Gulongai unmanned inspection system and method based on multi-sensor fusion

Abstract

The application discloses a Gulei cover unmanned inspection system and method based on multi-sensor fusion, and relates to the technical field of intelligent operation and maintenance of urban infrastructure.The system comprises the following steps: inspection path planning: according to Gulei cover distribution data and historical abnormal record of an area to be inspected, an optimal inspection route is generated through an inspection path planning module; multi-modal data synchronous acquisition: an unmanned inspection carrier travels along the planned path, and when approaching each Gulei cover target, a multi-modal sensing module synchronously starts data acquisition to obtain visible light image data and three-dimensional laser point cloud data; step S3, edge real-time processing: the edge computing processing module performs real-time processing on the collected multi-modal original data; intelligent early warning: according to an abnormal judgment result and a severity classification, early warning information is automatically generated. The application realizes automatic and unmanned inspection and real-time early warning of various abnormal states of urban road Gulei covers, such as displacement, inclination, loss, damage and abnormal opening.

Description

Technical Field

[0001] This invention relates to the field of intelligent operation and maintenance technology for urban infrastructure, and in particular to the Guligai unmanned inspection system and method based on multi-sensor fusion. Background Technology

[0002] Manhole covers are an important component of urban underground pipeline systems, widely distributed along city roads, sidewalks, and public areas. They play a vital role in protecting underground pipeline facilities and ensuring the safe passage of pedestrians and vehicles. Statistics show that there are hundreds of millions of manhole covers of various types in Chinese cities, and their safety status directly affects public safety and urban operational efficiency.

[0003] However, due to the large number and wide distribution of manhole covers, coupled with the long-term effects of vehicle traffic, natural erosion, and human damage, it is common for manhole covers to exhibit safety hazards such as displacement, tilting, damage, missing parts, and abnormal opening. Abnormal manhole covers can not only cause pedestrians to fall into them and suffer personal injury, but also lead to serious consequences such as vehicle damage and traffic accidents. In recent years, safety accidents caused by abnormal manhole covers have been frequently reported, becoming a major hidden danger to urban public safety.

[0004] Currently, the inspection of manhole covers mainly relies on manual patrols. Inspectors must visually check the condition of each manhole cover along the street. This method has the following significant drawbacks: First, it is inefficient; manual patrols have limited coverage, long patrol cycles, and make it difficult to achieve high-frequency, full-coverage inspections. Second, it is labor-intensive; inspectors work outdoors for extended periods, significantly constrained by weather conditions and nighttime lighting. Third, it is highly subjective; inspection results depend on personal experience and judgment, easily leading to missed inspections and misjudgments. Fourth, it lacks real-time capability, failing to enable immediate detection and rapid response to anomalies. Fifth, it has a low level of data digitization, making it difficult to establish a systematic inspection record and data analysis system.

[0005] While some existing technologies offer solutions for monitoring manhole covers based on fixed sensors, such as installing tilt sensors, displacement sensors, or RFID tags, these solutions require individual modification and installation for each manhole cover, resulting in extremely high initial deployment costs. Furthermore, sensor maintenance and power supply remain persistent challenges. Additionally, these solutions can only monitor manhole covers with installed sensors, failing to cover those without. Furthermore, some solutions attempt to utilize vehicle-mounted cameras for manhole cover status detection, but most remain at the level of single-image recognition, offering limited accuracy in complex environments and lacking systematic inspection planning and multi-dimensional anomaly detection capabilities.

[0006] Therefore, there is an urgent need for a cost-effective, wide-coverage, high-accuracy, and real-time unmanned inspection system and method for abnormal conditions in Guligai, in order to solve the above-mentioned problems in the existing technology. Summary of the Invention

[0007] This invention provides an unmanned inspection system and method for manhole covers based on multi-sensor fusion, which can be used to realize automated and unmanned inspection and real-time early warning of various abnormal states of manhole covers on urban roads, such as displacement, tilt, missing, damaged, and abnormal opening.

