, they have reduced the intensity and risks of inspection work. However, existing inspection robots still face many challenges when crossing obstacles such as power towers. Most traditional inspection robots use mechanical arms to complete the cross-tower function. This method not only has a complex mechanical structure, but is also prone to jamming and failure during the cross-tower process, which seriously affects the inspection efficiency and the service life and safety of the robot.

[0004] Therefore, developing a control method for an inspection robot that does not rely on a robotic arm and can efficiently and safely cross power towers has become an urgent problem to be solved in the current power industry.

[0005] Defects and shortcomings of the existing technology: At present, the power overhead line inspection robots on the market have the following defects and deficiencies in cross-tower control: 1. The tower-crossing method is single and complex: the existing technology mainly relies on complex mechanical structures, such as telescopic arms, climbing mechanisms, etc. These mechanisms not only increase the weight and cost of the robot, but are also prone to failure during use and have high maintenance costs.

[0006] 2. Low efficiency in crossing towers: Due to the complexity of the mechanical structure, existing robots need to be adjusted and operated multiple times during the process of crossing towers, resulting in a long crossing tower time and low inspection efficiency.

[0007] 3. Insufficient safety: The current cross-tower method has great safety hazards during operation. Once the mechanical structure fails or the operation is improper, it may cause serious accidents such as the robot falling or damaging the line.

[0008] 4. Poor adaptability: The existing cross-tower method has high requirements on the branch environment and requires the reconstruction of existing lines to build walking tracks, which greatly increases the reconstruction cost of existing facilities.

[0009] 5. Low level of intelligence: Existing inspection robots still require a lot of manual intervention during the cross-tower process and cannot achieve true autonomous operation capabilities. Summary of the invention The purpose of the present invention is to provide a method, device, equipment and medium for controlling an overhead line inspection robot across towers, so as to solve the technical problems that the existing overhead line inspection robots need manual assistance to cross towers by setting up mechanical arms, and the mechanical arms are prone to failures and the operation and maintenance costs are high.

[0010] To achieve the above object, the present invention adopts the following technical solutions: In a first aspect, the present invention provides a method for controlling an overhead line inspection robot across towers, comprising the following steps: Acquire image data of overhead lines in a preset area; Inputting the image data into a preset image processing model to obtain a topographic map within a preset area; Obtain the location data of the inspection robot, and generate a planned path based on the location data of the inspection robot and the terrain map; The inspection robot is controlled to cross the tower according to the planned path.

[0011] A further improvement of the present invention is that the preset image processing model comprises the following steps: Get training data; Perform data augmentation on training data; The enhanced training data is divided into training set, validation set and test set; The image processing model is trained using the YOLOv5 network based on the training set and the test set; The image processing model is tested according to the test set. If it passes the test, the image processing model is output; if it fails the test, the training data is re-acquired.

[0012] A further improvement of the present invention is that the step of testing the image processing model according to the test set, outputting the image processing model if the test is passed, and reacquiring the training data if the test is not passed specifically includes: Input the test set into the image processing model and obtain several recognition results; Obtain a pole tower database, judge a number of recognition results according to the pole tower database, and obtain a recognition accuracy rate; If the recognition accuracy is higher than the first threshold, the test passes; If the accuracy is less than or equal to the first threshold, the test fails.

[0013] A further improvement of the present invention is that the data enhancement includes scaling, horizontal flipping, vertical flipping, rotating 90 degrees, translating, darkening, brightening, salt and pepper noise, Gaussian noise and filter transformation.

[0014] A further improvement of the present invention is that the step of controlling the inspection robot to cross the tower according to the planned path includes the following steps: Obtain the planned path, and adjust the flight posture and speed of the inspection robot according to the planned path to complete the cross-tower action.

[0015] A further improvement of the present invention is that the step of obtaining the position data of the inspection robot and generating a planned path according to the position data of the inspection robot and the terrain map includes the following steps: Obtain the location data of the inspection robot and mark the starting point on the topographic map based on the location data; The terrain map shows the location of obstacles and the type of towers; According to the tower type, query the preset tower type database to obtain the location point of the tower; Mark several target nodes on the topographic map according to the location of obstacles and towers; Generate several planned routes based on the starting point and target nodes.

[0016] A further improvement of the present invention is that the tower type database is generated according to tower size data.

[0017] In a second aspect, the present invention provides a cross-tower control device for an overhead line inspection robot, comprising: Data acquisition module: used to acquire image data of overhead lines in a preset area; Image recognition module: used to input image data into a preset image processing model to obtain a topographic map within a preset area; Path generation module: used to obtain the location data of the inspection robot and generate a planned path based on the location data of the inspection robot and the terrain map; Control module: used to control the inspection robot to cross the tower according to the planned path.

