Image acquisition control method and system based on inspection robot

Abstract

The invention discloses an image acquisition control method and system based on an inspection robot, and relates to the technical field of inspection image acquisition control, and the method comprises the following steps: obtaining an image threshold corresponding to an inspection target and a basic image acquisition parameter corresponding to a target inspection time sequence based on historical inspection data; acquiring a real-time inspection image based on the basic image acquisition parameter, and updating the image acquisition parameter based on the basic image acquisition parameter according to the image threshold and the real-time inspection image; acquiring an inspection speed adjustment coefficient of the inspection robot based on the target inspection time sequence and the inspection target distance; and obtaining update fluctuation according to the image acquisition parameter update value of the previous time period, and updating the basic image acquisition parameter of the next inspection target according to the update fluctuation and the inspection speed adjustment coefficient. The method has the beneficial effects that the accuracy of image acquisition parameters is ensured under the condition that the adjustment amplitude is reduced, and the accuracy of image acquisition and the inspection efficiency are realized.

Description

Technical Field

[0001] This application relates to the technical field of inspection image acquisition control, and particularly to a method and system for controlling image acquisition based on an inspection robot. Background Art

[0002] Traditional inspection robot image acquisition control methods often rely on preset fixed parameters for image acquisition. This method can meet the basic requirements when the inspection target and environmental conditions are relatively stable. However, once the inspection target or environmental conditions change, such as light intensity, weather conditions, target distance, etc., the fixed image acquisition parameters are difficult to ensure the quality and clarity of the images, which may lead to incomplete, blurred or distorted image information, seriously affecting subsequent analysis and judgment.

[0003] In related technologies, fixed light source control is adopted, and the power of the light source is generally very large, resulting in a large power consumption of the inspection robot system. Fixed light source control has a better imaging effect when the line is darker, but has a poor imaging effect when the light source is sufficient or there is water vapor reflection. For the coding of the visual sensor, a fixed line frequency is used. When running at a low speed and only a small line frequency is required, the image is prone to stretching, and when running at a high speed, data is easily lost, making the acquired image data incomplete or inaccurate.

[0004] The patent "A Method for Event Monitoring in the Environment of a Tunnel Rail Inspection Robot", publication number: CN119152452A, publication date: December 17, 2024, specifically discloses that the method includes: S1. Install a monocular camera on the track to capture and preprocess the road surface image in real time; S2. Extract image features through the CSPDarknet network, and use the detection head and segmentation head to obtain information such as target bounding boxes and segmentation mask coordinates; S3. Perform nearest neighbor matching on the segmentation mask points of the front and rear frames and combine IMU data to predict the camera pose change; S4. Estimate the depth map based on the encoder-decoder model to generate multi-view depth information; S5. Detect tunnel events based on the camera and target object information and trigger an alarm. This solution improves the inspection accuracy by improving the motion pose and depth estimation of the bullet screen camera, but there is still a problem that fixed image acquisition parameters are difficult to ensure the quality and clarity of the images. Summary of the Invention

[0005] In view of the problem in the prior art that fixed image acquisition parameters are difficult to ensure the quality and clarity of images, the present application provides an image acquisition control method and system based on an inspection robot. By obtaining the image threshold corresponding to the inspection target and the basic image acquisition parameters corresponding to the target inspection time sequence from historical inspection data, the initial image acquisition parameters approximately meet the current environmental requirements. At the same time, based on the difference between the real-time inspection image and the image threshold, the image acquisition parameters are dynamically fine-tuned, ensuring the accuracy of the image acquisition parameters while reducing the adjustment range, thereby achieving the accuracy of image acquisition and the inspection efficiency.