[0008] According to one aspect of this disclosure, a multi-sensor fusion-based unmanned inspection system for Guligai is provided, the system comprising: a multimodal perception module, an edge computing processing module, and an inspection path planning module; The inspection route planning module is configured to automatically generate the optimal inspection route based on the distribution map of Guligai and historical anomaly data using an intelligent optimization algorithm, and supports dynamic adjustment of the route based on real-time feedback; the intelligent optimization algorithm comprehensively considers multiple objective optimization factors: inspection coverage and path length; The multimodal perception module is installed on the unmanned inspection vehicle and includes a high-resolution visual acquisition unit and a three-dimensional laser scanning unit; A high-resolution visual acquisition unit is configured to acquire visible light images and video data of Guligai and its surrounding area; A three-dimensional laser scanning unit is configured to acquire three-dimensional point cloud data of the manhole cover and its surrounding road surface, for accurately detecting the height difference, tilt angle and edge gap of the manhole cover. The three-dimensional laser scanning unit is a solid-state lidar or a mechanical lidar. The edge computing processing module is connected to the multimodal sensing module and is configured to perform real-time processing and analysis of the collected multimodal data, including: The image preprocessing submodule is used to perform distortion correction and enhancement on visible light images.

[0009] In one possible implementation, the system further includes an unmanned inspection vehicle basic module, which includes a chassis drive unit, an autonomous navigation unit, an obstacle avoidance unit, and a power management unit.

[0010] In one possible implementation, the edge computing module includes: a point cloud preprocessing submodule for denoising, registering, and segmenting 3D point cloud data; a Yolov11 target detection model configured to detect the Guligai target region; a multimodal feature fusion module configured to fuse visible light image features and 3D geometric deformation features; and a Guligai anomaly level determination module configured to comprehensively determine the anomaly type and severity of Guligai based on the multimodal fusion features.

[0011] In one possible implementation, the manhole cover anomaly level determination module extracts the following appearance features from the visible light image: cracks, gaps, damaged areas, missing areas and open areas, and extracts the height difference, tilt angle, edge gap and horizontal displacement of the manhole cover relative to the surrounding road surface from the three-dimensional point cloud; the height difference, tilt angle, edge gap and horizontal displacement are geometric features. The extracted appearance and geometric features are processed through a multimodal feature fusion network to form a comprehensive feature vector, which is then input into the anomaly level determination module of the gable cover to obtain the category confidence scores corresponding to the normal, displaced, tilted, damaged, missing, and abnormally open states. Based on the comparison results of each measurement parameter with the preset judgment threshold, a comprehensive risk score is calculated according to the anomaly type, the parameter exceeding the limit, the scope of the anomaly's impact, and the category confidence level, thereby determining the anomaly type and severity. When multiple abnormal conditions are met simultaneously, multiple abnormal labels are output, and the highest risk level among them is used as the final warning level.

[0012] The unmanned inspection method for Guligai based on multi-sensor fusion includes the following steps: Step S1, Inspection Path Planning: Based on the distribution data of the gorges in the area to be inspected and historical anomaly records, the optimal inspection route is generated through the inspection path planning module, and inspection parameters are configured, including inspection speed, data collection interval and sensing mode. Step S2, Multimodal Data Synchronous Acquisition: The unmanned inspection vehicle travels along the planned path. When it approaches each Guligai target, the multimodal perception module synchronously starts data acquisition to obtain visible light image data and three-dimensional laser point cloud data. Step S3, Real-time processing at the edge: The edge computing processing module processes the collected multimodal raw data in real time; Step S4, Intelligent Early Warning: Based on the anomaly judgment result and severity classification, an early warning message is automatically generated, and the early warning notification and maintenance work order are pushed to the management platform and relevant responsible persons through the early warning and work order distribution module.

[0013] In one possible implementation, S3 includes: S3.1, timestamp matching is performed on the synchronously acquired visible light images and 3D point clouds, and coordinate system transformation and spatial alignment are completed according to the sensor calibration parameters; S3.2, the image preprocessing module performs distortion correction, noise reduction, brightness normalization and contrast enhancement on the visible light image; S3.3, the point cloud preprocessing module performs outlier removal, downsampling, ground point segmentation, and multi-frame point cloud registration on the 3D point cloud; S3.4, the YOLOv11 target detection model is used to identify the manhole cover target, obtain the target category, target location and detection confidence, and extract the corresponding image and point cloud region of interest; the following appearance features are extracted from the image region of interest: cracks, gaps, rust, damaged area, missing area and abnormal opening area; the height difference, tilt angle, edge gap, horizontal displacement and surface flatness of the manhole cover relative to the surrounding road surface are extracted from the point cloud region of interest; the height difference, tilt angle, edge gap and horizontal displacement are geometric features; S3.5, the extracted appearance features and geometric features are input into the multimodal feature fusion network, the attention mechanism is used to calculate the weights of image features and point cloud features respectively, and the weighted features are concatenated or mapped to generate multimodal fusion features.

