Camera point location offset detection and automatic calibration method and equipment suitable for high-voltage equipment monitoring

Abstract

The invention relates to a camera point location offset detection and automatic calibration method suitable for high-voltage equipment monitoring. The method comprises the following steps: constructing current preset bit reference primitive information; constructing a mapping relation between a unit PT step length and an imaging space in the current scene; constructing a multi-layer surrounding type auxiliary reference primitive information matrix; judging whether the camera deviates or not; searching optimal matching meta-information; and calculating a deviation rectification PT value, and carrying out PTZ camera control and deviation rectification. According to the method, high-precision automatic correction of the preset position of the camera in high-voltage equipment monitoring is realized, key feature points of power equipment can be stably extracted in a complex field environment by fusing local feature matching and a dynamic key point selection mechanism, and the matching accuracy is greatly improved; actual deployment shows that the repeated positioning precision of the preset position can be improved to + / -2 pixels from original + / -15 pixels, the manual rechecking frequency is remarkably reduced, and differential requirements of equipment of different voltage grades on the inspection precision are met.

Description

Technical Field

[0001] This invention relates to the field of power system monitoring technology, and in particular to a method and device for detecting and automatically calibrating camera position offsets suitable for monitoring high-voltage equipment. Background Technology

[0002] High-voltage equipment such as transformers, reactors, and converter transformers are core components of power transmission and transformation projects. Their safe and stable operation is a crucial foundation for ensuring my country's energy and power development and implementing large-scale energy optimization strategies. Traditional monitoring technologies rely on manual inspections or fixed sensors, making it difficult to achieve high-precision monitoring of equipment fault characteristics under complex on-site operating environments. Currently, dedicated cameras are mainly used for non-contact monitoring, enabling real-time capture and analysis of equipment temperature, appearance defects, and abnormal conditions, providing important data for equipment condition assessment.

[0003] However, in actual operation, due to mechanical factors such as wear and tear of the pan-tilt drive components, aging of the synchronous belt, and accumulation of rotation angle errors caused by external vibrations, as well as environmental factors such as structural deformation caused by extreme weather and temperature changes, or human error during maintenance leading to inaccurate camera preset positions, the camera cannot be aligned with key equipment when the preset position is called, and equipment defects cannot be effectively identified, which seriously affects the accurate monitoring of the status of high-voltage equipment. Summary of the Invention

[0004] To address the problem of monitoring tasks failing to function properly due to preset position offset caused by the camera's own mechanical structure or external forces, the primary objective of this invention is to provide a camera position offset detection and automatic calibration method suitable for high-voltage equipment monitoring. This method enables high-precision automatic correction of the camera's preset position in high-voltage equipment monitoring and can stably extract key feature points of power equipment in complex field environments.

[0005] To achieve the above objectives, the present invention adopts the following technical solution: a method for detecting and automatically calibrating camera position offset suitable for monitoring high-voltage equipment, the method comprising the following sequential steps:

[0006] (1) Move the PTZ camera to the specified preset position and construct the reference primitive information of the current preset position;

[0007] (2) Move the PT camera by a unit PT step and use image matching technology to construct the mapping relationship between the unit PT step and the imaging space in the current scene;

[0008] (3) Based on the mapping relationship between the unit PT step size and the imaging space in the current scene, calculate the PT value that needs to be moved, move the PT camera and construct a multi-layer surround auxiliary reference primitive information matrix;

[0009] (4) Using image matching technology, periodically compare the image captured at the specified preset position with the reference primitive information of the preset position to determine whether the camera is offset. If the camera's imaging view is offset, proceed to the next step; otherwise, do not take any action.

[0010] (5) Compare the image captured at the specified preset position with the preset position reference primitive information and the auxiliary reference primitive information matrix to find the best matching primitive information;

[0011] (6) Calculate the correction PT value based on the best matching metadata and the mapping relationship between the unit PT step size and the imaging space in the current scene, and control and correct the PT camera.

