Preset bit self-correction method and system based on image offset and pan-tilt equipment parameterization

Abstract

The invention discloses a preset bit self-correction method and system based on image offset and pan-tilt equipment parameterization. The method comprises the following steps: constructing a preset bit reference image library; establishing a unified PTZ parameterized ledger management system of the holder; quantitatively analyzing pixel deviation values of the real-time screenshot and the reference image in horizontal and vertical directions by utilizing an image deviation model in an intelligent analysis system; based on the deviation value, the PTZ parameter of the holder is dynamically adjusted through an adaptive iterative algorithm, a holder camera is driven to rotate, the real-time image and the reference image are converged and consistent, and therefore automatic correction of the preset position is achieved. According to the invention, the problem of preset bit offset caused by equipment aging, replacement or human errors is effectively solved, the dependence of system maintenance on human experience and the maintenance cost are greatly reduced, and the automation degree and reliability of the electric power inspection system are improved.

Description

Technical Field

[0001] This invention relates to the field of intelligent monitoring technology, and in particular to a preset position self-correction method and system based on image offset and pan-tilt device parameterization. Background Technology

[0002] In the power sector, various self-inspection and patrol systems are widely used. These systems primarily employ PTZ cameras, self-inspection robots, and drones for joint land-air inspections. With the rapid development of deep learning and AI technologies in recent years, unmanned inspections are now possible in most scenarios. The intelligent analysis of these systems allows for automatic reporting and alerts after identifying problems, reducing the need for manual judgment in many situations. However, because most video surveillance systems were built relatively early, the firmware versions of the front-end equipment vary, and some equipment is not maintained in a timely manner. This leads to problems such as blurry screenshots, PTZ malfunctions, or worn gears preventing accurate screenshots. The clarity and accuracy of screenshots are crucial analytical criteria for existing intelligent analysis systems in the power industry.

[0003] In traditional power sector monitoring systems, the configuration of monitoring screens and preset positions relies primarily on manual judgment. During long-term maintenance, personnel turnover and replacement of front-end equipment can alter the original monitoring screens and preset positions, leading to various misjudgments and false alarms in the power industry's intelligent analysis system. This impacts the time and manpower costs of inspection and maintenance. Furthermore, the maintenance of monitoring screens and preset positions is extremely costly in the long run due to their large number and complex scenarios.

[0004] In terms of equipment, existing monitoring equipment in the power industry, such as substations, was installed relatively early. Coupled with the lack of timely maintenance of some equipment, the field of view of the front-end PTZ cameras gradually shifts due to factors such as zooming, PTZ tilt / horizontal rotation, and mechanical wear, resulting in inconsistencies between the video surveillance image and the preset position image. Furthermore, when replacing with new equipment, it is necessary to reconfigure the preset positions.

[0005] In terms of personnel, unlike traditional external monitoring systems, the power industry is a highly specialized field. For example, in substation scenarios, personnel need a clear understanding of various scenes and components; otherwise, they cannot select and configure monitoring screens and preset positions, including naming conventions. Therefore, traditional power monitoring systems rely heavily on long-term experience for equipment and monitoring points, requiring a relatively high level of expertise and increasing training costs. During training or maintenance, human error can lead to changes in monitoring screens and preset positions. Since the original monitoring screens are unknown, preset positions can only be configured based on memory and experience. Over time, accumulated errors cause preset screen misalignment.

[0006] Therefore, the existing power industry inspection systems and intelligent analysis systems rely heavily on the clarity and accuracy of the preset positions of the front-end equipment PTZ cameras, and the cost of maintaining and correcting the preset position image configuration is very high and inefficient. Summary of the Invention

[0007] The purpose of this invention is to provide a preset position self-correction method and system based on image offset and gimbal device parameterization. Based on intelligent analysis system offset technology, and in conjunction with the PTZ parameters of the control front-end device, precise rotation is performed to achieve the preset position self-correction function.

[0008] The technical solution adopted in this invention is:

[0009] A preset position self-correction method based on image offset and gimbal device parameterization includes the following steps:

[0010] Build a benchmark image library: Take screenshots of preset position images from the gimbal camera, and store the corrected and anomaly-marked preset position images into the benchmark image library;

[0011] Preset position ledger parameterization: Establish a PTZ parameter system for each PTZ camera based on the relationship between the longitudinal and lateral directions and the zoom, as well as the focal length; add PTZ parameters to the preset position ledger of the front-end equipment and unify the PTZ parameters of different manufacturers into a standard range; where P is the horizontal rotation angle, T is the vertical pitch angle, and Z is the zoom factor.

[0012] Historical patrol screenshot analysis: Retrieve historical patrol screenshots from the historical database of patrol tasks within a specified time window and determine the offset; when the historical patrol screenshot is determined to be offset, the corresponding PTZ camera jumps to the pre-position and waits for the pre-position self-correction; when the pre-position of the PTZ camera in the current time window is not offset, generate a report result and report it to the superior management system.

