Camera focusing system and method based on multi-sensor data fusion
By using a multi-sensor data fusion system, combined with a gyroscope and a TOF laser sensor, the problem of focusing accuracy and imaging offset of the camera under complex postures is solved, achieving high-precision and stable image capture.
Patent Information
- Application Number
- CN202511516909.7
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-10-23
- Publication Date
- 2026-02-10
AI Technical Summary
When existing cameras are bent, tilted, or rotated, traditional optical focusing suffers from delays, image shifts, and distance measurement errors, resulting in decreased focusing accuracy.
A multi-sensor data fusion system is adopted, combining a gyroscope sensor and a TOF laser sensor. Through data preprocessing, fusion and control modules, the system enables real-time attitude perception and distance measurement of the camera, and dynamic adjustment of focal length and image correction.
Improving focusing accuracy under complex postures ensures image sharpness and uprightness, thereby enhancing system stability and reliability.
Smart Images
Figure CN121509811A_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of camera imaging and intelligent control technology, specifically to a camera focusing system and method based on multi-sensor data fusion. Background Technology
[0002] Existing camera focusing technologies mostly rely on a single sensor. During shooting, when the camera bends, tilts, or rotates, traditional optical focusing needs to adjust the focal length through image contrast analysis, causing the target imaging position to shift and resulting in focusing delay. When the camera is placed horizontally or shooting from above, the image is prone to being inverted due to the tilt, resulting in imaging shift. When a single TOF laser sensor is tilted, the straight-line distance measured may differ from the actual imaging plane distance, leading to decreased focusing accuracy and distance measurement deviation. Summary of the Invention
[0003] The purpose of this invention is to provide a camera focusing system and method based on multi-sensor data fusion to solve the problems mentioned in the background art.
[0004] To solve the above-mentioned technical problems, the present invention provides the following technical solution: a camera focusing system based on multi-sensor data fusion, the system comprising a data acquisition module, a data processing module, a data fusion module, and a focusing control module; The data acquisition module consists of a gyroscope sensor and a TOF laser sensor. The gyroscope sensor is used to acquire real-time attitude data of the camera, and the TOF laser sensor is used to acquire straight-line distance data between the camera and the target. The real-time attitude data includes tilt angle, rotation direction, and rotation angular velocity.
[0005] The data processing module is connected to the data acquisition module. The data processing module is used to preprocess the attitude data and straight-line distance data. The data preprocessing includes outlier filtering, data standardization, and time sequence alignment. The outlier filtering includes: Outlier filtering of gyroscope sensor attitude data involves outlier removal and linear interpolation using mean and standard deviation formulas. Mean formula: ; Standard deviation formula: ; in, The mean; Standard deviation; For the first There are n pose data points, where n is the number of pose data points. When a certain frame of data satisfies Values exceeding the attitude data outlier detection threshold are considered attitude data outliers; this threshold is set by the system and is the default value. After removing the outliers in the attitude data, the outliers are replaced by the linear interpolation results of the two frames before and after the outlier to ensure the smoothness of the attitude data; the linear interpolation results are the average of the attitude data of the two frames before and after the outlier. The TOF laser sensor employs a dual verification mechanism for filtering straight-line distance data; this dual verification mechanism includes threshold verification and continuous verification. The threshold verification includes: When the difference between a single line-of-sight distance data and the previous frame data exceeds the line-of-sight distance threshold, it is marked as suspicious data; the line-of-sight distance threshold is set by the system. The continuous verification includes: When the number of consecutive occurrences of suspicious data exceeds the consecutive verification threshold, it is determined to be an abnormal value of the straight-line distance data. The consecutive verification threshold is set by the system. For abnormal values of straight-line distance data, the average straight-line distance of the first two frames in the cache is used as the replacement.
[0006] The data standardization includes: The units of attitude data and straight-line distance data are unified, and the tilt angle unit output by the gyroscope sensor is converted from degrees to radians according to the radian angle conversion formula; Radius conversion formula: 1 degree = π / 180 rad; The unit for rotational angular velocity is uniformly rad / s; The unit of linear distance data output by the TOF sensor is standardized to meters; the dimensional differences in the raw data from different sensors are eliminated. The standardized dataset is then normalized according to the actual working scenario of the camera; the normalization process includes: The system is configured with a range of tilt angle, a range of rotational angular velocity, and a range of straight-line distance. The tilt angle range is [-π / 2, π / 2], covering the bending range of the camera; The rotational angular velocity range is [-2π, 2π] rad / s, which is suitable for the attitude change rate of daily shooting; The straight-line distance range is [0.3, 10] meters. Data exceeding this range will be truncated to the boundary value to avoid interference from extreme values on the fusion algorithm.