[0018] In a third aspect, the present invention provides a computer device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor implements the above-mentioned cross-tower control method for an overhead line inspection robot when executing the computer program.

[0019] In a fourth aspect, the present invention provides a computer-readable storage medium, which stores a computer program. When the computer program is executed by a processor, it implements the above-mentioned cross-tower control method of an overhead line inspection robot.

[0020] Compared with the prior art, the present invention has at least the following beneficial effects: The present invention obtains image data for image recognition to generate a topographic map, and then generates a planned path based on the topographic map, thereby controlling the inspection robot to cross the tower. Compared with the prior art, it does not require a mechanical arm or manual intervention, thus saving costs. The present invention verifies the image recognition model through the accuracy rate, effectively improving the accuracy of image recognition; Autonomy and intelligence: This method uses technologies such as video perception, AI algorithms and drone aircraft to achieve a high degree of autonomy and intelligence, reducing dependence on manual operations.

[0021] Efficiency and accuracy: Through precise algorithm for route planning, the robot can quickly and accurately determine the best path across the tower, activate the drone flight mechanism to accurately complete the tower crossing function, greatly improving the inspection efficiency and accuracy.

[0022] Safety: This method uses precise route planning and mature drone technology, and has the capabilities of real-time judgment and autonomous obstacle avoidance. Problems found during execution can be corrected in a timely manner, ensuring the safety of the robot in the process of crossing towers and reducing the risk of collision or deviation from the route.

[0023] Adaptability: Since the mature UAV aircraft is used as the actuator for tower crossing, it is not restricted by different tower types. In addition, the algorithm can dynamically adjust the route according to the real-time image data, so this method can adapt to all cross-tower scenarios, improving the adaptability and flexibility of the inspection robot. BRIEF DESCRIPTION OF THE DRAWINGS

[0024] The accompanying drawings constituting a part of the present invention are used to provide a further understanding of the present invention. The exemplary embodiments of the present invention and their descriptions are used to explain the present invention and do not constitute an improper limitation of the present invention. In the accompanying drawings: Figure 1 This is a flow chart of a method for controlling a cross-tower overhead line inspection robot according to the present invention; Figure 2 The present invention is a structural block diagram of a cross-tower control device of an overhead line inspection robot. DETAILED DESCRIPTION

[0025] The present invention will be described in detail below with reference to the accompanying drawings and in combination with embodiments. It should be noted that the embodiments and features in the embodiments of the present invention can be combined with each other without conflict.

[0026] The following detailed description is an exemplary description, which is intended to provide further detailed description of the present invention. Unless otherwise specified, all technical terms used in the present invention have the same meaning as those generally understood by those skilled in the art to which the present invention belongs. The terms used in the present invention are only for describing specific embodiments, and are not intended to limit the exemplary embodiments according to the present invention.

[0027] Example 1 A method for controlling an overhead line inspection robot across towers, such as Figure 1 As shown, the following steps are included: S1. Acquire image data of overhead lines in a preset area; S2, inputting the image data into a preset image processing model to obtain a topographic map within a preset area; S3, obtaining the position data of the inspection robot, and generating a planned path according to the position data of the inspection robot and the terrain map; S4. Control the inspection robot to cross the tower according to the planned path.

[0028] Specifically, the inspection robot used in this embodiment includes a camera, a detection module, a communication module, a positioning module and a UAV flight module, etc. The camera is used to obtain image data of the overhead route during the inspection process, and to identify the poles / obstacles based on the image data; the positioning module is used to obtain the location information of the inspection robot to facilitate the subsequent generation of planned paths; the communication module is used for data transmission; the detection module includes several types of sensors for inspection through the detection module, and the types and quantities of sensors in the detection module are set according to the actual inspection situation; the UAV flight module is used to drive the inspection robot to fly, which is convenient for crossing towers and has a better shooting angle.

[0029] Specifically, the preset image processing model includes the following steps: S21, obtaining training data; Specifically, the training data is constructed by annotating photos collected by drones on the spot and open source datasets on the Internet; S22, perform data enhancement on the training data; Specifically, data enhancement in S22 is to expand the data set by adopting geometric transformation and image transformation to enhance the robustness of the model; data enhancement allows limited data to generate more data, increase the number and diversity of training samples (noise data), and improve the robustness of the model. It is generally used for training sets. Neural networks require a large number of parameters. Many neural networks have millions of parameters, and a large amount of data is required for training to make these parameters work correctly. However, in many actual projects, it is difficult for us to find sufficient data to complete the task. Randomly changing training samples can reduce the model's dependence on certain attributes, thereby improving the generalization ability of the model. The data enhancement methods used include: scaling, horizontal flipping, vertical flipping, rotating 90 degrees, translating, darkening, brightening, salt and pepper noise, Gaussian noise, and 12 types of filter transformations; S23, dividing the enhanced training data into a training set, a validation set, and a test set; S24, using the YOLOv5 network to train the image processing model according to the training set and the test set; Specifically, edge detection or contour extraction methods are used in the image processing model to extract features of towers / obstacles, and deep learning models (such as convolutional neural networks) that perform well in the field of image recognition are used to train special algorithms.

