Searchlighting control system of unmanned aerial vehicle

Abstract

The invention relates to the technical field of unmanned aerial vehicle searchlighting control, in particular to an unmanned aerial vehicle searchlighting control system, which comprises a multi-source data acquisition module for acquiring a water surface image and unmanned aerial vehicle flight data; the water surface type analysis module is used for judging a calm or wave water surface according to the image gray variance and the edge density; the searchlighting output sub-control module is used for selecting a lamp control strategy according to a water surface type, identifying a high-confidence target in combination with multiple features and determining a wick power distribution parameter; the searchlighting output execution module drives the lampwick to illuminate according to the parameters; the target real-time tracking module is used for calculating the projection angle adjustment amount of the searchlight to aim at the target; according to the system and the method, the problem that an existing searchlight system cannot dynamically adapt to water surface optical characteristics, suppress reflection interference and accurately illuminate in a complex water environment at night can be solved, and the accuracy and efficiency of target recognition are remarkably improved.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of unmanned aerial vehicle searchlight control, in particular to a searchlight control system of unmanned aerial vehicle. BACKGROUND

[0002] With the in-depth application of unmanned aerial vehicle technology in the field of inspection and rescue, the intelligent level of the airborne searchlight system is put forward with higher requirements in the night water inspection and rescue task. The searchlight carried by the unmanned aerial vehicle as the key task load, its illumination control performance directly affects the accuracy of target identification and the efficiency of task execution. However, the adaptability of the existing unmanned aerial vehicle searchlight control system in the complex water surface environment still has significant deficiencies. Especially in the scene of superimposed multiple interference factors such as strong reflection, dynamic wave and low illumination, it is difficult to realize stable and accurate illumination regulation and control.

[0003] Among them, the searchlight control technology mainly focuses on basic angle adjustment and fixed power output, and lacks dynamic perception and response ability to the optical properties of the water surface. The current mainstream scheme adopts a unified brightness illumination strategy, which does not distinguish the reflection difference between calm water and wave water, resulting in that the image acquisition process is easily disturbed by mirror reflection or random splash highlights, causing overexposure or underexposure of the target area, and seriously weakening the recognition ability of the vision system. Although some improved schemes introduce infrared or laser assisted positioning, the core target is obstacle warning or device self-positioning, which cannot analyze the specific contrast between the target and the water surface reflection in the water rescue scene.

[0004] The existing technology has multiple limitations in searchlight control. First, the illumination intensity adjustment depends on preset gears or simple photosensitive feedback, which cannot dynamically adjust the output power of each lamp core of the searchlight according to the real-time water surface state. Second, there is a lack of effective discrimination mechanism for the type of water surface, and it is unable to adopt differentiated control strategies for the uniform reflection of calm water and the non-steady-state scattering of wave water. Third, target identification and illumination regulation are disconnected, and the searchlight irradiation area and potential target position are not linked, resulting in waste of effective illumination resources in non-key areas. Finally, in the strong reflection environment, there is a lack of environmental light compensation and multi-feature fusion target verification mechanism, which is easy to misjudge the splash as the target to be rescued, and then cause false illumination focusing. The above problems make the existing system difficult to meet the actual application requirements in scenes with high precision and high reliability requirements such as night water rescue, and an intelligent searchlight control system that integrates water surface state perception, target credible identification and searchlight fine regulation is urgently needed. SUMMARY

[0005] In view of the deficiencies of the prior art, the purpose of the present application is to provide a searchlight control system of an unmanned aerial vehicle, which can solve the problem that the searchlight system of the unmanned aerial vehicle in the prior art cannot dynamically adapt to the change of water surface optical characteristics, effectively suppress reflection interference and realize adaptive accurate illumination of the target area in complex scenes such as night water patrol and rescue, the present application realizes the cooperative control of the searchlight array through multi-source data sensing and intelligent decision-making, thereby improving the accuracy and efficiency of target identification in complex optical medium environment.