[0007] The timing alignment includes: The data is timestamped to record the absolute time of data acquisition; the data includes gyroscope sensor attitude data and TOF laser sensor straight-line distance data for each frame. Sampling frequency adaptation: Since there is a difference between the sampling frequency of the gyroscope sensor (50Hz) and the sampling frequency of the TOF laser sensor (100Hz), an interpolation frame-filling strategy is used to achieve timing alignment. The interpolation and frame-filling strategy includes: using the timestamp of the gyroscope sensor attitude data as a reference, downsampling the straight-line distance data of the TOF laser sensor, and using the average of two frames to replace one frame; Ultimately, the time interval between the two data sets was unified to 50Hz, ensuring that corresponding gyroscope sensor attitude data and TOF laser sensor linear distance data exist at the same timestamp. This achieves precise matching in the spatiotemporal dimensions.
[0008] The data fusion module is connected to the data processing module. The data fusion module uses a weighted fusion algorithm to fuse the pre-processed gyroscope sensor attitude data and TOF laser sensor straight-line distance data to generate fused data. The fused data includes the target's projection distance on the camera imaging plane and attitude feature parameters. The weighted fusion algorithm includes: The fusion weights of the gyroscope sensor attitude data and the TOF laser sensor straight-line distance data are determined by a dynamic weighting strategy based on the angular velocity of the gyroscope sensor. The dynamic weighting strategy includes: When the absolute value of the rotational angular velocity of the gyroscope sensor attitude data is less than the stability threshold of the gyroscope sensor attitude data, it is marked as stable. The TOF laser sensor linear distance data has higher stability. At this time, the weight of the TOF laser sensor linear distance data is assigned to 0.7, and the weight of the gyroscope attitude data is assigned to 0.3, with the TOF laser sensor linear distance data as the main reference. When the absolute value of the rotational angular velocity of the gyroscope sensor attitude data exceeds the gyroscope sensor attitude data stability threshold, it is marked as a change in the gyroscope sensor attitude data. The gyroscope sensor responds to attitude changes more in real time. At this time, the weight of the gyroscope attitude data is increased to 0.6, and the weight of the TOF laser sensor straight-line distance data is reduced to 0.4. The attitude data compensates for the latency of the TOF data, ensuring the dynamic adaptability of the fusion result.
[0009] The stability threshold for the gyroscope sensor attitude data is set by the system; the default value is 0.3 rad / s.
[0010] The projection distance of the target onto the camera's imaging plane includes: A projection distance formula is defined by combining dynamic weighting coefficients and geometric imaging principles. The projection distance of the target on the camera's imaging plane is then calculated using this formula. The projection distance formula is: ; in The distance of the target projected onto the camera's imaging plane; d represents the weight of the straight-line distance data of the TOF laser sensor; d represents the straight-line distance data of the TOF laser sensor. Weights for gyroscope attitude data; The tilt angle for gyroscope attitude data; This is the correction factor.
[0011] The attitude feature parameters include: The rotation direction is determined by the sign of the rotational angular velocity of the gyroscope attitude data; when the rotational angular velocity of the gyroscope attitude data is positive, the rotation direction is clockwise, and when the rotational angular velocity of the gyroscope attitude data is negative, the rotation direction is counterclockwise. Using the initial rotation angle as a reference, defined as zero degrees; the rotation angular velocity of the gyroscope attitude data is used to obtain the camera's rotation angle around the optical axis through a time integration formula; the time integration formula is: ; in The rotation angle; The rotational angular velocity of the gyroscope's attitude data; The focusing control module is connected to the data fusion module. The focusing control module drives the camera lens to adjust the focal length according to the projection distance of the target on the camera imaging plane. The focusing control module triggers the switching between landscape and portrait modes and image orientation correction according to the posture feature parameters.