[0013] Pre-position self-correction: Obtain a real-time screenshot of the pre-position of the PTZ camera and determine the offset; when there is an offset in the real-time screenshot, adjust the PTZ parameters through the PTZ control system and perform iterative correction operations until the deviation value is within the preset threshold to achieve pre-position self-correction, and finally generate a report result and report it to the superior management system.

[0014] The method for determining the offset is as follows: calculate the deviation value between the current screenshot and the reference image. When the deviation value exceeds the set deviation threshold, it is determined that there is an offset in the preset position. The deviation value includes the P-axis deviation, T-axis deviation and Z-axis deviation.

[0015] Furthermore, building a benchmark image library also includes: dynamically updating the preset benchmark images that are added or removed from the library, and combining this with manual analysis to regularly maintain the accuracy of the benchmark image library.

[0016] Furthermore, the unified range of the PTZ parameter system is defined as follows:

[0017] The P parameter range is 0~3600, corresponding to 0~360 degrees of horizontal rotation;

[0018] The T-parameter range is 0~3600, corresponding to 0~360 degrees of vertical pitch.

[0019] The Z parameter ranges from 0 to 400, corresponding to a multiplier of 0 to 40.

[0020] Furthermore, during offset determination: screenshots with PTZ parameter deviation values ​​exceeding the preset range are judged as irrelevant images and are not subject to offset correction.

[0021] Furthermore, the steps of the iterative correction operation are as follows:

[0022] S1, Initialization parameters: Set the maximum number of iterations to avoid infinite loops; During the iteration process, first adjust the direction of the larger deviation value between the P-axis and T-axis until the deviations of both the P-axis and T-axis are within the threshold.

[0023] Preset initial P-axis coordinate P1; ;

[0024] S2, Construct the objective function for pre-position self-correction and determine the convergence condition. The expression for the objective function is: Where X represents the target offset, and the sign of X indicates the direction of the deviation; P is the initial P-axis coordinate of the requested pre-position self-correction; the convergence condition is... ,in Let t be the target offset in round t, and ε be a preset reasonable threshold.

[0025] S3, Direction Detection: Confirm the direction of deviation convergence through a multi-step detection sequence, including positive and negative detection, and calculate the improvement index of each direction to obtain the decision adjustment direction;

[0026] Specifically, S3 includes the following steps:

[0027] S3-1, Calculate the forward detection sequence, forward error, and corresponding positive improvement index;

[0028] Forward detection sequence: ;

[0029] in, These are the coordinates of the P-axis after the first offset; Let be the coordinate value of the P-axis after the i-th positive direction offset; offset in the positive direction The target offset from the reference image; Functions for analyzing the host machine. This is to improve the threshold, i.e., the step size. The overall process involves shifting in one direction and sending a request to the intelligent analysis host to retrieve the deviation value from the baseline image. If the deviation decreases, it indicates the direction is correct; if the deviation increases, it indicates the direction is incorrect.

[0030] Positive error: ;

[0031] in, This represents the error between the function return value after the i-th positive direction offset and the target offset value. Let t be the target offset in round t;

[0032] Positive improvement indicators: ;

[0033] Where N is the number of probes, This represents the absolute value of the difference between the two values ​​before and after the first time.

[0034] S3-2, Calculate the negative detection sequence and the corresponding negative improvement index;

[0035] Negative detection sequence: ;

[0036] in, The coordinate value of the P-axis after the i-th negative direction offset; Offset in the negative direction The target offset from the reference image; Functions for analyzing the host machine. This is to improve the threshold, i.e., the step size;

[0037] Positive error: ;

[0038] in, The error between the function return value after the i-th negative direction offset and the target offset value;

[0039] Negative improvement indicators: ;

[0040] S3-3, Based on the improvement indicators in each direction, the decision adjustment direction is obtained; the direction decision function is:

[0041] ;

[0042] in, To guide the direction, convergence occurs in either the positive or negative direction; The threshold value is a positive real number, representing the minimum acceptable level of efficiency improvement per unit step size. To improve the threshold, it is used to filter out minor improvements.

[0043] S4, based on the direction detection results and decision adjustments, adaptively calculate the initial step size by adjusting the direction d. The steps include:

[0044] S4-1, Based on the direction detection results, calculate the positive sensitivity estimate and the negative sensitivity estimate;

[0045] Positive sensitivity estimation: ;

[0046] Negative sensitivity estimation: ;

[0047] S4-2, Determine the comprehensive sensitivity S based on the decision adjustment direction d:

[0048] Overall sensitivity:

[0049] S4-3, Calculate the initial step size: ;in, This represents the maximum step size in a single operation, i.e., the maximum offset value P for each movement, such as 10. This represents the minimum step size in a single move, i.e., the minimum offset value P for each movement, such as 10; This is the target offset, and this value is generally determined by the site configuration, such as a threshold of 3% offset. The normalized bandwidth is 1000 pixels. It is 30. Offset value Analyze the actual pixel deviation values ​​returned by the host in real time. Utilize previous data. The returned deviation value is 100, which is (100-10) / 7.8. That is, the initial step size should be set to 1 or 2.