[0006] To achieve the above technical objectives, a technical solution provided by the present application is an image acquisition control method based on an inspection robot, including the following steps: S1: Obtain the inspection target and the target inspection time sequence of the inspection robot based on the target inspection route; S2: Obtain the image threshold corresponding to the inspection target and the basic image acquisition parameters corresponding to the target inspection time sequence based on historical inspection data; S3: Obtain the real-time inspection image based on the basic image acquisition parameters, update the image acquisition parameters according to the image threshold and the real-time inspection image based on the basic image acquisition parameters, and re-perform image acquisition with the updated value of the image acquisition parameters; S4: Obtain the inspection robot inspection speed adjustment coefficient based on the target inspection time sequence and the distance of the inspection target; S5: Obtain the update fluctuation according to the updated value of the image acquisition parameters in the previous period, and update the basic image acquisition parameters of the next inspection target with the update fluctuation and the inspection speed adjustment coefficient.

[0007] Further, S1 includes: constructing an inspection topology graph with the historical inspection targets in the historical inspection data; obtaining the inspection target by matching the target inspection route with the inspection topology graph; obtaining the target inspection time sequence of each inspection target based on the target inspection route and the target inspection time.

[0008] Further, S2 includes: obtaining the image threshold corresponding to the inspection target and the basic image acquisition parameters corresponding to the target inspection time sequence based on historical inspection data includes: obtaining the historical image data corresponding to each inspection target, calculating the complexity dimension according to the historical image data of each inspection target, and obtaining the image threshold according to the complexity dimension; obtaining the historical image acquisition parameters corresponding to each target inspection time sequence, and obtaining the basic image acquisition parameters according to the similarity of historical inspection environment parameters and historical image acquisition parameters.

[0009] Further, S3 includes: obtaining the real-time inspection image based on the basic image acquisition parameters; if the real-time inspection image meets the image threshold, save the real-time inspection image as the target inspection image; if the real-time inspection image does not meet the image threshold, update the image acquisition parameters according to the difference between the real-time inspection image and the image threshold; re-perform image acquisition with the updated value of the image acquisition parameters.

[0010] Further, S4 includes: calculating the target inspection time sequence corresponding to the current time sequence and the next inspection target, and calculating the inspection interval time; calculating the inspection robot inspection speed adjustment coefficient based on the inspection interval time and the inspection target distance.

[0011] Further, S5 includes: obtaining the updated value of the image acquisition parameter in the previous period, and using the unified fluctuation value of the updated value of the image acquisition parameter in the previous period as the update fluctuation; updating the basic image acquisition parameter of the next inspection target according to the update fluctuation, the inspection speed adjustment coefficient, and the speed line frequency influence relationship.

[0012] Further, calculating the complexity dimension according to the historical image data of each inspection target, and obtaining the image threshold according to the complexity dimension includes: obtaining the historical image data of each inspection target, and calculating the detail richness, texture complexity, and corresponding weight coefficients according to the similar objects in the historical image data; obtaining the minimum complexity dimension according to the valid images in the historical image data, and using the minimum complexity dimension as the image threshold.

[0013] Further, calculating the detail richness, texture complexity, and corresponding weight coefficients according to the similar objects in the historical image data includes: dividing the historical image data into valid images and invalid images; obtaining the important inspection objects according to the similar objects in the valid images and the similar objects in the invalid images; calculating the detail richness, texture complexity, and corresponding weight coefficients of the important inspection objects based on the valid images and the invalid images and the important inspection objects.

[0014] Further, the complexity dimension includes detail richness, texture complexity, importance dimension value, and combined dimension value.

[0015] Another technical solution provided by this application is a patrol robot image acquisition control system for implementing the method as described above, including: a vision sensor for performing image acquisition based on the image acquisition parameters; a main control unit for obtaining the basic image acquisition parameters according to the historical patrol data and transmitting them to the vision sensor; a storage unit for storing the patrol images collected by the vision sensor; and a processing unit for analyzing the real-time patrol images, obtaining the updated value of the image acquisition parameters, and transmitting them to the vision sensor.