[0012] The camera lens is adjusted by a drive motor built into the lens, which moves the lens group closer to or further away along the optical axis. When the projection distance of the target on the camera's imaging plane increases, the drive motor drives the lens group to move away along the optical axis, increasing the focal length; when the projection distance of the target on the camera's imaging plane decreases, the drive motor drives the lens group to move closer along the optical axis, decreasing the focal length. The landscape / portrait switching is determined by a landscape / portrait switching trigger condition; the landscape / portrait switching trigger condition is: If the rotation angle of the posture feature parameter is within the horizontal / vertical screen switching trigger threshold, the horizontal / vertical screen switching trigger condition is determined to be met; if the rotation angle of the posture feature parameter is outside the horizontal / vertical screen switching trigger threshold, the horizontal / vertical screen switching trigger condition is determined not to be met. The screen orientation switching trigger threshold is set to [85°, 95°] based on the attitude feature parameters output by the data fusion module; The image orientation correction compensates for image skew caused by camera tilt through inverse compensation; the orientation compensation includes correction angle calculation and image geometric correction execution. The correction angle calculation is based on the tilt angle to calculate data that is equal in value but opposite in direction to the tilt angle. The image geometric correction process involves performing a geometric rotation on the original image; the geometric rotation process involves rotating the image in the opposite direction according to the calculated correction angle to eliminate image tilt.
[0013] A camera focusing method based on multi-sensor data fusion is characterized by the following steps: S001, the data acquisition module consists of a gyroscope sensor and a TOF laser sensor. The gyroscope sensor is used to acquire real-time attitude data of the camera, and the TOF laser sensor is used to acquire straight-line distance data between the camera and the target being photographed. The real-time attitude data includes tilt angle, rotation direction, and rotation angular velocity. S002, The data processing module is connected to the data acquisition module. The data processing module is used to preprocess the attitude data and straight-line distance data. The data preprocessing includes outlier filtering, data standardization, and time sequence alignment. S003. The data fusion module is connected to the data processing module. The data fusion module uses a weighted fusion algorithm to fuse the pre-processed gyroscope sensor attitude data and TOF laser sensor straight-line distance data to generate fused data. The weighted fusion algorithm determines the fusion weights of the gyroscope sensor attitude data and TOF laser sensor straight-line distance data based on a dynamic weighting strategy of the gyroscope sensor's angular velocity. The fused data includes the target's projection distance on the camera's imaging plane and attitude feature parameters. S004. The focus control module is connected to the data fusion module. The focus control module drives the camera lens to adjust the focal length according to the projection distance of the target on the camera imaging plane. The focal length adjustment is achieved by the built-in drive motor of the lens driving the lens group to move closer or further away along the optical axis. The focus control module triggers the horizontal / vertical screen switching and image orientation correction according to the posture characteristic parameters. The horizontal / vertical screen switching determines the vertical or horizontal screen mode through the horizontal / vertical screen switching trigger conditions. The image orientation correction compensates for the image distortion caused by the camera tilt through reverse compensation.
[0014] Compared with the prior art, the beneficial effects achieved by the present invention are: Multi-sensor fusion improves focusing accuracy. When the camera bends, tilts, or rotates during shooting, it combines gyroscope attitude perception and TOF laser distance measurement to solve the measurement deviation of a single sensor under complex postures and achieve fast focusing. Clear imaging is achieved in any posture, such as shooting from the front, from above, or from the side, and the output image is always upright to ensure image quality. Multi-module collaborative optimization improves system stability and reliability. Attached Figure Description
[0015] The accompanying drawings are provided to further illustrate the invention and form part of the specification. They are used in conjunction with embodiments of the invention to explain the invention and do not constitute a limitation thereof. In the drawings: Fig. 1 This is a schematic diagram of the camera focusing system based on multi-sensor data fusion according to the present invention; Fig. 2 This is a schematic diagram of the camera focusing method based on multi-sensor data fusion according to the present invention. Detailed Implementation
[0016] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.
[0017] Please see Figs. 1-2 The present invention provides the following technical solution: A camera focusing system based on multi-sensor data fusion includes a data acquisition module, a data processing module, a data fusion module, and a focusing control module. The data acquisition module consists of a gyroscope sensor and a TOF laser sensor. The gyroscope sensor is used to acquire real-time attitude data of the camera, and the TOF laser sensor is used to acquire straight-line distance data between the camera and the target. The real-time attitude data includes tilt angle, rotation direction, and rotation angular velocity.