[0050] S5, perform iterative parameter adjustments until convergence, including the following steps:

[0051] S5-1, Parameter Update: Update the preset bit according to the current iteration number k. And calculate the offset by the corresponding function evaluation value. ;

[0052] S5-2, Convergence Check: After each iteration, check whether the offset meets the threshold condition. If it does, terminate the correction and return. As the final preset position; otherwise, continue with the subsequent steps; specifically, if The algorithm terminates and returns. .

[0053] S5-3, Direction Verification and Adjustment: Calculate the error change trend of the most recent M iterations. When the error trend shows that the error continues to increase, execute S3, and use the current point... Based on this, adjust the step size to Update direction d;

[0054] Specifically, record the most recent Error change trend in each iteration:

[0055] ;

[0056] like If the error continues to increase, then direction detection will be re-executed; using the current point... Based on, step size =( , Perform multi-step direction detection and update direction. ;

[0057] S5-4, Overshoot Detection and Step Size Adjustment: Detects whether the current function evaluation value exceeds the target value, i.e. If so, it means the target value has been exceeded, and the step size for the next step is halved. =0.5 And fine-tune the direction = Otherwise, it indicates that the overshoot did not occur, and the next step size is adaptively adjusted based on the error change of the current iteration, i.e.:

[0058] { ;

[0059] in , This is the step size adjustment factor.

[0060] S5-5, Update Iteration Status: Settings Repeat steps S5-1 to S5-5 until the convergence condition is met or the maximum number of iterations is reached. .

[0061] Convergence check: After each iteration, check whether the deviation value meets the threshold condition. If it does, terminate the correction.

[0062] Furthermore, the convergence condition for pre-position self-correction is:

[0063] , ;

[0064] in, The offset in the P-axis direction that the self-correction system ultimately returns; A reasonable threshold; The normalized image width is used to represent the screenshot. The size of screenshots taken from the field equipment may not be consistent, so normalization is necessary. The reference image may also have inconsistent length and width, so it is generally normalized, such as to 960*512. This is a reasonable percentage setting, such as k=0.03, which means the pixel error is within 0.03.

[0065] Furthermore, the method also includes a self-correcting task management function:

[0066] It provides a task configuration interface and supports batch distribution of correction tasks to multiple PTZ camera locations;

[0067] Monitor the correction process log in real time and report the correction results, including equipment malfunction, successful correction, exceeding the maximum number of iterations, or irrelevant image.

[0068] Furthermore, the method also includes gimbal control access management:

[0069] Set the gimbal control priority, with the self-correction function having a lower priority than the inspection task. The self-correction function will be executed after the inspection task is completed. Use the latest screenshot from the inspection task as the real-time screenshot to reduce the time the gimbal is occupied.

[0070] Furthermore, the method of the present invention is applicable to power industry monitoring systems, including substation inspection systems, transmission and distribution line monitoring systems, unmanned operation and maintenance monitoring stations, or intelligent inspection systems.

[0071] A preset position self-correction system based on image offset and gimbal device parameterization includes:

[0072] The reference image library module is used to store and manage preset reference images of the PTZ camera;

[0073] The parameter management module is used to maintain the PTZ parameter system of the PTZ camera;

[0074] The intelligent analysis host is used to calculate the deviation between the screenshot and the reference image and to determine whether there is an offset.

[0075] The self-correction control module is used to execute the iterative correction algorithm and adjust the PTZ parameters.

[0076] The task management module is used to configure and monitor correction tasks, providing a task configuration interface, real-time log monitoring, and correction result reporting functions.

[0077] This invention, employing the above technical solutions, offers the following advantages over existing technologies: 1. Parameterizing the direction of preset positions facilitates management, enabling quantification and datafication. It eliminates reliance on traditional experience and memory, and reduces the need for highly skilled maintenance personnel. Even with a large number of preset positions and complex scenarios, quantification and parameterization management are possible. 2. Parameterizing the direction of preset positions allows for the restoration of the original inspection point image when camera equipment replacement or human error causes changes in the inspection point's view. This significantly reduces maintenance costs. 3. Based on a specially trained image offset model using an existing intelligent analysis system, the deviation between real-time screenshots and the baseline image can be quantified. This not only accurately analyzes the preset position image offset but also serves as a preset position offset alarm that can be directly sent for manual judgment and correction. 4. Based on the intelligent analysis system's offset technology, combined with the PTZ parameters of the control front-end equipment, precise rotation is achieved to realize the preset position self-correction function. 5. Through practical engineering applications, it can be found that some preset positions are not set reasonably, or there are abnormal images or abnormal PTZ devices. Both inspection systems and monitoring systems can effectively identify existing problems. Attached Figure Description

[0078] The present invention will be further described in detail below with reference to the accompanying drawings and specific embodiments;

[0079] Figure 1 This is a schematic diagram of the preset position self-correction method and system based on image offset and gimbal device parameterization of the present invention;

[0080] Figure 2 This is a schematic diagram illustrating the working principle of the present invention. Detailed Implementation

[0081] To make the objectives, technical solutions, and advantages of the embodiments of this application clearer, the technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings.