[0014] In one possible implementation, S4 includes: S4.1 Input the multimodal fusion features into the anomaly discrimination model and output the category confidence scores for normal, shifted, tilted, damaged, missing, and abnormally open states; S4.2, comprehensively compares the geometric parameters, appearance parameters, and category confidence with the preset judgment conditions, determines the severity based on the parameter exceedance range and the scope of abnormal influence, and generates a judgment result that includes the Guli Gai identifier, geographical location, abnormality type, severity, detection parameters, confidence level, and collection time.

[0015] In one possible implementation, the anomaly determination result includes anomaly type, severity, anomaly confidence, anomaly location, and anomaly parameters; Among them, the abnormality types include at least normal, displacement, tilt, damage, missing and abnormal opening; the severity can be divided into four levels: normal, general abnormality, relatively severe abnormality and severe abnormality. Abnormal parameters include the measured values ​​of height difference, tilt angle, edge gap, horizontal displacement, and damaged area. The system determines the severity level based on the type of abnormality and the degree to which the abnormal parameters exceed the preset threshold, and generates corresponding prompt, general, severe, or emergency warning information.

[0016] Compared with the prior art, the beneficial effects of the present invention are: (1) By adopting an unmanned inspection carrier equipped with a multimodal sensor array, the fully automated and unmanned inspection of the abnormal state of the manhole cover was realized, which greatly improved the inspection efficiency and coverage. A single inspection can cover thousands of manhole covers on tens of kilometers of roads, thus improving the inspection efficiency.

[0017] (2) By integrating sensing methods such as visible light and three-dimensional laser scanning, multi-dimensional and all-round detection of the abnormal state of Guligai was realized, effectively overcoming the limitations of single detection methods.

[0018] (3) No individual modification is required for each manhole cover, the deployment cost is controllable, the inspection range can be flexibly expanded, and it is suitable for various urban road scenarios. Attached Figure Description

[0019] Figure 1 A block diagram of a multi-sensor fusion-based unmanned inspection system for Guligai, according to an embodiment of this disclosure, is shown.

[0020] Figure 2 A schematic block diagram of a multimodal feature fusion network according to an embodiment of the present disclosure is shown. Detailed Implementation

[0021] Various exemplary embodiments, features, and aspects of this disclosure will now be described in detail with reference to the accompanying drawings. The same reference numerals in the drawings denote elements that have the same or similar functions. Although various aspects of the embodiments are shown in the drawings, they are not necessarily drawn to scale unless specifically indicated otherwise.

[0022] The term “exemplary” as used herein means “serving as an example, embodiment, or illustration.” Any embodiment illustrated herein as “exemplary” is not necessarily to be construed as superior to or better than other embodiments.

[0023] Furthermore, to better illustrate this disclosure, numerous specific details are set forth in the following detailed description. Those skilled in the art will understand that this disclosure can be practiced without certain specific details. In some instances, methods, means, components, and circuits well known to those skilled in the art have not been described in detail in order to highlight the main points of this disclosure.

[0024] The Guligai unmanned inspection system based on multi-sensor fusion includes: a multimodal perception module, an edge computing processing module, and an inspection path planning module. The inspection route planning module is configured to automatically generate the optimal inspection route based on the distribution map of Guligai and historical anomaly data using an intelligent optimization algorithm, and supports dynamic adjustment of the route based on real-time feedback; the intelligent optimization algorithm comprehensively considers multiple objective optimization factors such as inspection coverage and route length. The multimodal perception module is installed on the unmanned inspection vehicle and includes a high-resolution visual acquisition unit and a three-dimensional laser scanning unit; A high-resolution visual acquisition unit is configured to acquire visible light images and video data of Guligai and its surrounding area; A three-dimensional laser scanning unit is configured to acquire three-dimensional point cloud data of the manhole cover and its surrounding road surface, for accurately detecting the height difference, tilt angle and edge gap of the manhole cover. The three-dimensional laser scanning unit is a solid-state lidar or a mechanical lidar. The edge computing processing module is connected to the multimodal sensing module and is configured to perform real-time processing and analysis of the collected multimodal data, including: The image preprocessing submodule is used to perform distortion correction and enhancement on visible light images.