[0006] To achieve the above object, the present application provides the following technical scheme: A searchlight control system of an unmanned aerial vehicle, comprising: A multi-source data acquisition module acquires water surface images photographed by a visual camera carried by the unmanned aerial vehicle and acquires flight data information of the unmanned aerial vehicle; A water surface type analysis module judges the water surface type according to the gray variance and edge density of the water surface image, the water surface type including calm water surface and wave water surface; A searchlight output sub-control module selects a lamp control strategy according to the water surface type, the lamp control strategy including reflection rate analysis according to the water surface image and identification of a high-confidence target through multi-feature fusion combined with the flight data information of the unmanned aerial vehicle, and determination of power distribution parameters of each lamp filament according to the position of the high-confidence target and the acquired parameters; A searchlight output execution module drives the multiple lamp filaments of the searchlight to illuminate according to the power distribution parameters; A target real-time tracking module calculates a projection angle adjustment amount of the searchlight in real time according to the position of the high-confidence target or the center point of the water surface image, so that the center point of the water surface image always coincides with the high-confidence target; An image verification feedback module calculates an image definition index of the water surface image every certain number of frames, and triggers system reevaluation and adjustment instructions according to the image definition index.

[0007] Further, the lamp control strategy includes a theoretical sub-strategy and a regulation sub-strategy, the theoretical sub-strategy matching the calm water surface, the theoretical sub-strategy including determination of the total power of the searchlight according to a preset mapping table and uniform distribution of the total power to each lamp filament; the regulation sub-strategy matching the wave water surface, the regulation sub-strategy including determination of the output power of each lamp filament of the searchlight according to a pre-trained neural network model.

[0008] Further, the theoretical sub-strategy includes a preset mapping table, which includes a one-to-one correspondence of total power, height and ambient light intensity, the searchlight current height and the ambient current light intensity are included in the unmanned aerial vehicle flight data information, and the total power of the theory is indexed in the mapping table according to the searchlight current height and the ambient current light intensity, if there is no completely matched index entry in the mapping table, the total power of the theory is calculated by a multi-dimensional linear interpolation method.

[0009] Further, the regulation sub-strategy includes converting the pixel coordinates of the high-confidence target into the target azimuth angle deviation and the pitch angle deviation, analyzing the current height of the water wave according to the water surface image, and then inputting the searchlight current height, the current height of the water wave, the target azimuth angle deviation, the pitch angle deviation and the ambient current light intensity into the pre-trained neural network model to calculate the power distribution parameters of each lampwick by forward propagation.

[0010] Further, the light control strategy includes a reflectivity analysis step, which includes retrieving an initialized preset standard ambient light intensity, calculating an ambient light compensation coefficient with the ambient current light intensity, and calculating the water surface reflectivity distribution map after ambient compensation based on the ambient light compensation coefficient.

[0011] Further, the light control strategy further includes a reflectivity gradient determination step, a texture feature auxiliary screening step and a shape feature secondary screening step. The reflectivity gradient determination step calculates the gradient of the ambient-compensated water surface reflectivity distribution map in the horizontal and vertical directions using the Sobel operator, and synthesizes the gradient modulus value, if the gradient modulus value of a certain pixel is greater than the preset gradient threshold, it is preliminarily marked as a candidate abnormal point. The texture feature auxiliary screening step calculates the local binary pattern texture feature of the local neighborhood where the candidate abnormal point is located, and matches it with the templates in the preset target texture template library, if the matching degree is greater than or equal to the preset threshold, the candidate abnormal point is retained and upgraded to a suspected abnormal point. The shape feature secondary screening step obtains the circumscribed contour of the suspected abnormal point by a contour extraction algorithm, and calculates the area and aspect ratio of the minimum circumscribed rectangle of each contour, if the area is less than the preset area threshold, the suspected abnormal point is determined as an interference point and is removed, and the remaining points are confirmed as final abnormal points.

[0012] Further, the light control strategy further includes an abnormal block clustering step and a target confidence evaluation step. The abnormal block clustering step clusters the final abnormal points into one or more abnormal blocks by setting a neighborhood radius and a minimum number of points, so that spatially closely connected abnormal points are clustered into clusters, and each abnormal block is regarded as a suspected target to be tracked. The target confidence evaluation step extracts the center coordinates, area value and reflectivity mean value of each abnormal block, and constructs a confidence function according to the matching degree of the texture, the area value and the reflectivity mean value to calculate the target confidence, and determines whether the abnormal block is a high-confidence target according to the target confidence.

[0013] Further, the target real-time tracking module includes a searchlight tracking strategy, which includes comparing the center coordinates of the high-confidence target with the center coordinates of the visual image, calculating a pixel deviation value, and calculating the target azimuth angle adjustment amount and the target elevation angle adjustment amount of the searchlight according to the visual camera setting parameters and the current height of the searchlight through geometric relationship.