[0082] like Figure 1As shown in Figure 2, this invention discloses a preset position self-correction method and system based on image offset and PTZ device parameterization. It configures a reference image for the preset position and adds parameterization support based on the PTZ camera manufacturer of the front-end device, establishing a PTZ system based on the longitudinal, lateral, and zoom relationships of each PTZ camera. An intelligent analysis system analyzes the difference between the reference image and the real-time screenshot under the PTZ system and provides the deviation pixel value in the PT direction. By checking whether the deviation value is within a reasonable range, it is determined whether the preset position image has shifted. Then, by controlling the PTZ parameters of the front-end device to perform precise rotation, preset position correction is achieved. After multiple correction actions, when the deviation value is within a reasonable threshold, the current image is set to the original preset position, thus realizing the preset position self-correction function.

[0083] A preset position self-correction method based on image offset and gimbal device parameterization includes the following steps:

[0084] Build a benchmark image library: Take screenshots of preset position images from the gimbal camera, and store the corrected and anomaly-marked preset position images into the benchmark image library;

[0085] Preset position ledger parameterization: Establish a PTZ parameter system for each PTZ camera based on the longitudinal and lateral relationships, zoom, and focal length of each PTZ camera; add PTZ parameters to the preset position ledger of the front-end equipment and unify the PTZ parameters of different manufacturers into a standard range; where P is the horizontal rotation angle, T is the vertical pitch angle, and Z is the zoom factor; specifically, through the existing preset position ledger and after establishing a benchmark image library, add preset position ledger parameterization management to establish a PTZ system based on the longitudinal and lateral relationships, zoom, and focal length of each PTZ camera.

[0086] Historical patrol screenshot analysis: Retrieve historical patrol screenshots from the historical database of patrol tasks within a specified time window and determine the offset; when the historical patrol screenshot is determined to be offset, the corresponding PTZ camera jumps to the pre-position and waits for the pre-position self-correction; when the pre-position of the PTZ camera in the current time window is not offset, generate a report result and report it to the superior management system.

[0087] Pre-position self-correction: Obtain a real-time screenshot of the pre-position of the PTZ camera and determine the offset; when there is an offset in the real-time screenshot, adjust the PTZ parameters through the PTZ control system and perform iterative correction operations until the deviation value is within the preset threshold to achieve pre-position self-correction, and finally generate a report result and report it to the superior management system.

[0088] The method for determining the offset is as follows: calculate the deviation value between the current screenshot and the reference image. When the deviation value exceeds the set deviation threshold, it is determined that there is an offset in the preset position. The deviation value includes the P-axis deviation, T-axis deviation and Z-axis deviation.

[0089] Furthermore, building the benchmark image library also includes dynamically updating newly added or removed preset position benchmark images, and combining this with manual review to regularly maintain the accuracy of the benchmark image library. Specifically, using the existing preset position ledger and the system's screenshot module, all existing preset position images in the system are stored. Manual marking of equipment malfunctions and gimbal rotation malfunctions is then performed. For image malfunctions, especially for preset position images, corrections are required before re-entry into the library. New and removed images must be processed promptly.

[0090] Furthermore, the unified range of the PTZ parameter system is defined as follows:

[0091] The P parameter range is 0~3600, corresponding to 0~360 degrees of horizontal rotation;

[0092] The T-parameter range is 0~3600, corresponding to 0~360 degrees of vertical pitch.

[0093] The Z parameter ranges from 0 to 400, corresponding to a multiplier of 0 to 40.

[0094] Specifically, PTZ parameters are added to the preset position ledger of the front-end equipment. Since the PTZ relationships of the front-end equipment's PTZ devices vary between manufacturers, a unified range is needed. The P parameter (horizontal parameter 0~3600, representing 0~360 degrees), T parameter (vertical parameter representing 0~360 degrees), and Z parameter (magnification parameter 0~400, representing 0~40x) are internally converted according to the manufacturer. This eliminates the need to differentiate types based on the manufacturer, simplifying management.