[0012] In step (1), the pan-tilt camera running to the designated preset position means that the high-voltage equipment monitoring system platform connects to the pan-tilt camera to collect video information required for high-voltage equipment monitoring. The high-voltage equipment monitoring system platform sends control commands to the pan-tilt camera to adjust the PTZ of the pan-tilt camera to the preset position. The PTZ includes the pan value of horizontal rotation, the value of vertical rotation, and the zoom value of the lens. The construction of the current preset position reference image information means that the high-voltage equipment monitoring system platform sends commands to obtain the current PTZ information of the pan-tilt camera and the current viewpoint snapshot image, stores the pan-tilt camera number, PTZ and preset position number data, and snapshot image to form a set of reference image information.

[0013] Step (2) specifically refers to: the moving unit PT step size refers to the preset step size specified by sending control commands to the PT camera through the high-voltage equipment monitoring system platform; the image matching technology refers to taking two images before and after inputting the moving unit PT step size, extracting features from the two images respectively, inputting the extracted features into the matching network for matching, selecting the final output feature matching information of the two images, and providing the two images as a reference. and Extract key points and their descriptors respectively:

[0014] image descriptor for: ;

[0015] image descriptor for: ;

[0016] in, For image The coordinates of the i-th key point in the middle. For image The descriptor of the i-th key point in the image, where N and M are the image's... , The number of key points, where D is the descriptor dimension; For image The coordinates of the j-th key point in the middle, For image The descriptor of the j-th key point;

[0017] Image Key point coordinates and descriptors The features are fused into initial features through feature embedding. :

[0018] ;

[0019] in, For sine and cosine position coding functions, This is a vector concatenation operation. For linear projection layers, C is the feature dimension;

[0020] Similarly, we obtain the image. initial features ;

[0021] The Transformer model includes self-attention components, cross-attention components, and a lightweight feedforward network. The input model is iterated, and the Transformer model iterates through L layers, where the first... Feature vectors obtained from the layer for:

[0022] ;

[0023] in, , For layer normalization, For a dimension reduction matrix, It is an upgraded matrix; All are bias terms, and GELU is the Gaussian error linear unit activation function; ;

[0024] In the final layer L of the Transformer model iteration, the similarity matrix between the two image features is calculated. :

[0025] ;

[0026] in, For vector dot product, It is the L2 norm. The cosine similarity matrix for all keypoint pairs. The image output by layer L Key point feature vectors The image output by layer L Key point feature vectors;

[0027] Perform dual-channel Softmax matching to calculate the image. The i-th key point and the image Softmax weights of all points :

[0028] ;

[0029] Similarly, compute images The j-th key point and the image Softmax weights of all points :

[0030] ;

[0031] in, For temperature parameters;

[0032] Define symmetric confidence :

[0033] ;

[0034] Finally, a matching selection is performed. Two image points can be predicted as matching points when they meet the following three conditions:

[0035] Condition 1: j is an image The nearest neighbor of i is the most similar point;

[0036] Condition 2: i is an image The nearest neighbor of j is the point that is most similar to j;

[0037] Condition 3: The symmetric confidence level is higher than the set confidence threshold;

[0038] A matching set is composed of multiple predicted matching points. for:

[0039] ;

[0040] in, The confidence threshold. For image Key point i relative to the image The similarity matrix; This means that when j takes the value k, It is the maximum value; For image Key point j relative to the image The similarity matrix; This means that when i takes the value k, It is the maximum value;

[0041] The mapping relationship between the unit PT step size and the imaging space in the current scene is constructed as follows: Matching points corresponding to the preceding and following imaging images are obtained through image matching technology; pixel offsets are calculated; and the relationship between the pixel offsets of the matching points in the preceding and following imaging images and the unit PT step size is analyzed.

[0042] ;

[0043] in, The difference in the horizontal rotation angle of the pan-tilt camera is represented by a unit P step. The difference in horizontal movement of pixel positions in the image before and after a unit step P is used to match the position of the pixels. This represents the corresponding mapping relationship for the horizontal rotation direction;

[0044] ;

[0045] in, The difference in vertical rotation angle of the pan-tilt camera is expressed in units of T steps. The vertical movement difference of pixel positions in the image before and after a unit step T is used to match the position of the pixels. That is, the corresponding mapping relationship of the vertical rotation direction;

[0046] , This establishes the mapping relationship between the unit PT step size and the imaging space in the current scene.