[0016] Advantages of this application: By matching the target inspection time sequence of the inspection robot with the historical inspection data, the image acquisition parameters of the corresponding historical situation at this time sequence are obtained, which are used as the basic image acquisition parameters for the corresponding inspection target, and the image threshold corresponding to the inspection target is obtained. By comparing the real-time inspection image with the image threshold, it is judged whether the currently acquired real-time inspection image meets the inspection requirements. If not, the image acquisition parameters are adjusted according to the real-time acquisition difference to ensure that the acquired inspection image meets the requirements. At the same time, based on the target inspection time sequence and the distance between inspection targets, the distance of the inspection robot is adjusted to ensure that the inspection time requirement is met, and the image acquisition parameters for the subsequent time period are adjusted according to the adjusted speed ratio and the unified parameter fluctuation situation in the previous time period, so as to undertake part of the update operation during the movement process, reduce the adjustment calculation amount during the image acquisition of the inspection target, and ensure the image acquisition efficiency while ensuring the accuracy of image acquisition. Description of the Drawings

[0017] Figure 1 It is a schematic flowchart of a method for controlling image acquisition based on an inspection robot in this application. Specific Embodiments

[0018] To make the purpose, technical solutions and advantages of this application clearer, the following further elaborates this application in conjunction with the drawings and embodiments. It should be understood that the specific embodiments described here are only the best embodiments of this application, which are only used to explain this application and do not limit the protection scope of this application. All other embodiments obtained by those of ordinary skill in the art without creative efforts fall within the protection scope of this application.

[0019] As Figure 1 shown, as Embodiment 1 of this application, a method for controlling image acquisition based on an inspection robot includes the following steps: S1: Obtain the inspection target of the inspection robot and the target inspection time sequence based on the target inspection route; S2: Obtain the image threshold corresponding to the inspection target and the basic image acquisition parameters corresponding to the target inspection time sequence based on the historical inspection data; S3: Obtain the real-time inspection image based on the basic image acquisition parameters, update the image acquisition parameters according to the image threshold and the real-time inspection image based on the basic image acquisition parameters, and re-perform image acquisition with the updated value of the image acquisition parameters; S4: Obtain the inspection robot inspection speed adjustment coefficient based on the target inspection time sequence and the inspection target distance; S5: Obtain the update fluctuation according to the updated value of the image acquisition parameters in the previous time period, and update the basic image acquisition parameters of the next inspection target with the update fluctuation and the inspection speed adjustment coefficient.

[0020] In this embodiment, by matching the target inspection time sequence of the inspection robot with the historical inspection data, the image acquisition parameters of the corresponding historical situation at this time sequence are obtained, which are used as the basic image acquisition parameters for the corresponding inspection target, and the image threshold corresponding to the inspection target is obtained. By comparing the real-time inspection image with the image threshold, it is judged whether the currently acquired real-time inspection image meets the inspection requirements. If not, the image acquisition parameters are adjusted according to the real-time acquisition difference to ensure that the acquired inspection image meets the requirements. At the same time, the distance of the inspection robot is adjusted based on the target inspection time sequence and the distance between inspection targets to ensure that the inspection time requirement is met, and the image acquisition parameters for the subsequent time period are adjusted according to the adjusted speed ratio and the unified parameter fluctuation situation in the previous time period, so as to perform part of the update operation during the movement process, reduce the adjustment operation amount during the image acquisition of the inspection target, and ensure the image acquisition efficiency while ensuring the accuracy of image acquisition.

[0021] Among them, obtaining the inspection target and the target inspection time sequence of the inspection robot based on the target inspection route includes: Constructing an inspection topology map with the historical inspection targets in the historical inspection data; Obtaining the inspection target by matching the target inspection route with the inspection topology map; Obtaining the target inspection time sequence of each inspection target based on the target inspection route and the target inspection time.