[0018] The data processing module is connected to the data acquisition module. The data processing module is used to preprocess the attitude data and straight-line distance data. The data preprocessing includes outlier filtering, data standardization, and time sequence alignment. The outlier filtering includes: Outlier filtering of gyroscope sensor attitude data involves outlier removal and linear interpolation using mean and standard deviation formulas. Mean formula: ; Standard deviation formula: ; in, The mean; Standard deviation; For the first There are n pose data points, where n is the number of pose data points. When a certain frame of data satisfies Values exceeding the attitude data outlier detection threshold are considered attitude data outliers; this threshold is set by the system and is the default value. After removing the outliers in the attitude data, the outliers are replaced by the linear interpolation results of the two frames before and after the outlier to ensure the smoothness of the attitude data; the linear interpolation results are the average of the attitude data of the two frames before and after the outlier. In this embodiment, the gyroscope attitude data shows a rotational angular velocity of 0.2 rad / s; the rotation direction is clockwise; and the tilt angle data are 30°, 30.1°, 30°, 29.9°, and 30°. Calculate the mean tilt angle: ; Calculate the standard deviation of the tilt angle: ; Outlier threshold: All data meet the requirements. ≤0.19, no outliers.
[0019] The TOF laser sensor employs a dual verification mechanism for filtering straight-line distance data; this dual verification mechanism includes threshold verification and continuous verification. The threshold verification includes: When the difference between a single line-of-sight distance data and the previous frame data exceeds the line-of-sight distance threshold, it is marked as suspicious data; the line-of-sight distance threshold is set by the system. The continuous verification includes: When the number of consecutive occurrences of suspicious data exceeds the consecutive verification threshold, it is determined to be an abnormal value of the straight-line distance data. The consecutive verification threshold is set by the system. For abnormal values of straight-line distance data, the average straight-line distance of the first two frames in the cache is used as the replacement.
[0020] In this embodiment, the straight-line distance threshold is set to 0.5m, and the TOF laser straight-line distance data processing is as follows: Ten frames of straight-line distance data were continuously collected: 3.0m, 3.0m, 3.1m, 3.0m, 3.0m, 3.0m, 3.1m, 3.0m, 3.0m, 3.0m; where m is the unit meter.
[0021] Threshold verification: The difference between all single straight-line distance data and the previous frame data does not exceed the straight-line distance threshold, and there is no suspicious data; Continuous verification: Retain the original data.
[0022] The data standardization includes: The units of attitude data and straight-line distance data are unified, and the tilt angle unit output by the gyroscope sensor is converted from degrees to radians according to the radian angle conversion formula; Radius conversion formula: 1 degree = π / 180 rad; The unit for rotational angular velocity is uniformly rad / s; The unit of linear distance data output by the TOF sensor is standardized to meters; the dimensional differences in the raw data from different sensors are eliminated. The standardized dataset is then normalized according to the actual working scenario of the camera; the normalization process includes: The system is configured with a range of tilt angle, a range of rotational angular velocity, and a range of straight-line distance. The tilt angle range is [-π / 2, π / 2], covering the bending range of the camera; The rotational angular velocity range is [-2π, 2π] rad / s, which is suitable for the attitude change rate of daily shooting; The straight-line distance range is [0.3, 10] meters. Data exceeding this range will be truncated to the boundary value to avoid interference from extreme values on the fusion algorithm.
[0023] In this embodiment, the tilt angle is converted into the following values according to the radian angle conversion formula: 0.523 rad, 0.525 rad, 0.523 rad, 0.521 rad, 0.523 rad; Rotational angular velocity: 0.2 rad / s, which meets the unit requirements; TOF laser linear distance: 3.0m, which meets the unit's requirements; The timing alignment includes: The data is timestamped to record the absolute time of data acquisition; the data includes gyroscope sensor attitude data and TOF laser sensor straight-line distance data for each frame. Sampling frequency adaptation: Since there is a difference between the sampling frequency of the gyroscope sensor (50Hz) and the sampling frequency of the TOF laser sensor (100Hz), an interpolation frame-filling strategy is used to achieve timing alignment. The interpolation and frame-filling strategy includes: using the timestamp of the gyroscope sensor attitude data as a reference, downsampling the straight-line distance data of the TOF laser sensor, and using the average of two frames to replace one frame; Ultimately, the time interval between the two data sets was unified to 50Hz, ensuring that corresponding gyroscope sensor attitude data and TOF laser sensor linear distance data exist at the same timestamp. This achieves precise matching in the spatiotemporal dimensions.