[0030] S25. Test the image processing model according to the test set. If the image processing model passes the test, output the image processing model. If the model fails the test, return to step S21.

[0031] Specifically, S25 includes the following steps: S251, input the test set into the image processing model to obtain several recognition results; S252, obtaining a pole tower database, judging a number of recognition results according to the pole tower database, and obtaining a recognition accuracy rate; S253, if the recognition accuracy is higher than the first threshold, the test passes; If the accuracy is less than or equal to the first threshold, the test fails and the process returns to step S21.

[0032] The experiment measures the actual effect of the algorithm test by accuracy. Accuracy is equal to the number of samples N correctly detected by the algorithm in the picture. True Divide by the total number of samples of each category in the image, that is, the number of correctly detected samples N True and the number of samples with detection errors N False , which is specifically expressed as follows:

[0033] Specifically, the pole tower database is a conventional database in the art, and the pole tower database includes the types and sizes of pole towers.

[0034] Specifically, the topographic map shows obstacles, tower types and tower sizes on overhead lines.

[0035] Specifically, in the step of obtaining the position data of the inspection robot and generating a planned path based on the position data of the inspection robot and the terrain map, the current position and the target position are determined based on the perception information and dynamic path planning is performed to generate an optimal route from the current position to safely cross the tower obstacles. The route should ensure that the robot maintains a safe distance from obstacles during the tower crossing process.

[0036] Based on technologies such as GPS+RTK, the precise position of a robot can be obtained, and the pole towers in the field of view can be detected through image data, the corresponding categories can be identified, and their size and area in the image can be determined; based on image processing technology and the absolute position limitation of GPS, the robot's more accurate position and posture relative to the inspection tower can be determined; at the same time, the tower type information determined by the identification information and positioning information can be searched and consulted in the pre-constructed tower type database to obtain a series of key points required for the pole tower; with the current position as the starting point and a series of key points as the target points, dynamic planning of the route is carried out.

[0037] You can also set the objective function to generate the optimal planning path with the minimum moving distance as the goal.

[0038] Specifically, the step of controlling the inspection robot to cross the tower according to the planned path includes the following steps: The inspection robot obtains the planned path and adjusts its flight attitude and speed according to the planned path, so that the robot can fly stably along the planned route and complete the tower crossing action. During the tower crossing process, the robot will automatically identify and avoid obstacles such as wires, poles, and towers to achieve safe and stable flight, and make intelligent decisions, such as automatic route calculation, target identification, and target tracking.

[0039] After the robot successfully crosses the tower obstacle, the robot's position and posture are confirmed through the onboard sensors to ensure the smooth completion of the tower crossing mission and prepare for the next stage of inspection mission.

[0040] Example 2 A cross-tower control device for an overhead line inspection robot, such as Figure 2 As shown, including: Data acquisition module: used to acquire image data of overhead lines in a preset area; Image recognition module: used to input image data into a preset image processing model to obtain a topographic map within a preset area; Path generation module: used to obtain the location data of the inspection robot and generate a planned path based on the location data of the inspection robot and the terrain map; Control module: used to control the inspection robot to cross the tower according to the planned path.

[0041] Specifically, the inspection robot used in this embodiment includes a camera, a detection module, a communication module, a positioning module and a UAV flight module, etc. The camera is used to obtain image data of the overhead route during the inspection process, and to identify the poles / obstacles based on the image data; the positioning module is used to obtain the location information of the inspection robot to facilitate the subsequent generation of planned paths; the communication module is used for data transmission; the detection module includes several types of sensors for inspection through the detection module, and the types and quantities of sensors in the detection module are set according to the actual inspection situation; the UAV flight module is used to drive the inspection robot to fly, which is convenient for crossing towers and has a better shooting angle.