[0095] Furthermore, during offset judgment: screenshots with PTZ parameter deviation values ​​exceeding a preset range are considered irrelevant images and no correction processing is performed. This invention deploys an image offset technology model that supports real-time response. Since the correction function is a continuous set of control analysis results, the return time should not exceed 3 seconds. It can accurately determine the deviation values ​​between the real-time image and the reference image on the P-axis (horizontal) and T-axis (vertical), as well as the zoom deviation value. Images with excessive zoom or offset are considered irrelevant and the results are returned directly without processing. Correction is only performed after the deviation value is less than the value of the reference image and the image is overlaid.

[0096] Furthermore, the steps of the iterative correction operation are as follows:

[0097] S1, Initialization parameters: Set the maximum number of iterations to avoid infinite loops; During the iteration process, first adjust the direction of the larger deviation value between the P-axis and T-axis until the deviations of both the P-axis and T-axis are within the threshold.

[0098] Preset initial P-axis coordinate P1; ;

[0099] S2, Construct the objective function for pre-position self-correction and determine the convergence condition. The expression for the objective function is: Where X represents the target offset, and the sign of X indicates the direction of the deviation. X is of type Int and can be either positive or negative; the sign represents different directions (left or right). P is the initial P-axis coordinate for requesting pre-position self-correction. The convergence condition is... ,in Let t be the target offset in round t, and ε be a preset reasonable threshold.

[0100] S3, Direction Detection: A multi-step detection sequence is used to confirm the direction of deviation convergence, including positive and negative detection, and improvement indicators for each direction are calculated to determine the decision adjustment direction. Because direction exists, and in engineering applications, a single test may lead to directional errors or insignificant changes, multi-step detection is necessary to confirm the direction. N should be 2 to 4 steps for a reasonable number of steps. Specifically, S3 includes the following steps:

[0101] S3-1, Calculate the forward detection sequence, forward error, and corresponding positive improvement index;

[0102] Forward detection sequence: ;

[0103] in, These are the coordinates of the P-axis after the first offset; Let be the coordinate value of the P-axis after the i-th positive direction offset; offset in the positive direction The target offset from the reference image; Functions for analyzing the host machine. This is to improve the threshold, i.e., the step size. The overall process involves shifting in one direction and sending a request to the intelligent analysis host to retrieve the deviation value from the baseline image. If the deviation decreases, it indicates the direction is correct; if the deviation increases, it indicates the direction is incorrect.

[0104] Positive error: ;

[0105] in, This represents the error between the function return value after the i-th positive direction offset and the target offset value. Let t be the target offset in round t;

[0106] Positive improvement indicators: ;

[0107] Where N is the number of probes, This represents the absolute value of the difference between the two values ​​before and after the first time.

[0108] S3-2, Calculate the negative detection sequence and the corresponding negative improvement index;

[0109] Negative detection sequence: ;

[0110] in, The coordinate value of the P-axis after the i-th negative direction offset; Offset in the negative direction The target offset from the reference image; Functions for analyzing the host machine. This is to improve the threshold, i.e., the step size;

[0111] Positive error: ;

[0112] in, The error between the function return value after the i-th negative direction offset and the target offset value;

[0113] Negative improvement indicators: ;

[0114] The absolute value of the difference between the first and second times.

[0115] Specifically, the meaning of the entire function for positive or negative improvement indicators is that the difference between the return value of each step and the target value is calculated, divided by the sum of the actual step size (each step size * the nth step size), and then divided by the number of probes (average).

[0116] For example, the following is an illustration: Assume: =100, =60 (pixels) (initial point), target value It is 10 pixels. - ;

[0117] Assumption , =50; = - =40;

[0118] Assumption ; ;

[0119] Assumption ; ;

[0120] Then there is, ;

[0121] + + )

[0122] = + + )

[0123] = +7.5+6)

[0124] = )

[0125] =7.8 (approximately equal to).

[0126] S3-3, Based on the improvement indicators in each direction, the decision adjustment direction is obtained; the direction decision function is:

[0127] ;

[0128] in, To guide the direction, convergence occurs in either the positive or negative direction; The threshold value is a positive real number, representing the minimum acceptable level of efficiency improvement per unit step size. To improve the threshold, it is used to filter out minor improvements.

[0129] S4, based on the direction detection results and decision adjustments, adaptively calculate the initial step size by adjusting the direction d. This includes the following steps:

[0130] S4-1, Based on the direction detection results, calculate the positive sensitivity estimate and the negative sensitivity estimate;

[0131] Positive sensitivity estimation: ;

[0132] Negative sensitivity estimation: ;

[0133] S4-2, Determine the comprehensive sensitivity S based on the decision adjustment direction d:

[0134] Overall sensitivity:

[0135] S4-3, Calculate the initial step size: ;

[0136] S5, perform iterative parameter adjustments until convergence, including the following steps:

[0137] S5-1, Parameter Update: Update the preset position based on the current iteration number k. And calculate the offset by the corresponding function evaluation value. ;

[0138] S5-2, Convergence Check: After each iteration, check whether the offset meets the threshold condition. If it does, terminate the correction and return. As the final preset position; otherwise, continue with the subsequent steps; specifically, if The algorithm terminates and returns. .