[0024] In this embodiment, the TOF laser straight-line distance data is obtained by continuously collecting 10 frames of straight-line distance data: 3.0m, 3.0m, 3.1m, 3.0m, 3.0m, 3.0m, 3.1m, 3.0m, 3.0m, 3.0m; the average of two frames is used to replace one frame; therefore, the straight-line distance data becomes: 3.0m, 3.05m, 3.0m, 3.05m, 3.0m. The data fusion module is connected to the data processing module. The data fusion module uses a weighted fusion algorithm to fuse the pre-processed gyroscope sensor attitude data and TOF laser sensor straight-line distance data to generate fused data. The fused data includes the target's projection distance on the camera imaging plane and attitude feature parameters. The weighted fusion algorithm includes: The fusion weights of the gyroscope sensor attitude data and the TOF laser sensor straight-line distance data are determined by a dynamic weighting strategy based on the angular velocity of the gyroscope sensor. The dynamic weighting strategy includes: When the absolute value of the rotational angular velocity of the gyroscope sensor attitude data is less than the stability threshold of the gyroscope sensor attitude data, it is marked as stable. The TOF laser sensor linear distance data has higher stability. At this time, the weight of the TOF laser sensor linear distance data is assigned to 0.7, and the weight of the gyroscope attitude data is assigned to 0.3, with the TOF laser sensor linear distance data as the main reference. When the absolute value of the rotational angular velocity of the gyroscope sensor attitude data exceeds the gyroscope sensor attitude data stability threshold, it is marked as a change in the gyroscope sensor attitude data. The gyroscope sensor responds to attitude changes more in real time. At this time, the weight of the gyroscope attitude data is increased to 0.6, and the weight of the TOF laser sensor straight-line distance data is reduced to 0.4. The attitude data compensates for the latency of the TOF data, ensuring the dynamic adaptability of the fusion result.
[0025] The stability threshold for the gyroscope sensor attitude data is set by the system; the default value is 0.3 rad / s.
[0026] In this embodiment, 0.2 is less than 0.3. When the absolute value of the gyroscope rotational angular velocity is less than the gyroscope sensor attitude data stability threshold, it is marked as gyroscope sensor attitude data being stable. The TOF laser sensor linear distance data has higher stability. At this time, the weight of the TOF laser sensor linear distance data is assigned to 0.7, and the weight of the gyroscope attitude data is assigned to 0.3, with the TOF laser sensor linear distance data as the main reference. The projection distance of the target onto the camera's imaging plane includes: A projection distance formula is defined by combining dynamic weighting coefficients and geometric imaging principles. The projection distance of the target on the camera's imaging plane is then calculated using this formula. The projection distance formula is: ; in The distance of the target projected onto the camera's imaging plane; d represents the weight of the straight-line distance data of the TOF laser sensor; d represents the straight-line distance data of the TOF laser sensor. Weights for gyroscope attitude data; The tilt angle for gyroscope attitude data; This is the correction factor.
[0027] In this embodiment, the correction factor is 1; ; The attitude feature parameters include: The rotation direction is determined by the sign of the rotational angular velocity of the gyroscope attitude data; when the rotational angular velocity of the gyroscope attitude data is positive, the rotation direction is clockwise, and when the rotational angular velocity of the gyroscope attitude data is negative, the rotation direction is counterclockwise. Using the initial rotation angle as a reference, defined as zero degrees; the rotation angular velocity of the gyroscope attitude data is used to obtain the camera's rotation angle around the optical axis through a time integration formula; the time integration formula is: ; in The rotation angle; The rotational angular velocity of the gyroscope's attitude data; In this embodiment, the initial rotation angle is 0°, the screen orientation is vertical, and the integration time is 20ms (0.02s); rotation angle: .
[0028] The focusing control module is connected to the data fusion module. The focusing control module drives the camera lens to adjust the focal length according to the projection distance of the target on the camera imaging plane. The focusing control module triggers the switching between landscape and portrait modes and image orientation correction according to the posture feature parameters.