[0025] In one possible implementation, the system further includes an unmanned inspection vehicle basic module, which includes a chassis drive unit, an autonomous navigation unit, an obstacle avoidance unit, and a power management unit.

[0026] In one possible implementation, the edge computing module includes: a point cloud preprocessing submodule for denoising, registering, and segmenting 3D point cloud data; a Yolov11 target detection model configured to detect the Guligai target region; a multimodal feature fusion module configured to fuse visible light image features and 3D geometric deformation features; and a Guligai anomaly level determination module configured to comprehensively determine the anomaly type and severity of Guligai based on the multimodal fusion features.

[0027] In one possible implementation, the manhole cover anomaly level determination module extracts the following appearance features from the visible light image: cracks, gaps, damaged areas, missing areas and open areas, and extracts the height difference, tilt angle, edge gap and horizontal displacement of the manhole cover relative to the surrounding road surface from the three-dimensional point cloud; the height difference, tilt angle, edge gap and horizontal displacement are geometric features. The extracted appearance and geometric features are processed through a multimodal feature fusion network to form a comprehensive feature vector, which is then input into the anomaly level determination module of the gable cover to obtain the category confidence scores corresponding to the normal, displaced, tilted, damaged, missing, and abnormally open states. Based on the comparison results of each measurement parameter with the preset judgment threshold, a comprehensive risk score is calculated according to the anomaly type, the parameter exceeding the limit, the scope of the anomaly's impact, and the category confidence level, thereby determining the anomaly type and severity. When multiple abnormal conditions are met simultaneously, multiple abnormal labels are output, and the highest risk level among them is used as the final warning level.

[0028] The unmanned inspection method for Guligai based on multi-sensor fusion includes the following steps: Step S1, Inspection Path Planning: Based on the distribution data of the gorges in the area to be inspected and historical anomaly records, the optimal inspection route is generated through the inspection path planning module, and inspection parameters are configured, including inspection speed, data collection interval and sensing mode. Step S2, Multimodal Data Synchronous Acquisition: The unmanned inspection vehicle travels along the planned path. When it approaches each Guligai target, the multimodal perception module synchronously starts data acquisition to obtain visible light image data and three-dimensional laser point cloud data. Step S3, Real-time processing at the edge: The edge computing processing module processes the collected multimodal raw data in real time; Step S4, Intelligent Early Warning: Based on the anomaly judgment result and severity classification, an early warning message is automatically generated, and the early warning notification and maintenance work order are pushed to the management platform and relevant responsible persons through the early warning and work order distribution module.

[0029] In one possible implementation, S3 includes: S3.1, timestamp matching is performed on the synchronously acquired visible light images and 3D point clouds, and coordinate system transformation and spatial alignment are completed according to the sensor calibration parameters; S3.2, the image preprocessing module performs distortion correction, noise reduction, brightness normalization and contrast enhancement on the visible light image; S3.3, the point cloud preprocessing module performs outlier removal, downsampling, ground point segmentation, and multi-frame point cloud registration on the 3D point cloud; S3.4, the YOLOv11 target detection model is used to identify the manhole cover target, obtain the target category, target location and detection confidence, and extract the corresponding image and point cloud region of interest; the following appearance features are extracted from the image region of interest: cracks, gaps, rust, damaged area, missing area and abnormal opening area; geometric features such as the height difference, tilt angle, edge gap, horizontal displacement and surface flatness of the manhole cover relative to the surrounding road surface are extracted from the point cloud region of interest; the height difference, tilt angle, edge gap and horizontal displacement are geometric features; S3.5, the extracted appearance features and geometric features are input into the multimodal feature fusion network, the attention mechanism is used to calculate the weights of image features and point cloud features respectively, and the weighted features are concatenated or mapped to generate multimodal fusion features.

[0030] In one possible implementation, S4 includes: S4.1 Input the multimodal fusion features into the anomaly discrimination model and output the category confidence scores of states such as normal, shifted, tilted, damaged, missing, and abnormally open. S4.2, comprehensively compares the geometric parameters, appearance parameters, and category confidence with the preset judgment conditions, determines the severity based on the parameter exceedance range and the scope of abnormal influence, and generates a judgment result that includes the Guli Gai identifier, geographical location, abnormality type, severity, detection parameters, confidence level, and collection time.