[0139] S5-3, Direction Verification and Adjustment: Calculate the error change trend of the most recent M iterations. When the error trend shows that the error continues to increase, execute S3, and use the current point... Based on this, adjust the step size to Update direction d;

[0140] Specifically, record the most recent Error change trend in each iteration:

[0141] ;

[0142] like If the error continues to increase, then direction detection will be re-executed; using the current point... Based on, step size =( , Perform multi-step direction detection and update direction. ;

[0143] S5-4, Overshoot Detection and Step Size Adjustment: Detects whether the current function evaluation value exceeds the target value, i.e. If so, it means the target value has been exceeded, and the step size for the next step is halved. =0.5 And fine-tune the direction = Otherwise, it indicates that the overshoot did not occur, and the next step size is adaptively adjusted based on the error change of the current iteration, i.e.:

[0144] { ;

[0145] in , This is the step size adjustment factor.

[0146] S5-5, Update Iteration Status: Settings Repeat steps S5-1 to S5-5 until the convergence condition is met or the maximum number of iterations is reached. .

[0147] Convergence check: After each iteration, check whether the deviation value meets the threshold condition. If it does, terminate the correction.

[0148] Specifically, since the correction function will invoke the front-end device's PTZ (pan-tilt-zoom) unit, if other systems or users invoke the PTZ at this time, it may cause errors in the preset position screenshot. Therefore, it is necessary to unify PTZ control permissions and set priorities. If there is a patrol system, patrol tasks should be controlled first, and other users should be prioritized over the self-correction function. The correction function will call preset positions extensively and occupy PTZ control time for a long time. Therefore, the self-correction function should not be used too frequently; it should be used at weekly or monthly intervals. To address this, if a patrol system exists in the system, a screenshot of the patrol task (the latest screenshot of the patrol points within seven days) can be sent as the current screenshot to the intelligent analysis system. If the intelligent analysis system returns a result indicating that correction is not needed, it can be skipped. This method can significantly save time spent on self-correction and PTZ control system time. If the intelligent analysis system returns an offset or irrelevant result, the corresponding preset position should be called again, and a real-time image should be captured again for correction. For points with deviations, the PTZ control system moves according to the PTZ value and returns the value from the intelligent analysis system to reduce the deviation value until a reasonable threshold is reached, at which point the correction is successful. The maximum number of corrections is set; otherwise, it will loop indefinitely.

[0149] Further, X and Y represent the offset values ​​in the P and T directions, respectively, in pixels. Control logic is used to converge the preset position to a reasonable range. In engineering, the direction with the larger X or Y value returned by the P-axis or T-axis should be converged first, until both are within a reasonable range. Here, P-axis convergence is used as the calculation method. The convergence condition for pre-position self-correction is:

[0150] , ;

[0151] in, The offset in the P-axis direction that the self-correction system ultimately returns; A reasonable threshold; The normalized image width is used because the size of screenshots from the field equipment may not be consistent, so normalization is necessary. The reference image may also have inconsistent length and width, so it is generally normalized, such as to 960*512. This is a reasonable percentage setting, such as k=0.03, which means the pixel error is within 0.03.

[0152] Furthermore, the method also includes a self-correcting task management function:

[0153] It provides a task configuration interface and supports batch distribution of correction tasks to multiple PTZ camera locations;

[0154] Monitor the correction process log in real time and report the correction results, including equipment malfunction, successful correction, exceeding the maximum number of iterations, or irrelevant image.

[0155] Specifically, the self-correction system has a task configuration interface, real-time log monitoring, and correction result reporting (equipment abnormality, correction success, correction exceeding the maximum limit number of times, irrelevant images). It can also batch issue correction tasks to multiple points simultaneously, thereby supporting the large-scale deployment of substation or distribution station inspection systems.

[0156] Furthermore, the method also includes gimbal control access management:

[0157] Set the gimbal control priority, with the self-correction function having a lower priority than the inspection task. The self-correction function will be executed after the inspection task is completed. Use the latest screenshot from the inspection task as the real-time screenshot to reduce the time the gimbal is occupied.

[0158] Furthermore, the method of the present invention is applicable to power industry monitoring systems, including substation inspection systems, transmission and distribution line monitoring systems, unmanned operation and maintenance monitoring stations, or intelligent inspection systems.

[0159] A preset position self-correction system based on image offset and gimbal device parameterization includes:

[0160] The reference image library module is used to store and manage preset reference images of the PTZ camera;

[0161] The parameter management module is used to maintain the PTZ parameter system of the PTZ camera;

[0162] The intelligent analysis host is used to calculate the deviation between the screenshot and the reference image and to determine whether there is an offset.