[0029] The camera lens is adjusted by a drive motor built into the lens, which moves the lens group closer to or further away along the optical axis. When the projection distance of the target on the camera's imaging plane increases, the drive motor drives the lens group to move away along the optical axis, increasing the focal length; when the projection distance of the target on the camera's imaging plane decreases, the drive motor drives the lens group to move closer along the optical axis, decreasing the focal length. The landscape / portrait switching is determined by a landscape / portrait switching trigger condition; the landscape / portrait switching trigger condition is: If the rotation angle of the posture feature parameter is within the horizontal / vertical screen switching trigger threshold, the horizontal / vertical screen switching trigger condition is determined to be met; if the rotation angle of the posture feature parameter is outside the horizontal / vertical screen switching trigger threshold, the horizontal / vertical screen switching trigger condition is determined not to be met. The screen orientation switching trigger threshold is set to [85°, 95°] based on the attitude feature parameters output by the data fusion module; In this embodiment, the user rotates the camera to 92°. Landscape / portrait screen switching trigger condition judgment: If the rotation angle of the posture feature parameter is within the landscape / portrait screen switching trigger threshold, then the landscape / portrait screen switching trigger condition is satisfied; 92°∈[85°,95°], the switching condition is satisfied; when switching from portrait to landscape, the focal length parameter remains unchanged and the image is not blurred.
[0030] The image orientation correction compensates for image skew caused by camera tilt through inverse compensation; the orientation compensation includes correction angle calculation and image geometric correction execution. The correction angle calculation is based on the tilt angle to calculate data that is equal in value but opposite in direction to the tilt angle. The image geometric correction process involves performing a geometric rotation on the original image; the geometric rotation process involves rotating the image in the opposite direction according to the calculated correction angle to eliminate image tilt.
[0031] In this embodiment, the user uses a camera to take a picture from an overhead view. Correction angle calculation: Calculate the correction angle that is equal in value but opposite in direction to the tilt angle; the tilt angle is 30°, so the correction angle is -30°. Geometric correction is performed: the original image is rotated by -30° according to the calculated correction angle to eliminate image tilt caused by the tilt angle; Correction effect: The horizontal deviation of the output image is ≤0.5°, which conforms to the human eye's upright visual habits.
[0032] It should be noted that, in this document, relational terms such as "first" and "second" are used only to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such process, method, article, or apparatus.
[0033] Finally, it should be noted that the above descriptions are merely preferred embodiments of the present invention and are not intended to limit the present invention. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art can still modify the technical solutions described in the foregoing embodiments or make equivalent substitutions for some of the technical features. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.
Claims
1. A camera focusing system based on multi-sensor data fusion, characterized in that, include: Data acquisition module, data processing module, data fusion module, and focus control module; The data acquisition module consists of a gyroscope sensor and a TOF laser sensor. The gyroscope sensor is used to acquire real-time attitude data of the camera, and the TOF laser sensor is used to acquire straight-line distance data between the camera and the target being photographed. The data processing module is connected to the data acquisition module. The data processing module is used to preprocess the attitude data and straight-line distance data. The data preprocessing includes outlier filtering, data standardization, and time sequence alignment. The data fusion module is connected to the data processing module. The data fusion module uses a weighted fusion algorithm to fuse the pre-processed gyroscope sensor attitude data and TOF laser sensor straight-line distance data to generate fused data. The fused data includes the target's projection distance on the camera imaging plane and attitude feature parameters. The focusing control module is connected to the data fusion module. The focusing control module drives the camera lens to adjust the focal length according to the projection distance of the target on the camera imaging plane. The focusing control module triggers the switching between landscape and portrait modes and image orientation correction according to the posture feature parameters.
2. The camera focusing system based on multi-sensor data fusion according to claim 1, characterized in that, The real-time attitude data includes: Inclination angle, rotation direction, and rotational angular velocity.