[0042] Specifically, the preset image processing model includes the following steps: S21, obtaining training data; Specifically, the training data is constructed by annotating photos collected by drones on the spot and open source datasets on the Internet; S22, perform data enhancement on the training data; Specifically, data enhancement in S22 is to expand the data set by adopting geometric transformation and image transformation to enhance the robustness of the model; data enhancement allows limited data to generate more data, increase the number and diversity of training samples (noise data), and improve the robustness of the model. It is generally used for training sets. Neural networks require a large number of parameters. Many neural networks have millions of parameters, and a large amount of data is required for training to make these parameters work correctly. However, in many actual projects, it is difficult for us to find sufficient data to complete the task. Randomly changing training samples can reduce the model's dependence on certain attributes, thereby improving the generalization ability of the model. The data enhancement methods used include: scaling, horizontal flipping, vertical flipping, rotating 90 degrees, translating, darkening, brightening, salt and pepper noise, Gaussian noise, and 12 types of filter transformations; S23, dividing the enhanced training data into a training set, a validation set, and a test set; S24, using the YOLOv5 network to train the image processing model according to the training set and the test set; Specifically, edge detection or contour extraction methods are used in the image processing model to extract features of towers / obstacles, and deep learning models (such as convolutional neural networks) that perform well in the field of image recognition are used to train special algorithms.

[0043] S25. Test the image processing model according to the test set. If the image processing model passes the test, output the image processing model. If the model fails the test, return to step S21.

[0044] Specifically, S25 includes the following steps: S251, input the test set into the image processing model to obtain several recognition results; S252, obtaining a pole tower database, judging a number of recognition results according to the pole tower database, and obtaining a recognition accuracy rate; S253, if the recognition accuracy is higher than the first threshold, the test passes; If the accuracy is less than or equal to the first threshold, the test fails and the process returns to step S21.

[0045] The experiment measures the actual effect of the algorithm test by accuracy. Accuracy is equal to the number of samples N correctly detected by the algorithm in the picture. True Divide by the total number of samples of each category in the picture, that is, the number of correct samples detected N True and the number of samples with detection errors N False , which is specifically expressed as follows:

[0046] Specifically, the pole tower database is a conventional database in the art, and the pole tower database includes the types and sizes of pole towers.

[0047] Specifically, the topographic map shows obstacles, tower types and tower sizes on overhead lines.

[0048] Specifically, in the step of obtaining the position data of the inspection robot and generating a planned path based on the position data of the inspection robot and the terrain map, the current position and the target position are determined based on the perception information and dynamic path planning is performed to generate an optimal route from the current position to safely cross the tower obstacles. The route should ensure that the robot maintains a safe distance from obstacles during the tower crossing process.

[0049] Specifically, the step of obtaining the position data of the inspection robot and generating a planned path according to the position data of the inspection robot and the terrain map includes the following steps: Obtain the location data of the inspection robot and mark the starting point on the topographic map based on the location data; The terrain map shows the location of obstacles and the type of towers; According to the tower type, query the preset tower type database to obtain the location point of the tower; Mark several target nodes on the topographic map according to the location of obstacles and towers; Generate several planned routes based on the starting point and target nodes Based on technologies such as GPS+RTK, the precise position of a robot can be obtained, and the pole towers in the field of view can be detected through image data, the corresponding categories can be identified, and their size and area in the image can be determined; based on image processing technology and the absolute position limitation of GPS, the robot's more accurate position and posture relative to the inspection tower can be determined; at the same time, the tower type information determined by the identification information and positioning information can be searched and consulted in the pre-constructed tower type database to obtain a series of key points required for the pole tower; with the current position as the starting point and a series of key points as the target points, dynamic planning of the route is carried out.

[0050] Specifically, the step of controlling the inspection robot to cross the tower according to the planned path includes the following steps: The inspection robot obtains the planned path and adjusts its flight attitude and speed according to the planned path, so that the robot can fly stably along the planned route and complete the tower crossing action. During the tower crossing process, the robot will automatically identify and avoid obstacles such as wires, poles, and towers to achieve safe and stable flight, and make intelligent decisions, such as automatic route calculation, target identification, and target tracking.

[0051] After the robot successfully crosses the tower obstacle, the robot's position and posture are confirmed through the onboard sensors to ensure the smooth completion of the tower crossing mission and prepare for the next stage of inspection mission.

[0052] Example 3 A computer device includes a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, a cross-tower control method for an overhead line inspection robot of Example 1 is implemented.

[0053] Example 4 A computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, a cross-tower control method of an overhead line inspection robot of embodiment 1 is implemented.

[0054] It is known from common technical knowledge that the present invention can be implemented by other embodiments that do not deviate from its spirit or essential features. Therefore, the above disclosed embodiments are only illustrative in all respects and are not exclusive. All changes within the scope of the present invention or within the scope equivalent to the present invention are included in the present invention.