[0163] The self-correction control module is used to execute the iterative correction algorithm and adjust the PTZ parameters.

[0164] The task management module is used to configure and monitor correction tasks, providing a task configuration interface, real-time log monitoring, and correction result reporting functions.

[0165] This invention, employing the above technical solutions, offers the following advantages over existing technologies: 1. Parameterizing the direction of preset positions facilitates management, enabling quantification and datafication. It eliminates reliance on traditional experience and memory, and reduces the need for highly skilled maintenance personnel. Even with a large number of preset positions and complex scenarios, quantification and parameterization management are possible. 2. Parameterizing the direction of preset positions allows for the restoration of the original inspection point image when camera equipment replacement or human error causes changes in the inspection point's view. This significantly reduces maintenance costs. 3. Based on a specially trained image offset model using an existing intelligent analysis system, the deviation between real-time screenshots and the baseline image can be quantified. This not only accurately analyzes the preset position image offset but also serves as a preset position offset alarm that can be directly sent for manual judgment and correction. 4. Based on the intelligent analysis system's offset technology, combined with the PTZ parameters of the control front-end equipment, precise rotation is achieved to realize the preset position self-correction function. 5. Through practical engineering applications, it can be found that some preset positions are not set reasonably, or there are abnormal images or abnormal PTZ devices. Both inspection systems and monitoring systems can effectively identify existing problems.

[0166] Obviously, the described embodiments are only a part of the embodiments of this application, not all of them. Without conflict, the embodiments and features in the embodiments of this application can be combined with each other. The components of the embodiments of this application described and illustrated herein can generally be arranged and designed in various different configurations. Therefore, the detailed description of the embodiments of this application is not intended to limit the scope of the claimed application, but merely to illustrate selected embodiments of this application. All other embodiments obtained by those skilled in the art based on the embodiments of this application without inventive effort are within the scope of protection of this application.

Claims

1. A preset position self-correction method and system based on image offset and gimbal device parameterization, characterized in that: Includes the following steps: Build a benchmark image library: Take screenshots of preset position images from the gimbal camera, and store the corrected and anomaly-marked preset position images into the benchmark image library; Preset position ledger parameterization: Establish a PTZ parameter system for each PTZ camera based on the relationship between the longitudinal and lateral directions and the zoom, as well as the focal length; add PTZ parameters to the preset position ledger of the front-end equipment and unify the PTZ parameters of different manufacturers into a standard range; where P is the horizontal rotation angle, T is the vertical pitch angle, and Z is the zoom factor. Historical patrol screenshot analysis: Retrieve historical patrol screenshots from the historical database of patrol tasks within a specified time window and determine the offset; when the historical patrol screenshot is determined to be offset, the corresponding PTZ camera jumps to the pre-position and waits for the pre-position self-correction; when the pre-position of the PTZ camera in the current time window is not offset, generate a report result and report it to the superior management system. Pre-position self-correction: Obtain a real-time screenshot of the pre-position of the PTZ camera and determine the offset; when there is an offset in the real-time screenshot, adjust the PTZ parameters through the PTZ control system and perform iterative correction operations until the deviation value is within the preset threshold to achieve pre-position self-correction, and finally generate a report result and report it to the superior management system. The method for determining the offset is as follows: calculate the deviation value between the current screenshot and the reference image. When the deviation value exceeds the set deviation threshold, it is determined that there is an offset in the preset position. The deviation value includes the P-axis deviation, T-axis deviation and Z-axis deviation.

2. The preset position self-correction method based on image offset and gimbal device parameterization according to claim 1, characterized in that: Building a benchmark image library also includes: dynamically updating the preset benchmark images that are added or removed from the library, and combining this with manual analysis to regularly maintain the accuracy of the benchmark image library.

3. The preset position self-correction method based on image offset and gimbal device parameterization according to claim 1, characterized in that: When determining offset: screenshots with PTZ parameter deviation values ​​exceeding the preset range are judged as irrelevant images and are not subject to offset correction.

4. The preset position self-correction method based on image offset and gimbal device parameterization according to claim 1, characterized in that: The steps of the iterative correction operation are as follows: S1, Initialize parameters: Set the maximum number of iterations and calculate the initial P-axis coordinates. , ; S2, Construct the objective function for pre-position self-correction and determine the convergence condition. The expression for the objective function is: Where X represents the target offset, and the sign of X indicates the direction of the deviation; P is the initial P-axis coordinate of the requested pre-position self-correction; the convergence condition is... ,in Let t be the target offset in round t, and ε be a preset reasonable threshold. S3, Direction Detection: Confirm the direction of deviation convergence through a multi-step detection sequence, including positive and negative detection, and calculate the improvement index of each direction to obtain the decision adjustment direction; S4, based on the direction detection results and decision adjustments, adaptively calculate the initial step size by adjusting the direction d. , S5, perform iterative parameter adjustments until convergence.