[0047] The specific step (3) refers to: calculating the PT value that each auxiliary image needs to move according to the mapping relationship between the unit PT step size and the imaging space in the current scene, and constructing a multi-layer surrounding auxiliary reference primitive information matrix, wherein the multi-layer is two layers.

[0048] The step (4) specifically refers to: conducting a test once a week during the normal operation of high-voltage equipment monitoring, comparing the image captured by the designated preset position during the normal monitoring process with the reference primitive information of the preset position of the PTZ camera, and determining whether the pixel-level Euclidean distance between the matching points of the two images is greater than the set pixel offset threshold. If the determination result is yes, it is determined to be a preset position offset.

[0049] Step (5) specifically refers to: based on the pixel-level deviation obtained by comparing the normal monitoring image with the baseline primitive information, and based on the mapping relationship between the unit PT step size and the imaging space in the current scene, deriving the corresponding PT adjustment. and , The difference in vertical rotation angle of the pan-tilt camera is expressed in units of T steps. The difference in the horizontal rotation angle of the pan-tilt camera is represented by the unit step size P. The four reference primitives with the highest probability are found in the multi-layer surrounding auxiliary reference primitive information matrix, and image matching technology is used to match them to obtain the reference primitive information with the smallest pixel-level deviation as the best matching primitive information.

[0050] The step (6) specifically refers to: calculating the difference between the current camera PTZ information and the PTZ in the best matching meta information, obtaining the correction PT value, and then resetting the preset position information after correcting the camera pan-tilt unit using the correction PT value to complete the overall correction action.

[0051] Another object of the present invention is to provide an electronic device comprising:

[0052] Processor; and

[0053] A memory storing computer program instructions that, when executed by the processor, cause the processor to perform the camera position offset detection and automatic calibration method described above for monitoring high-voltage equipment.

[0054] The present invention also provides a computer-readable storage medium having stored thereon computer program instructions, which, when executed by a processor, cause the processor to perform the camera position offset detection and automatic calibration method applicable to high-voltage equipment monitoring as described above.

[0055] As can be seen from the above technical solution, the beneficial effects of the present invention are as follows: First, the present invention achieves high-precision automatic correction of the preset position of the camera in high-voltage equipment monitoring. By integrating local feature matching and dynamic key point selection mechanisms, the present invention can stably extract key feature points of power equipment in complex field environments, and the matching accuracy is greatly improved. Second, through bidirectional consistency verification, mismatches are effectively filtered, and the preset position coordinate correction error is controlled within ±3 pixels, meeting the requirement of Q / GDW 12073-2020 standard for image overlap of substation equipment ≥90%. Third, a closed-loop control pan-tilt pose compensation system is formed. By establishing a nonlinear mapping relationship between PTZ control parameters and pixel offset, accurate compensation of the servo motor is achieved. Actual deployment shows that the present invention can improve the preset position repeatability accuracy from the original ±15 pixels to ±2 pixels, significantly reducing the frequency of manual verification and adapting to the differentiated requirements of inspection accuracy for equipment of different voltage levels. Attached Figure Description

[0056] Figure 1 This is a flowchart of the method of the present invention;

[0057] Figure 2This is a schematic diagram illustrating the construction perspective of the baseline map and multi-level auxiliary baseline maps;

[0058] Figure 3 This is a flowchart illustrating the image matching process. Detailed Implementation

[0059] like Figure 1 , Figure 3 As shown, a method for detecting and automatically calibrating camera position offset suitable for monitoring high-voltage equipment includes the following steps in sequence:

[0060] (1) Move the PTZ camera to the specified preset position and construct the reference primitive information of the current preset position;

[0061] (2) Move the PT camera by a unit PT step and use image matching technology to construct the mapping relationship between the unit PT step and the imaging space in the current scene;

[0062] (3) Based on the mapping relationship between the unit PT step size and the imaging space in the current scene, calculate the PT value that needs to be moved, move the PT camera and construct a multi-layer surround auxiliary reference primitive information matrix;

[0063] (4) Using image matching technology, periodically compare the image captured at the specified preset position with the reference primitive information of the preset position to determine whether the camera is offset. If the camera's imaging view is offset, proceed to the next step; otherwise, do not take any action.