[0022] Since there are necessary inspection targets in the overall inspection process, such as important operating equipment, fault-prone areas, etc., and the rest of the areas do not need to be focused on during the inspection process, an inspection topology map is constructed according to the historical inspection targets in the historical inspection data, which is convenient for the operator to set the target inspection route according to the inspection needs, or directly obtain the inspection target according to the topological points in the target inspection route after the operator sets the target inspection route, without additional selection, improving the operator's operation experience. At the same time, the speed of the inspection robot during the inspection process is reasonably allocated according to the historical inspection data and the target inspection time, which is reflected by the target inspection time sequence of each inspection target, that is, the time sequence for the inspection robot to reach each inspection target is initially set according to the historical inspection situation. Therefore, if there is a problem with the image acquisition of a certain inspection target during the inspection process and the time is delayed, the speed of the inspection robot can be directly adjusted according to the target inspection time sequence of the next inspection target to recover the delayed time and ensure that the actual inspection time meets the target inspection time, facilitating the operator's subsequent arrangements.

[0023] Obtaining the image threshold corresponding to the inspection target and the basic image acquisition parameters corresponding to the target inspection time sequence based on the historical inspection data includes: Obtain the historical image data corresponding to each inspection target, calculate the complexity dimension according to the historical image data of each inspection target, and obtain the image threshold according to the complexity dimension; Obtain the historical image acquisition parameters corresponding to each target inspection time sequence, and obtain the basic image acquisition parameters according to the similarity of historical inspection environment parameters and historical image acquisition parameters.

[0024] In this embodiment, calculate the complexity dimension according to the historical image resolution corresponding to each inspection target, use the complexity dimension as the image threshold, and according to the historical image acquisition parameters corresponding to the target inspection time sequence of each inspection target, match similar historical situations according to the similarity of historical inspection environment parameters, and use the image parameters of this historical situation as the basic image acquisition parameters of this inspection target.

[0025] Specifically, calculating the complexity dimension according to the historical image data of each inspection target and obtaining the image threshold according to the complexity dimension includes: Obtain the historical image data of each inspection target, and calculate the detail richness, texture complexity and corresponding weight coefficients according to the similar objects in the historical image data; Obtain the minimum complexity dimension according to the valid images in the historical image data, and use the minimum complexity dimension as the image threshold.

[0026] The complexity dimension includes at least detail richness and texture complexity. Obtain the image data of each inspection target in the historical time sequence. At this time, the historical image data includes at least valid images and invalid images. A valid image is an image currently collected that belongs to an image available for inspection, and an invalid image is an image currently collected that belongs to an image not available for inspection, such as being unable to obtain the actual state of the inspection target based on this image, etc. Calculating the detail richness, texture complexity and corresponding weight coefficients according to the similar objects in the historical image data includes: Divide the historical image data into valid images and invalid images; Obtain the important inspection objects according to the similar objects in the valid images and the similar objects in the invalid images; Calculate the detail richness, texture complexity and corresponding weight coefficients of the important inspection objects based on the valid images and the invalid images for the important inspection objects.

[0027] Through object detection algorithms, such as YOLO, Faster R-CNN, etc., detect similar objects in valid images and invalid images respectively. In this embodiment, obtaining the important inspection objects according to the similar objects in the valid images and the similar objects in the invalid images includes: Obtain the effective frequency of the similar objects appearing in the valid images according to the similarity of each object in the valid images; Obtain the invalid frequency of similar objects in the invalid image according to the similarity of each object in the invalid image; When the valid frequency is greater than the invalid frequency, then the similar object belongs to the important object for inspection.

[0028] For example, for object O, there is a valid frequency of F(O_valid) and an invalid frequency of F(O_invalid). When F(O_valid) > F(O_invalid), this object is considered an important object for inspection.

[0029] Calculate the detail richness, texture complexity, and corresponding weight coefficients of the important objects for inspection based on the valid image and the invalid image, including: Calculate the detail richness of the important objects for inspection in the valid image and the invalid image according to the image processing algorithm; Calculate the texture complexity of the important objects for inspection in the valid image and the invalid image according to the texture analysis algorithm; Use the principal component analysis algorithm to calculate the weight coefficient of detail richness and the weight coefficient of texture complexity based on the valid image and the invalid image.

[0030] Specifically, the detail richness is: D = E + C; where D is the detail richness, E is the edge density, and C is the number of corner points. The image processing algorithm can use the Sobel operator and Canny edge detection to obtain the detail richness of the important objects for inspection by calculating the detail information amount of the object.