[0031] In one possible implementation, the anomaly determination result includes anomaly type, severity, anomaly confidence, anomaly location, and anomaly parameters; Among them, the abnormality types include at least normal, displacement, tilt, damage, missing and abnormal opening; the severity can be divided into four levels: normal, general abnormality, relatively severe abnormality and severe abnormality. Abnormal parameters include the measured values ​​of height difference, tilt angle, edge gap, horizontal displacement, and damaged area. The system determines the severity level based on the type of abnormality and the degree to which the abnormal parameters exceed the preset threshold, and generates corresponding prompt, general, severe, or emergency warning information.

[0032] Compared with the prior art, the beneficial effects of the present invention are: (1) By adopting an unmanned inspection carrier equipped with a multimodal sensor array, the fully automated and unmanned inspection of the abnormal state of the manhole cover was realized, which greatly improved the inspection efficiency and coverage. A single inspection can cover thousands of manhole covers on tens of kilometers of roads, thus improving the inspection efficiency.

[0033] (2) By integrating sensing methods such as visible light and three-dimensional laser scanning, multi-dimensional and all-round detection of the abnormal state of Guligai was realized, effectively overcoming the limitations of single detection methods.

[0034] (3) No individual modification is required for each manhole cover, the deployment cost is controllable, the inspection range can be flexibly expanded, and it is suitable for various urban road scenarios.

[0035] The various embodiments of this disclosure have been described above. These descriptions are exemplary and not exhaustive, nor are they limited to the disclosed embodiments. Many modifications and variations will be apparent to those skilled in the art without departing from the scope and spirit of the described embodiments. The terminology used herein is chosen to best explain the principles, practical application, or technical improvements to the embodiments in the market, or to enable others skilled in the art to understand the embodiments disclosed herein.

Claims

1. A multi-sensor fusion-based unmanned inspection system for Guli Gai, characterized in that, The system includes: a multimodal perception module, an edge computing processing module, and an inspection path planning module; The inspection route planning module is configured to automatically generate the optimal inspection route based on the distribution map of Guligai and historical anomaly data using an intelligent optimization algorithm, and supports dynamic adjustment of the route based on real-time feedback; the intelligent optimization algorithm comprehensively considers multiple objective optimization factors: inspection coverage and path length. The multimodal perception module is installed on the unmanned inspection vehicle and includes a high-resolution visual acquisition unit and a three-dimensional laser scanning unit; A high-resolution visual acquisition unit is configured to acquire visible light images and video data of Guligai and its surrounding area; A three-dimensional laser scanning unit is configured to acquire three-dimensional point cloud data of the manhole cover and its surrounding road surface, for accurately detecting the height difference, tilt angle and edge gap of the manhole cover. The three-dimensional laser scanning unit is a solid-state lidar or a mechanical lidar. The edge computing processing module is connected to the multimodal sensing module and is configured to perform real-time processing and analysis of the collected multimodal data, including: The image preprocessing submodule is used to perform distortion correction and enhancement on visible light images.

2. The unmanned inspection system for Guli Gai based on multi-sensor fusion as described in claim 1, characterized in that, The system also includes an unmanned inspection vehicle basic module, which includes a chassis drive unit, an autonomous navigation unit, an obstacle avoidance unit, and a power management unit.

3. The unmanned inspection system for Guli Gai based on multi-sensor fusion as described in claim 1, characterized in that, In the edge computing module: the point cloud preprocessing submodule is used to perform noise reduction, registration and segmentation on 3D point cloud data; the Yolov11 target detection model is configured to detect the Guligai target region; The multimodal feature fusion module is configured to fuse visible light image features and three-dimensional geometric deformation features; The Guligai anomaly level determination module is configured to comprehensively determine the anomaly type and severity of Guligai based on multimodal fusion features.