[0064] (5) Compare the image captured at the specified preset position with the preset position reference primitive information and the auxiliary reference primitive information matrix to find the best matching primitive information;

[0065] (6) Calculate the correction PT value based on the best matching metadata and the mapping relationship between the unit PT step size and the imaging space in the current scene, and control and correct the PT camera.

[0066] In step (1), the pan-tilt camera running to the designated preset position means that the high-voltage equipment monitoring system platform connects to the pan-tilt camera to collect video information required for high-voltage equipment monitoring. The high-voltage equipment monitoring system platform sends control commands to the pan-tilt camera to adjust the PTZ of the pan-tilt camera to the preset position. The PTZ includes the pan value of horizontal rotation, the value of vertical rotation, and the zoom value of the lens. The construction of the current preset position reference image information means that the high-voltage equipment monitoring system platform sends commands to obtain the current PTZ information of the pan-tilt camera and the current viewpoint snapshot image, stores the pan-tilt camera number, PTZ and preset position number data, and snapshot image to form a set of reference image information.

[0067] Step (2) specifically refers to: the moving unit PT step size refers to the preset step size specified by sending control commands to the PT camera through the high-voltage equipment monitoring system platform; the image matching technology refers to taking two images before and after inputting the moving unit PT step size, extracting features from the two images respectively, inputting the extracted features into the matching network for matching, selecting the final output feature matching information of the two images, and providing the two images as a reference. and Extract key points and their descriptors respectively:

[0068] image descriptor for: ;

[0069] image descriptor for: ;

[0070] in, For image The coordinates of the i-th key point in the middle. For image The descriptor of the i-th key point in the image, where N and M are the image's... , The number of key points, where D is the descriptor dimension; For image The coordinates of the j-th key point in the middle, For image The descriptor of the j-th key point;

[0071] Image Key point coordinates and descriptors The features are fused into initial features through feature embedding. :

[0072] ;

[0073] in, For sine and cosine position coding functions, This is a vector concatenation operation. For linear projection layers, C is the feature dimension;

[0074] Similarly, we obtain the image. initial features ;

[0075] The Transformer model includes self-attention components, cross-attention components, and a lightweight feedforward network. The input model is iterated, and the Transformer model iterates through L layers, where the first... Feature vectors obtained from the layer for:

[0076] ;

[0077] in, , For layer normalization, For a dimension reduction matrix, It is an upgraded matrix; All are bias terms, and GELU is the Gaussian error linear unit activation function; ;

[0078] In the final layer L of the Transformer model iteration, the similarity matrix between the two image features is calculated. :

[0079] ;

[0080] in, For vector dot product, It is the L2 norm. The cosine similarity matrix for all keypoint pairs. The image output by layer L Key point feature vectors The image output by layer L Key point feature vectors;

[0081] Perform dual-channel Softmax matching to calculate the image. The i-th key point and the image Softmax weights of all points :

[0082] ;

[0083] Similarly, compute images The j-th key point and the image Softmax weights of all points :

[0084] ;

[0085] in, For temperature parameters;

[0086] Define symmetric confidence :

[0087] ;

[0088] Finally, a matching selection is performed. Two image points can be predicted as matching points when they meet the following three conditions:

[0089] Condition 1: j is an image The nearest neighbor of i is the most similar point;

[0090] Condition 2: i is an image The nearest neighbor of j is the point that is most similar to j;

[0091] Condition 3: The symmetric confidence level is higher than the set confidence threshold;

[0092] A matching set is composed of multiple predicted matching points. for:

[0093] ;

[0094] in, The confidence threshold. For image Key point i relative to the image The similarity matrix; This means that when j takes the value k, It is the maximum value; For image Key point j relative to the image The similarity matrix; This means that when i takes the value k, It is the maximum value;

[0095] The mapping relationship between the unit PT step size and the imaging space in the current scene is constructed as follows: Matching points corresponding to the preceding and following imaging images are obtained through image matching technology; pixel offsets are calculated; and the relationship between the pixel offsets of the matching points in the preceding and following imaging images and the unit PT step size is analyzed.