[0031] The texture complexity is: T = Contrast + Energy; where T is the texture complexity, Contrast is the contrast, and Energy is the energy. The texture analysis algorithm can use the gray-level co-occurrence matrix (GLCM) to calculate the contrast and energy to obtain the texture complexity of the important objects for inspection.

[0032] Furthermore, through the principal component analysis algorithm, associate the occurrence of each important object for inspection in the valid image and the invalid image with whether the image is valid, and calculate the weight coefficient of detail richness and the weight coefficient of texture complexity for each important object for inspection. Construct a feature matrix with each row representing an image, and the columns including important objects for inspection, detail richness, texture complexity, and the image valid truth value. If the image is a valid image, the image valid truth value is "1", and if the image is an invalid image, the image valid truth value is "0". Calculate the contribution of each feature to the image validity, that is, obtain the loadings of detail richness and texture complexity in the principal component through the principal component analysis algorithm, so as to obtain the weight coefficient of detail richness and the weight coefficient of texture complexity.

[0033] Therefore, the minimum complexity dimension is obtained from the valid images in the historical image data, and the minimum complexity dimension is used as the image threshold. When the complexity dimension of an image meets the image threshold, the image is considered a valid image and is saved.

[0034] Obtain the historical image acquisition parameters corresponding to each target inspection time sequence. Based on the similarity of historical inspection environment parameters and historical image acquisition parameters, the basic image acquisition parameters are obtained, including: Obtain the historical image acquisition parameters and historical inspection environment parameters according to the historical time sequences corresponding to each target inspection time sequence; Retrieve the corresponding historical image acquisition parameters as the basic image acquisition parameters according to the matching situation between the current inspection environment parameters and the historical inspection environment parameters.

[0035] Calculate the similarity between the current inspection environment and the historical environment using historical inspection environment parameters (such as light intensity, temperature, humidity, etc.). According to the similarity calculation result, select the image acquisition parameters corresponding to the most similar historical inspection time sequence as the basic image acquisition parameters. The similarity calculation based on the historical environment enables quick matching and can determine appropriate basic image acquisition parameters for the current inspection environment within a short time, allowing the inspection robot to quickly enter the data acquisition state, saving the time consumed by parameter adjustment, and thus improving the overall inspection efficiency.

[0036] In this embodiment, S3 includes: Obtain real-time inspection images based on the basic image acquisition parameters; If the real-time inspection image meets the image threshold, save the real-time inspection image as the target inspection image; If the real-time inspection image does not meet the image threshold, update the image acquisition parameters according to the difference between the real-time inspection image and the image threshold; re-perform image acquisition with the updated value of the image acquisition parameters.

[0037] When the inspection robot initially performs inspection image acquisition, perform image acquisition according to the basic image acquisition parameters. If the complexity dimension of the acquired image, i.e., the real-time inspection image, meets the image threshold, that is, meets the complexity dimension of the inspection target, it is considered that a valid image has been acquired, and the real-time inspection image is saved as the target inspection image. If the complexity dimension of the real-time inspection image does not meet the image threshold, update the image acquisition parameters according to the real-time inspection image and the image threshold, and re-perform image acquisition with the updated value of the image acquisition parameters.

[0038] At this time, the image acquisition parameters at least include the line frequency, exposure parameters, and the power of the light source controller, and the image acquisition parameters are updated according to the complexity dimension differences of the real-time inspection images. The line frequency refers to the frequency at which the electron beam scans from left to right on the screen, which is directly related to the stability and clarity of the image and indirectly affects the perception of detail richness. The exposure parameters include the exposure time and the aperture size, which affect the brightness and contrast of the image, thereby affecting the performance of detail richness and texture complexity. The power of the light source controller affects the brightness of the lighting system, thereby affecting the recognition of texture complexity and detail richness. The influence relationship between the camera parameters and the detail richness and texture complexity is obtained in advance based on historical image data combined with expert experience, so as to obtain the updated value of the image acquisition parameters according to the influence relationship between the camera parameters and the detail richness and texture complexity and the complexity dimension differences of the real-time inspection images, and fine-tune the basic image acquisition parameters through the complexity dimension differences to ensure the accuracy of the inspection-acquired images.