[0055] It will be appreciated by those skilled in the art that embodiments of the present invention may be provided as methods, systems, or computer program products. Therefore, the present invention may take the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware. Furthermore, the present invention may take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0056] The present invention is described with reference to flowcharts and / or block diagrams of methods, devices (systems), and computer program products according to embodiments of the present invention. It should be understood that each process and / or block in the flowchart and / or block diagram, as well as the combination of processes and / or blocks in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to produce a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1 A process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0057] These computer program instructions may also be stored in a computer readable memory capable of directing a computer or other programmable data processing device to operate in a specific manner, so that the instructions stored in the computer readable memory produce an article of manufacture including an instruction device, which implements the process Figure 1 A process or multiple processes and / or boxes Figure 1 A function specified in one or more boxes.

[0058] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operating steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing instructions for implementing the process in the computer or other programmable device. Figure 1 A process or multiple processes and / or boxes Figure 1 The steps for the functions specified in one or more boxes.

[0059] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention rather than to limit it. Although the present invention has been described in detail with reference to the above embodiments, ordinary technicians in the relevant field should understand that the specific implementation methods of the present invention can still be modified or replaced by equivalents. Any modification or equivalent replacement that does not depart from the spirit and scope of the present invention should be covered within the scope of protection of the claims of the present invention.

Claims

1. A method for controlling an overhead line inspection robot across towers, characterized in that: The following steps are involved: Acquire image data of overhead lines in a preset area; Inputting the image data into a preset image processing model to obtain a topographic map within a preset area; Obtain the location data of the inspection robot, and generate a planned path based on the location data of the inspection robot and the terrain map; The inspection robot is controlled to cross the tower according to the planned path.

2. The method for controlling an overhead line inspection robot across towers according to claim 1, characterized in that: The preset image processing model includes the following steps: Get training data; Perform data augmentation on training data; The enhanced training data is divided into training set, validation set and test set; The image processing model is trained using the YOLOv5 network based on the training set and the test set; The image processing model is tested according to the test set. If it passes the test, the image processing model is output; if it fails the test, the training data is re-acquired.

3. The method for controlling an overhead line inspection robot across towers according to claim 2, characterized in that: The step of testing the image processing model according to the test set, outputting the image processing model if the test is passed, and reacquiring training data if the test is not passed specifically includes: Input the test set into the image processing model and obtain several recognition results; Obtain a pole tower database, judge a number of recognition results according to the pole tower database, and obtain a recognition accuracy rate; If the recognition accuracy is higher than the first threshold, the test passes; If the accuracy is less than or equal to the first threshold, the test fails.

4. The method for controlling an overhead line inspection robot across towers according to claim 2, characterized in that: The data enhancement includes scaling, horizontal flipping, vertical flipping, rotating 90 degrees, translating, darkening, brightening, salt and pepper noise, Gaussian noise, and filter transformation.

5. The method for controlling an overhead line inspection robot across towers according to claim 1, characterized in that: The step of controlling the inspection robot to cross the tower according to the planned path includes the following steps: Obtain the planned path, and adjust the flight posture and speed of the inspection robot according to the planned path to complete the cross-tower action.

6. The method for controlling an overhead line inspection robot across towers according to claim 1, characterized in that: The step of obtaining the position data of the inspection robot and generating a planned path according to the position data of the inspection robot and the terrain map includes the following steps: Obtain the location data of the inspection robot and mark the starting point on the topographic map based on the location data; The terrain map shows the location of obstacles and the type of towers; According to the tower type, query the preset tower type database to obtain the location point of the tower; Mark several target nodes on the topographic map according to the location of obstacles and towers; Generate several planned routes based on the starting point and target nodes.

7. The method for controlling an overhead line inspection robot across towers according to claim 6, characterized in that: The tower type database is generated according to the tower size data.

8. An overhead line inspection robot cross-tower control device, characterized in that: include: Data acquisition module: used to acquire image data of overhead lines in a preset area; Image recognition module: used to input image data into a preset image processing model to obtain a topographic map within a preset area; Path generation module: used to obtain the location data of the inspection robot and generate a planned path based on the location data of the inspection robot and the terrain map; Control module: used to control the inspection robot to cross the tower according to the planned path.

9. A computer device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that: When the processor executes the computer program, the cross-tower control method of an overhead line inspection robot described in any one of claims 1 to 7 is implemented.

10. A computer-readable storage medium storing a computer program, characterized in that: When the computer program is executed by a processor, a cross-tower control method for an overhead line inspection robot according to any one of claims 1 to 7 is implemented.

Images