[0096] ;

[0097] in, The difference in the horizontal rotation angle of the pan-tilt camera is represented by a unit P step. The difference in horizontal movement of pixel positions in the image before and after a unit step P is used to match the position of the pixels. This represents the corresponding mapping relationship for the horizontal rotation direction;

[0098] ;

[0099] in, The difference in vertical rotation angle of the pan-tilt camera is expressed in units of T steps. The vertical movement difference of pixel positions in the image before and after a unit step T is used to match the position of the pixels. That is, the corresponding mapping relationship of the vertical rotation direction;

[0100] , This establishes the mapping relationship between the unit PT step size and the imaging space in the current scene.

[0101] Step (3) specifically refers to: calculating the PT value that each auxiliary image needs to move based on the mapping relationship between the unit PT step size and the imaging space in the current scene, and constructing a multi-layer surrounding auxiliary reference primitive information matrix, wherein the multi-layer consists of two layers, such as... Figure 2 As shown.

[0102] The step (4) specifically refers to: conducting a test once a week during the normal operation of high-voltage equipment monitoring, comparing the image captured by the designated preset position during the normal monitoring process with the reference primitive information of the preset position of the PTZ camera, and determining whether the pixel-level Euclidean distance between the matching points of the two images is greater than the set pixel offset threshold. If the determination result is yes, it is determined to be a preset position offset.

[0103] Step (5) specifically refers to: based on the pixel-level deviation obtained by comparing the normal monitoring image with the baseline primitive information, and based on the mapping relationship between the unit PT step size and the imaging space in the current scene, deriving the corresponding PT adjustment. and , The difference in vertical rotation angle of the pan-tilt camera is expressed in units of T steps. The difference in the horizontal rotation angle of the pan-tilt camera is represented by the unit step size P. The four reference primitives with the highest probability are found in the multi-layer surrounding auxiliary reference primitive information matrix, and image matching technology is used to match them to obtain the reference primitive information with the smallest pixel-level deviation as the best matching primitive information.

[0104] The step (6) specifically refers to: calculating the difference between the current camera PTZ information and the PTZ in the best matching meta information, obtaining the correction PT value, and then resetting the preset position information after correcting the camera pan-tilt unit using the correction PT value to complete the overall correction action.

[0105] In summary, this invention achieves high-precision automatic correction of camera preset positions in high-voltage equipment monitoring. By integrating local feature matching and dynamic key point selection mechanisms, this invention can stably extract key feature points of power equipment in complex field environments, significantly improving matching accuracy. Through bidirectional consistency verification, it effectively filters out mismatches, keeping the preset position coordinate correction error within ±3 pixels, meeting the Q / GDW 12073-2020 standard requirement for ≥90% image overlap for substation equipment. A closed-loop control pan-tilt-zoom (PTZ) pose compensation system is formed. By establishing a nonlinear mapping relationship between PTZ control parameters and pixel offset, precise compensation of the servo motor is achieved. Actual deployment shows that this invention can improve the preset position repeatability accuracy from ±15 pixels to ±2 pixels, significantly reducing the frequency of manual verification and adapting to the differentiated inspection accuracy requirements of equipment at different voltage levels. This invention has been piloted in three 500kV substations in China, reducing manual verification workload by an average of 75%, providing reliable technical support for high-voltage equipment monitoring.