[0039] Obtaining the inspection robot inspection speed adjustment coefficient based on the target inspection time sequence and the inspection target distance includes: Calculating the inspection interval time by calculating the current time sequence and the target inspection time sequence corresponding to the next inspection target; Calculating the inspection robot inspection speed adjustment coefficient based on the inspection interval time and the inspection target distance.

[0040] According to the distances between inspection targets and the target inspection time sequence, dynamically adjust the moving speed of the inspection robot to ensure that the inspection robot can reach each target inspection point on time while maintaining stable image acquisition quality.

[0041] Calculating the inspection robot inspection speed adjustment coefficient based on the inspection interval time and the inspection target distance as: Where, k is the inspection robot inspection speed adjustment coefficient, l is the inspection target distance, t is the inspection interval time, and v is the current speed of the inspection robot.

[0042] Obtaining the update fluctuation according to the updated value of the image acquisition parameters in the previous period, and updating the basic image acquisition parameters of the next inspection target with the update fluctuation and the inspection speed adjustment coefficient includes: Obtaining the updated value of the image acquisition parameters in the previous period, and using the unified fluctuation value of the updated value of the image acquisition parameters in the previous period as the update fluctuation; Updating the basic image acquisition parameters of the next inspection target according to the update fluctuation, the inspection speed adjustment coefficient, and the speed-line frequency influence relationship.

[0043] Obtain the updated value of the image acquisition parameters for the previous time period (i.e., the previous one or several inspection time sequences), calculate the unified fluctuation value according to the fluctuation situation of the difference between the updated value of the image acquisition parameters and the basic image acquisition parameters. In this embodiment, the unified fluctuation value is calculated based on the average difference fluctuation as the updated fluctuation.

[0044] Obtain the influence of speed adjustment on the line frequency according to the speed-line frequency influence relationship and the inspection speed adjustment coefficient. At the same time, update the basic image acquisition parameters according to the updated fluctuation. For example, when the speed is low, the line frequency needs to be reduced. A large line frequency will cause image distortion. At this time, the image brightness can be increased by increasing the exposure time, and the brightness of the supplementary light can be controlled by the controller to weaken to adjust the balance and reduce power consumption at the same time. When the speed is high, the line frequency needs to be increased. At this time, the exposure time or the brightness of the supplementary light is reduced to adjust the balance. After incorporating the updated fluctuation, obtain the image acquisition parameters that meet the speed requirements according to the influence of speed change on the line frequency, and further update the image acquisition parameters in combination with the updated fluctuation to ensure that the updated basic image acquisition parameters not only compensate for the influence brought by speed change but also have been updated to the image acquisition parameters incorporating the fluctuation of the previous time period when the inspection target is reached, so as to ensure the accuracy of subsequent image acquisition, reduce the adjustment of subsequent shooting, and improve the efficiency of inspection image acquisition.

[0045] As the second embodiment of the present application, obtaining the important inspection objects according to the similar objects in the valid images and the similar objects in the invalid images includes: Obtain the valid frequency of the similar objects appearing in the valid images according to the similarity situation of each object in the valid images; Obtain the invalid frequency of the similar objects appearing in the invalid images according to the similarity situation of each object in the invalid images; When the valid frequency is greater than the invalid frequency, then the similar object belongs to the important inspection object, and obtain the importance dimension value of the important inspection object according to the ratio of the valid frequency and the invalid frequency.

[0046] For example, for object O, there is a valid frequency of F(O_valid) and an invalid frequency of F(O_invalid). When F(O_valid) > F(O_invalid), it is considered that the object is an important inspection object. At this time, the importance dimension value of this important inspection object is Compare the number of times an object appears in the valid image with the number of times the object appears in the invalid image to obtain whether the object belongs to the important inspection object.