4. The unmanned inspection system for Guli Gai based on multi-sensor fusion as described in claim 1, characterized in that, The manhole cover anomaly level determination module extracts the following appearance features from visible light images: cracks, gaps, damaged areas, missing areas, and open areas. It also extracts the height difference, tilt angle, edge gap, and horizontal displacement of the manhole cover relative to the surrounding road surface from the 3D point cloud. The height difference, tilt angle, edge gap, and horizontal displacement are geometric features. The extracted appearance and geometric features are processed through a multimodal feature fusion network to form a comprehensive feature vector, which is then input into the anomaly level determination module of the gable cover to obtain the category confidence scores corresponding to the normal, displaced, tilted, damaged, missing, and abnormally open states. Based on the comparison results of each measurement parameter with the preset judgment threshold, a comprehensive risk score is calculated according to the anomaly type, the parameter exceeding the limit, the scope of the anomaly's impact, and the category confidence level, thereby determining the anomaly type and severity. When multiple abnormal conditions are met simultaneously, multiple abnormal labels are output, and the highest risk level among them is used as the final warning level.

5. A method for unmanned inspection of manhole covers based on multi-sensor fusion, characterized in that: Includes the following steps: Step S1, Inspection Path Planning: Based on the distribution data of the gorges in the area to be inspected and historical anomaly records, the optimal inspection route is generated through the inspection path planning module, and inspection parameters are configured, including inspection speed, data collection interval and sensing mode. Step S2, Multimodal Data Synchronous Acquisition: The unmanned inspection vehicle travels along the planned path. When it approaches each Guligai target, the multimodal perception module synchronously starts data acquisition to obtain visible light image data and three-dimensional laser point cloud data. Step S3, Real-time processing at the edge: The edge computing processing module processes the collected multimodal raw data in real time; Step S4, Intelligent Early Warning: Based on the anomaly judgment result and severity classification, an early warning message is automatically generated, and the early warning notification and maintenance work order are pushed to the management platform and relevant responsible persons through the early warning and work order distribution module.

6. The unmanned inspection method for Guli Gai based on multi-sensor fusion according to claim 5, characterized in that, S3 include: S3.1, timestamp matching is performed on the synchronously acquired visible light images and 3D point clouds, and coordinate system transformation and spatial alignment are completed according to the sensor calibration parameters; S3.2, the image preprocessing module performs distortion correction, noise reduction, brightness normalization and contrast enhancement on the visible light image; S3.3, the point cloud preprocessing module performs outlier removal, downsampling, ground point segmentation, and multi-frame point cloud registration on the 3D point cloud; S3.4, the YOLOv11 target detection model is used to identify the manhole cover target, obtain the target category, target location and detection confidence, and extract the corresponding image and point cloud region of interest; the following appearance features are extracted from the image region of interest: cracks, gaps, rust, damaged area, missing area and abnormal opening area; the height difference, tilt angle, edge gap, horizontal displacement and surface flatness of the manhole cover relative to the surrounding road surface are extracted from the point cloud region of interest; the height difference, tilt angle, edge gap and horizontal displacement are geometric features; S3.5, the extracted appearance features and geometric features are input into the multimodal feature fusion network, the attention mechanism is used to calculate the weights of image features and point cloud features respectively, and the weighted features are concatenated or mapped to generate multimodal fusion features.

7. The unmanned inspection method for Guli Gai based on multi-sensor fusion according to claim 5, characterized in that, S4 include: S4.1 Input the multimodal fusion features into the anomaly discrimination model and output the category confidence scores for normal, shifted, tilted, damaged, missing, and abnormally open states; S4.2, comprehensively compares the geometric parameters, appearance parameters, and category confidence with the preset judgment conditions, determines the severity based on the parameter exceedance range and the scope of abnormal influence, and generates a judgment result that includes the Guli Gai identifier, geographical location, abnormality type, severity, detection parameters, confidence level, and collection time.

8. The unmanned inspection method for Guli Gai based on multi-sensor fusion according to claim 5, characterized in that, The anomaly determination results include anomaly type, severity, anomaly confidence, anomaly location, and anomaly parameters; Among them, the abnormality types include at least normal, displacement, tilt, damage, missing and abnormal opening; the severity can be divided into four levels: normal, general abnormality, relatively severe abnormality and severe abnormality. Abnormal parameters include the measured values ​​of height difference, tilt angle, edge gap, horizontal displacement, and damaged area. The system determines the severity level based on the type of abnormality and the degree to which the abnormal parameters exceed the preset threshold, and generates corresponding prompt, general, severe, or emergency warning information.

Images