[0106] The foregoing has shown and described the basic principles, main features, and advantages of the present invention. Those skilled in the art should understand that the present invention is not limited to the above embodiments. The embodiments and descriptions in the specification are merely principles of the invention. Various changes and modifications can be made to the invention without departing from its spirit and scope, and all such changes and modifications fall within the scope of the claimed invention. The scope of protection claimed by the appended claims and their equivalents is defined.

Claims

1. A method for detecting and automatically calibrating camera position offset suitable for monitoring high-voltage equipment, characterized in that: The method includes the following steps in sequence: (1) Move the PTZ camera to the specified preset position and construct the reference primitive information of the current preset position; (2) Move the PT camera by a unit PT step and use image matching technology to construct the mapping relationship between the unit PT step and the imaging space in the current scene; (3) Based on the mapping relationship between the unit PT step size and the imaging space in the current scene, calculate the PT value that needs to be moved, move the PT camera and construct a multi-layer surrounding auxiliary reference primitive information matrix; (4) Using image matching technology, periodically compare the image captured at the specified preset position with the reference primitive information of the preset position to determine whether the camera is offset. If the camera's imaging view is offset, proceed to the next step; otherwise, do not take any action. (5) Compare the image captured at the specified preset position with the preset position reference primitive information and the auxiliary reference primitive information matrix to find the best matching primitive information; (6) Calculate the correction PT value based on the best matching metadata and the mapping relationship between the unit PT step size and the imaging space in the current scene, and control and correct the PT camera.

2. The method for camera position offset detection and automatic calibration applicable to high-voltage equipment monitoring according to claim 1, characterized in that: In step (1), the pan-tilt camera running to the designated preset position means that the high-voltage equipment monitoring system platform connects to the pan-tilt camera to collect video information required for high-voltage equipment monitoring. The high-voltage equipment monitoring system platform sends control commands to the pan-tilt camera to adjust the PTZ of the pan-tilt camera to the preset position. The PTZ includes the pan value of horizontal rotation, the value of vertical rotation, and the zoom value of the lens. The construction of the current preset position reference image information means that the high-voltage equipment monitoring system platform sends commands to obtain the current PTZ information of the pan-tilt camera and the current viewpoint snapshot image, stores the pan-tilt camera number, PTZ and preset position number data, and snapshot image to form a set of reference image information.

3. The method for camera position offset detection and automatic calibration applicable to high-voltage equipment monitoring according to claim 1, characterized in that: Step (2) specifically refers to: the moving unit PT step size refers to the preset step size specified by sending control commands to the PT camera through the high-voltage equipment monitoring system platform; the image matching technology refers to taking two images before and after inputting the moving unit PT step size, extracting features from the two images respectively, inputting the extracted features into the matching network for matching, selecting the final output feature matching information of the two images, and providing the two images as a reference. and Extract key points and their descriptors respectively: image descriptor for: ; image descriptor for: ; in, For image The coordinates of the i-th key point in the middle. For image The descriptor of the i-th key point in the image, where N and M are the image's... , The number of key points, where D is the descriptor dimension; For image The coordinates of the j-th key point in the middle, For image The descriptor of the j-th key point; Image Key point coordinates and descriptors The features are fused into initial features through feature embedding. : ； in, For sine and cosine position coding functions, This is a vector concatenation operation. For linear projection layers, C is the feature dimension; Similarly, we obtain the image. initial features ; The Transformer model includes self-attention components, cross-attention components, and a lightweight feedforward network. The input model is iterated, and the Transformer model iterates through L layers, where the first... Feature vectors obtained from the layer for: ； in, , For layer normalization, For a dimension reduction matrix, It is an upgraded matrix; All are bias terms, and GELU is the Gaussian error linear unit activation function; ; In the final layer L of the Transformer model iteration, the similarity matrix between the two image features is calculated. : ； in, For vector dot product, It is the L2 norm. The cosine similarity matrix for all keypoint pairs. The image output by layer L Key point feature vectors The image output by layer L Key point feature vectors; Perform dual-channel Softmax matching to calculate the image. The i-th key point and the image Softmax weights of all points : ； Similarly, compute images The j-th key point and the image Softmax weights of all points : ； in, For temperature parameters; Define symmetric confidence : ； Finally, a matching selection is performed. Two image points can be predicted as matching points when they meet the following three conditions: Condition 1: j is an image The nearest neighbor of i is the most similar point; Condition 2: i is an image The nearest neighbor of j is the point that is most similar to j; Condition 3: The symmetric confidence level is higher than the set confidence threshold; A matching set is composed of multiple predicted matching points. for: ； in, The confidence threshold. For image Key point i relative to the image The similarity matrix; This means that when j takes the value k, It is the maximum value; For image Key point j relative to the image The similarity matrix; This means that when i takes the value k, It is the maximum value; The mapping relationship between the unit PT step size and the imaging space in the current scene is constructed as follows: Matching points corresponding to the preceding and following imaging images are obtained through image matching technology; pixel offsets are calculated; and the relationship between the pixel offsets of the matching points in the preceding and following imaging images and the unit PT step size is analyzed. ； in, The difference in the horizontal rotation angle of the pan-tilt camera is represented by a unit P step. The difference in horizontal movement of pixel positions in the image before and after a unit step P is used to match the position of the pixels. This represents the corresponding mapping relationship for the horizontal rotation direction; ； in, The difference in vertical rotation angle of the pan-tilt camera is expressed in units of T steps. The vertical movement difference of pixel positions in the image before and after a unit step T is used to match the position of the pixels. That is, the corresponding mapping relationship of the vertical rotation direction; , This establishes the mapping relationship between the unit PT step size and the imaging space in the current scene.