[0047] Since there may be multiple objects existing together when the image has analysis value, but the image does not have analysis value when a single image exists alone, obtaining the important inspection objects according to the similar objects in the valid images and the similar objects in the invalid images further includes: When the effective frequency is less than or equal to the invalid frequency, the similar object belongs to the object to be determined, and the combined dimension value of the object to be determined is obtained according to the combined efficiency of all objects to be determined.

[0048] That is, when F(O_effective) ≤ F(O_invalid), the object is considered as an object to be determined. At this time, the historical image data of all objects to be determined is obtained, and the combined dimension value of the objects to be determined is obtained according to the combined efficiency between the objects to be determined. For example, there is an object H, and F(H_effective) ≤ F(H_invalid), but for the combination of objects O and H, F(H, O_effective) > F(H, O_invalid), and the combined efficiency of their combination is used as the combined dimension value of the object to be determined.

[0049] At this time, correspondingly, the complexity dimension includes the detail richness, texture complexity, importance dimension value, and combined dimension value.

[0050] The complexity dimension is calculated based on the historical image data of each inspection target, and the image threshold is obtained according to the complexity dimension, including: Calculating the detail richness, texture complexity, importance dimension value of the similar object, and the combined dimension value between the similar objects based on the historical image data of each inspection target; Calculating the sum of the detail richness, texture complexity, importance dimension value of the similar object, and the combined dimension value between the similar objects as the complexity dimension; Taking the minimum complexity dimension in the historical image data as the image threshold.

[0051] Taking the minimum complexity dimension of the effective images in the historical image data as the image threshold can screen out images with sufficient complexity and quality. The minimum complexity dimension represents the lowest image quality standard that meets the effective detection and analysis of the inspection target in the historical inspection. Using it as the image threshold corresponding to the inspection target not only ensures the adaptability of the image threshold but also ensures that the collected images meet the requirements of analysis and processing, ensuring the accuracy of image acquisition.

[0052] At the same time, considering the combined dimension value between the similar objects and incorporating the importance dimension value of the similar objects can avoid missing necessary acquisition objects or related acquisition objects, further improving the accuracy of image acquisition.

[0053] As the third embodiment of the present application, an image acquisition control system based on an inspection robot includes: A visual sensor that performs image acquisition based on image acquisition parameters; A main control unit that obtains basic image acquisition parameters according to historical inspection data and transmits them to the visual sensor; A storage unit for storing the inspection images collected by the visual sensor; A processing unit is configured to analyze real-time inspection images, obtain updated values of image acquisition parameters, and transmit them to a vision sensor.

[0054] In this embodiment, the main control unit is responsible for the calculation and processing of historical inspection data, while the processing unit only performs real-time update calculations based on real-time inspection images. The distributed computing unit shares the computing pressure, improves the processing efficiency of the processing unit, and further improves the image acquisition control efficiency to meet the inspection time requirements.

[0055] In some other cases, an image acquisition control system based on an inspection robot further includes: A light source controller is configured to control the output power based on image acquisition parameters to adjust the light source.

[0056] The vision sensor can be a line array camera, a area array camera, etc. Due to the requirements of resolution and accuracy, in this embodiment, the vision sensor is a line array camera array.

[0057] The above specific implementation manners are the preferred implementation manners of an image acquisition control method and system based on an inspection robot of the present application, and do not limit the specific implementation scope of the present application. The scope of the present application includes but is not limited to this specific implementation manner. Any equivalent changes made according to the shape and structure of the present application are within the protection scope of the present application.

Claims

1. A method for image acquisition and control based on a patrol robot, characterized in that: The steps include: S1: Acquire the inspection target and target inspection timing of the inspection robot based on the target inspection route; S2: Acquire the image threshold corresponding to the inspection target and the basic image acquisition parameters corresponding to the inspection timing of the target based on the historical inspection data; S3: acquiring a real-time inspection image based on the basic image acquisition parameters, updating the image acquisition parameters based on the basic image acquisition parameters according to the image threshold and the real-time inspection image, and re-executing the image acquisition with the updated image acquisition parameter values; S4: Obtaining the inspection speed adjustment coefficient of the inspection robot based on the target inspection timing and the inspection target distance; S5: Obtaining an update fluctuation according to the image acquisition parameter update value of the previous time period, and updating the basic image acquisition parameters of the next inspection target with the update fluctuation and the inspection speed adjustment coefficient.