5. The preset position self-correction method based on image offset and gimbal device parameterization according to claim 4, characterized in that: S3 includes the following steps: S3-1, Calculate the forward detection sequence, forward error, and corresponding positive improvement index; Forward detection sequence: ; in, These are the coordinates of the P-axis after the first offset; Let be the coordinate value of the P-axis after the i-th positive direction offset; offset in the positive direction The target offset from the reference image; The function return value used to analyze the host machine. This is to improve the threshold, i.e., the step size; Positive error: ; in, This represents the error between the function return value after the i-th positive direction offset and the target offset value. Let t be the target offset in round t; Positive improvement indicators: ; Where N is the number of probes, This represents the absolute value of the difference between the two values ​​before and after the first time. S3-2, Calculate the negative detection sequence and the corresponding negative improvement index; Negative detection sequence: ; in, The coordinate value of the P-axis after the i-th negative direction offset; Offset in the negative direction The target offset from the reference image; The function return value used to analyze the host machine. This is to improve the threshold, i.e., the step size; Positive error: ; in, The error between the function return value after the i-th negative direction offset and the target offset value; Negative improvement indicators: ; S3-3, Based on the improvement indicators in each direction, the decision adjustment direction is obtained; the direction decision function is: ； in, To guide the direction, convergence occurs in either the positive or negative direction; The threshold value is a positive real number, representing the minimum acceptable level of efficiency improvement per unit step size. To improve the threshold, it is used to filter out minor improvements.

6. The preset position self-correction method based on image offset and gimbal device parameterization according to claim 4, characterized in that: S4 includes the following steps: S4-1, Based on the direction detection results, calculate the positive sensitivity estimate and the negative sensitivity estimate; Positive sensitivity estimation: ; Negative sensitivity estimation: ; S4-2, Determine the comprehensive sensitivity S based on the decision adjustment direction d: Overall sensitivity: ; S4-3, Calculate the initial step size: ;in, This is the maximum step size in a single move, that is, the maximum value of the offset value P for each movement. This is the minimum step size in a single move, that is, the minimum offset value P for each movement. The target offset; Offset value Analyze the actual pixel deviation value returned by the host in real time.

7. The preset position self-correction method based on image offset and gimbal device parameterization according to claim 4, characterized in that: S5 includes the following steps: S5-1, Parameter Update: Update the preset bit according to the current iteration number k. And calculate the offset by the corresponding function evaluation value. ; S5-2, Convergence Check: After each iteration, check whether the offset meets the threshold condition. If it does, terminate the correction and return. This will be the final preset position; otherwise, continue with the subsequent steps. S5-3, Direction Verification and Adjustment: Calculate the error change trend of the most recent M iterations. When the error trend shows that the error continues to increase, execute S3, and use the current point... Based on this, adjust the step size to Update direction d; recent The error change trend of each iteration is as follows: ； S5-4, Overshoot Detection and Step Size Adjustment: Detects whether the current function evaluation value exceeds the target value, i.e. ; If so, it means the target value has been exceeded, and the step size for the next step is halved. =0.5 And fine-tune the direction = Otherwise, it indicates that the overshoot did not occur, and the next step size is adaptively adjusted based on the error change of the current iteration, i.e.: { ； in , This is the step size adjustment factor; S5-5, Update Iteration Status: Settings Repeat steps S5-1 to S5-5 until the convergence condition is met or the maximum number of iterations is reached. .

8. The preset position self-correction method based on image offset and gimbal device parameterization according to claim 1, characterized in that: The method also includes self-correction task management functionality: It provides a task configuration interface and supports batch distribution of correction tasks to multiple PTZ camera locations; Monitor the correction process log in real time and report the correction results, including equipment malfunction, successful correction, exceeding the maximum number of iterations, or irrelevant image.

9. The preset position self-correction method based on image offset and gimbal device parameterization according to claim 1, characterized in that: The method also includes PTZ control access management: Set the gimbal control priority, with the self-correction function having a lower priority than the inspection task. The self-correction function will be executed after the inspection task is completed. Use the latest screenshot from the inspection task as the real-time screenshot to reduce the time the gimbal is occupied.

10. A preset position self-correction system based on image offset and gimbal device parameterization, employing the preset position self-correction method based on image offset and gimbal device parameterization as described in any one of claims 1 to 9, characterized in that: include: The reference image library module is used to store and manage preset reference images of the PTZ camera; The parameter management module is used to maintain the PTZ parameter system of the PTZ camera; The intelligent analysis host is used to calculate the deviation between the screenshot and the reference image and to determine whether there is an offset. The self-correction control module is used to execute the iterative correction algorithm and adjust the PTZ parameters. The task management module is used to configure and monitor correction tasks, providing a task configuration interface, real-time log monitoring, and correction result reporting functions.

Images