4. The method for camera position offset detection and automatic calibration applicable to high-voltage equipment monitoring according to claim 1, characterized in that: The specific step (3) refers to: calculating the PT value that each auxiliary image needs to move according to the mapping relationship between the unit PT step size and the imaging space in the current scene, and constructing a multi-layer surrounding auxiliary reference primitive information matrix, wherein the multi-layer is two layers.

5. The method for camera position offset detection and automatic calibration applicable to high-voltage equipment monitoring according to claim 1, characterized in that: The step (4) specifically refers to: conducting a test once a week during the normal operation of high-voltage equipment monitoring, comparing the image captured by the designated preset position during the normal monitoring process with the reference primitive information of the preset position of the PTZ camera, and determining whether the pixel-level Euclidean distance between the matching points of the two images is greater than the set pixel offset threshold. If the determination result is yes, it is determined to be a preset position offset.

6. The method for camera position offset detection and automatic calibration applicable to high-voltage equipment monitoring according to claim 1, characterized in that: Step (5) specifically refers to: based on the pixel-level deviation obtained by comparing the normal monitoring image with the baseline primitive information, and based on the mapping relationship between the unit PT step size and the imaging space in the current scene, deriving the corresponding PT adjustment. and , The difference in vertical rotation angle of the pan-tilt camera is expressed in units of T steps. The difference in the horizontal rotation angle of the pan-tilt camera is represented by the unit step size P. The four reference primitives with the highest probability are found in the multi-layer surrounding auxiliary reference primitive information matrix, and image matching technology is used to match them to obtain the reference primitive information with the smallest pixel-level deviation as the best matching primitive information.

7. The method for camera position offset detection and automatic calibration applicable to high-voltage equipment monitoring according to claim 1, characterized in that: The step (6) specifically refers to: calculating the difference between the current camera PTZ information and the PTZ in the best matching meta information, obtaining the correction PT value, and then resetting the preset position information after correcting the camera pan-tilt unit using the correction PT value to complete the overall correction action.

8. An electronic device, comprising: processor; as well as A memory storing computer program instructions that, when executed by the processor, cause the processor to perform the camera position offset detection and automatic calibration method for monitoring high-voltage equipment as described in any one of claims 1-7.

9. A computer-readable storage medium having stored thereon computer program instructions, which, when executed by a processor, cause the processor to perform a camera position offset detection and automatic calibration method for monitoring high-voltage equipment as described in any one of claims 1-7.

Images