2. The image acquisition control method based on the inspection robot according to claim 1, characterized in that: The S1 includes: Construct an inspection topology map based on historical inspection targets in historical inspection data; Obtain inspection targets based on matching of target inspection routes and inspection topology graphs; The target inspection timing of each inspection target is obtained based on the target inspection route and target inspection time.

3. The image acquisition control method based on the inspection robot according to claim 1, characterized in that: The S2 includes: The image threshold corresponding to the inspection target and the basic image acquisition parameters corresponding to the inspection timing of the target are obtained based on the historical inspection data, including: Obtain historical image data corresponding to each inspection target, perform complexity dimension calculation based on the historical image data of each inspection target, and obtain image thresholds based on the complexity dimension; Obtain the historical image acquisition parameters corresponding to each target inspection sequence, and obtain the basic image acquisition parameters based on the similarity of historical inspection environment parameters and historical image acquisition parameters.

4. The image acquisition control method based on the inspection robot according to claim 1, characterized in that: The S3 includes: Acquire real-time inspection images based on basic image acquisition parameters; If the real-time inspection image meets the image threshold, the real-time inspection image is saved as the target inspection image; If the real-time inspection image does not meet the image threshold, the image acquisition parameters are updated according to the difference between the real-time inspection image and the image threshold; and the image acquisition is re-executed with the updated value of the image acquisition parameters.

5. The image acquisition control method based on the inspection robot according to claim 1, characterized in that: The S4 includes: Calculate the inspection interval time by calculating the target inspection timing corresponding to the current timing and the next inspection target; The inspection robot's inspection speed adjustment coefficient is calculated based on the inspection interval time and the inspection target distance.

6. The image acquisition control method based on the inspection robot according to claim 1, characterized in that: The S5 includes: Obtain the updated value of the image acquisition parameter in the previous time period, and use the uniform fluctuation value of the updated value of the image acquisition parameter in the previous time period as the updated fluctuation; The basic image acquisition parameters of the next inspection target are updated according to the update fluctuation and the inspection speed adjustment coefficient combined with the speed line frequency influence relationship.

7. The image acquisition control method based on the inspection robot according to claim 3, characterized in that: The calculation of complexity dimension based on the historical image data of each inspection target and obtaining the image threshold based on the complexity dimension include: Obtain historical image data of each inspection target, and calculate detail richness, texture complexity and corresponding weight coefficients based on similar objects in the historical image data; The minimum complexity dimension is obtained according to the valid images in the historical image data, and the minimum complexity dimension is used as the image threshold.

8. The image acquisition control method based on the inspection robot according to claim 7, characterized in that: The calculating of detail richness, texture complexity and corresponding weight coefficients according to similar objects of historical image data includes: dividing the historical image data into valid images and invalid images; Obtain important inspection objects based on similar objects in valid images and similar objects in invalid images; The detail richness, texture complexity and corresponding weight coefficients of the important inspection objects are calculated based on the valid images and the invalid images.

9. The image acquisition control method based on the inspection robot according to claim 1, characterized in that: The complexity dimensions include detail richness, texture complexity, importance dimension value and combined dimension value.

10. An inspection robot-based image acquisition and control system, used to implement the method according to any one of claims 1 to 9, characterized in that: include: A visual sensor, performing image acquisition based on the image acquisition parameters; The main control unit obtains basic image acquisition parameters based on historical inspection data and transmits them to the visual sensor; A storage unit, used for storing inspection images collected by the visual sensor; The processing unit is used to analyze the real-time inspection image, obtain the updated value of the image acquisition parameter and transmit it to the visual sensor.

Images