3. The camera focusing system based on multi-sensor data fusion according to claim 1, characterized in that, The outlier filtering includes: Outlier filtering of gyroscope sensor attitude data involves outlier removal and linear interpolation using mean and standard deviation formulas. Mean formula: ; Standard deviation formula: ; in, The mean; Standard deviation; For the first There are n pose data points, where n is the number of pose data points. When a certain frame of data satisfies Values exceeding the attitude data outlier detection threshold are considered attitude data outliers; this threshold is set by the system and is the default value. After removing the outliers in the attitude data, the outliers are replaced by the linear interpolation results of the two frames before and after the outlier to ensure the smoothness of the attitude data; the linear interpolation results are the average of the attitude data of the two frames before and after the outlier. The TOF laser sensor employs a dual verification mechanism for filtering straight-line distance data; this dual verification mechanism includes threshold verification and continuous verification. The threshold verification includes: When the difference between a single line-of-sight distance data and the previous frame data exceeds the line-of-sight distance threshold, it is marked as suspicious data; the line-of-sight distance threshold is set by the system. The continuous verification includes: When the number of consecutive occurrences of suspicious data exceeds the consecutive verification threshold, it is determined to be an abnormal value of the straight-line distance data. The consecutive verification threshold is set by the system. For abnormal values of straight-line distance data, the average straight-line distance of the first two frames in the cache is used as the replacement.
4. The camera focusing system based on multi-sensor data fusion according to claim 1, characterized in that, The data standardization includes: The units of attitude data and straight-line distance data are unified, and the tilt angle unit output by the gyroscope sensor is converted from degrees to radians according to the radian angle conversion formula; Radius conversion formula: 1 degree = π / 180 rad; The unit for rotational angular velocity is uniformly rad / s; The unit for the straight-line distance data output by the TOF sensor is standardized to meters; The standardized dataset is then normalized based on the actual working scenario of the camera; the normalization process includes: The system is configured with a range of tilt angle, a range of rotational angular velocity, and a range of straight-line distance. The tilt angle range is [-π / 2, π / 2], covering the bending range of the camera; The rotational angular velocity range is [-2π, 2π] rad / s, which is suitable for the attitude change rate of daily shooting; The straight-line distance range is [0.3, 10] meters. Data exceeding this range will be truncated to the boundary value to avoid interference from extreme values on the fusion algorithm.
5. The camera focusing system based on multi-sensor data fusion according to claim 1, characterized in that, The timing alignment includes: The data is timestamped to record the absolute time of data acquisition; the data includes gyroscope sensor attitude data and TOF laser sensor straight-line distance data for each frame. Sampling frequency adaptation: Since there is a difference between the sampling frequency of the gyroscope sensor (50Hz) and the sampling frequency of the TOF laser sensor (100Hz), an interpolation frame-filling strategy is used to achieve timing alignment. The interpolation and frame-filling strategy includes: using the timestamp of the gyroscope sensor attitude data as a reference, downsampling the straight-line distance data of the TOF laser sensor, and using the average of two frames to replace one frame; Ultimately, the time interval between the two data sets was unified to 50Hz, ensuring that corresponding gyroscope sensor attitude data and TOF laser sensor linear distance data exist at the same timestamp.
6. The camera focusing system based on multi-sensor data fusion according to claim 1, characterized in that, The weighted fusion algorithm includes: The fusion weights of the gyroscope sensor attitude data and the TOF laser sensor straight-line distance data are determined by a dynamic weighting strategy based on the angular velocity of the gyroscope sensor. The dynamic weighting strategy includes: When the absolute value of the rotational angular velocity of the gyroscope sensor attitude data is less than the stability threshold of the gyroscope sensor attitude data, it is marked as stable. The TOF laser sensor linear distance data has higher stability. At this time, the weight of the TOF laser sensor linear distance data is assigned to 0.7, and the weight of the gyroscope attitude data is assigned to 0.3, with the TOF laser sensor linear distance data as the main reference. When the absolute value of the rotational angular velocity of the gyroscope sensor attitude data exceeds the gyroscope sensor attitude data stability threshold, it is marked as a change in the gyroscope sensor attitude data. The gyroscope sensor responds to attitude changes in a more real-time manner. At this time, the weight of the gyroscope attitude data is increased to 0.6, and the weight of the TOF laser sensor straight-line distance data is reduced to 0.
4. The stability threshold for the gyroscope sensor attitude data is set by the system; the default value is 0.3 rad / s.
7. The camera focusing system based on multi-sensor data fusion according to claim 1, characterized in that, The projection distance of the target onto the camera's imaging plane includes: A projection distance formula is defined by combining dynamic weighting coefficients and geometric imaging principles. The projection distance of the target on the camera's imaging plane is then calculated using this formula. The projection distance formula is: ； in The distance of the target projected onto the camera's imaging plane; d represents the weight of the straight-line distance data of the TOF laser sensor; d represents the straight-line distance data of the TOF laser sensor. Weights for gyroscope attitude data; The tilt angle for gyroscope attitude data; This is the correction factor.
8. The camera focusing system based on multi-sensor data fusion according to claim 1, characterized in that, The attitude feature parameters include: The rotation direction is determined by the sign of the rotational angular velocity of the gyroscope attitude data; when the rotational angular velocity of the gyroscope attitude data is positive, the rotation direction is clockwise, and when the rotational angular velocity of the gyroscope attitude data is negative, the rotation direction is counterclockwise. Using the initial rotation angle as a reference, defined as zero degrees; the rotation angular velocity of the gyroscope attitude data is used to obtain the camera's rotation angle around the optical axis through a time integration formula; the time integration formula is: ； in The rotation angle; This represents the rotational angular velocity of the gyroscope's attitude data.
9. The camera focusing system based on multi-sensor data fusion according to claim 1, characterized in that, The focus control module includes: The camera lens is adjusted by a drive motor built into the lens, which moves the lens group closer to or further away along the optical axis. When the projection distance of the target on the camera's imaging plane increases, the drive motor drives the lens group to move away along the optical axis, increasing the focal length; when the projection distance of the target on the camera's imaging plane decreases, the drive motor drives the lens group to move closer along the optical axis, decreasing the focal length. The landscape / portrait switching is determined by a landscape / portrait switching trigger condition; the landscape / portrait switching trigger condition is: If the rotation angle of the posture feature parameter is within the horizontal / vertical screen switching trigger threshold, the horizontal / vertical screen switching trigger condition is determined to be met; if the rotation angle of the posture feature parameter is outside the horizontal / vertical screen switching trigger threshold, the horizontal / vertical screen switching trigger condition is determined not to be met. The screen orientation switching trigger threshold is set to [85°, 95°] based on the attitude feature parameters output by the data fusion module; The image orientation correction compensates for image skew caused by camera tilt through inverse compensation; the orientation compensation includes correction angle calculation and image geometric correction execution. The correction angle calculation is based on the tilt angle to calculate data that is equal in value but opposite in direction to the tilt angle. The image geometric correction process involves performing a geometric rotation on the original image; the geometric rotation process involves rotating the image in the opposite direction according to the calculated correction angle to eliminate image tilt.
10. A camera focusing method based on multi-sensor data fusion, characterized in that, The method includes the following steps: S001, the data acquisition module consists of a gyroscope sensor and a TOF laser sensor. The gyroscope sensor is used to acquire real-time attitude data of the camera, and the TOF laser sensor is used to acquire straight-line distance data between the camera and the target being photographed. The real-time attitude data includes tilt angle, rotation direction, and rotation angular velocity. S002, The data processing module is connected to the data acquisition module. The data processing module is used to preprocess the attitude data and straight-line distance data. The data preprocessing includes outlier filtering, data standardization, and time sequence alignment. S003. The data fusion module is connected to the data processing module. The data fusion module uses a weighted fusion algorithm to fuse the pre-processed gyroscope sensor attitude data and TOF laser sensor straight-line distance data to generate fused data. The weighted fusion algorithm determines the fusion weights of the gyroscope sensor attitude data and TOF laser sensor straight-line distance data based on a dynamic weighting strategy of the gyroscope sensor's angular velocity. The fused data includes the target's projection distance on the camera's imaging plane and attitude feature parameters. S004. The focus control module is connected to the data fusion module. The focus control module drives the camera lens to adjust the focal length according to the projection distance of the target on the camera imaging plane. The focal length adjustment is achieved by the built-in drive motor of the lens driving the lens group to move closer or further away along the optical axis. The focus control module triggers the horizontal / vertical screen switching and image orientation correction according to the posture characteristic parameters. The horizontal / vertical screen switching determines the vertical or horizontal screen mode through the horizontal / vertical screen switching trigger conditions. The image orientation correction compensates for the image distortion caused by the camera tilt through reverse compensation.