[0014] Further, the image verification feedback module includes a verification strategy, which includes frame processing the water surface image after searchlight control, analyzing the current frame image definition index, comparing the image definition index with the minimum definition threshold preset during initialization, and if the definition index calculated for three consecutive times is lower than the minimum definition threshold, triggering the system to re-evaluate and adjust the instruction to optimize the strategy in each module.

[0015] Further, the water surface type analysis module includes extracting an analysis region in the water surface image, calculating the gray variance of the analysis region, and obtaining the edge density of the analysis region by edge detection analysis of the edge pixel number, and comparing the gray variance and the edge density with the preset gray variance threshold and the edge density threshold, respectively, if the gray variance is less than or equal to the gray variance threshold, and the edge density is less than or equal to the edge density threshold, it is determined that the current water surface is a calm water surface, otherwise, it is determined that the current water surface is a wave water surface.

[0016] The beneficial effects of the present application: by setting the water surface type analysis, the environmental adaptability of the searchlight system can be improved, by setting the searchlight output sub-control module, the target recognition accuracy can be improved and fine lighting regulation can be realized, by setting the image verification feedback module, the image shooting can always maintain the consistency of the definition, specifically, through the water surface type analysis module, the calm and wave water surface is accurately distinguished according to the image gray variance and edge density, the corresponding matching theory sub-strategy and regulation sub-strategy are matched, the overexposure and underexposure problems caused by uniform lighting are avoided, the complex scenes such as night water strong reflection and dynamic wave are adapted; the light control strategy is combined with reflectivity analysis, multi-feature screening and confidence evaluation, effectively eliminates interference such as spray, accurately identifies high-confidence targets, solves the target misjudgment problem of the existing system; the searchlight output sub-control module determines the power distribution parameters of each lampwick for different water surface types through the mapping table interpolation or pre-training neural network, combines the target real-time tracking module to dynamically adjust the searchlight angle, avoids the waste of lighting resources, and ensures the accurate lighting of the target area; the image verification feedback module regularly detects the definition of the image, triggers the system to re-evaluate and adjust when the continuous unqualified, ensures that the lighting strategy always adapts to the environment and target state, improves the efficiency and reliability of the night water patrol rescue task execution. BRIEF DESCRIPTION OF DRAWINGS

[0017] Figure 1 is the overall flowchart in the present application; Figure 2 is the regulation sub-strategy flowchart when the water surface type is wave water surface in the present application; Figure 3 is the water surface type determination flowchart in the present application; Figure 4 is the searchlight and image analysis flowchart when the water surface type is calm water surface in the present application. DETAILED DESCRIPTION

[0018] The present application will be further described in detail below in combination with the drawings and examples. Wherein the same parts are denoted by the same reference numerals. It should be noted that the words "front", "back", "left", "right", "up" and "down" used in the following description refer to the directions in the drawings, and the words "bottom surface" and "top surface", "inner" and "outer" refer to the directions towards or away from the geometric center of a particular part.

[0019] Since the control of the searchlight carried by the unmanned aerial vehicle in the process of current unmanned aerial vehicle inspection and rescue can only realize the control of the angle, the control of the light source cannot be realized for specific environment, such as the water inspection and rescue mentioned in the application, the image acquisition will face the water reflection and the fluctuation of the water waves during the water inspection and rescue, so that the to-be-detected target cannot be clearly extracted and tracked during the image acquisition, therefore, the unmanned aerial vehicle searchlight control system with multi-source perception and intelligent decision is designed, specifically, the structure part includes the unmanned aerial vehicle body and the visual camera and the searchlight carried on the unmanned aerial vehicle, the searchlight is composed of an array of a plurality of independently controllable LED lamp cores, the calibration of the visual camera involves that a series of standard gray scale plates are shot under the controlled illumination environment, an accurate conversion model from the image gray value to the actual water reflectivity is established, and the accuracy of the subsequent reflectivity calculation is ensured, the conversion model usually shows a nonlinear function, which can be obtained through polynomial fitting.

[0020] The system part, as shown in Figure 1 the figure, includes a multi-source data acquisition module, a water surface type analysis module, a searchlight output sub-control module, a searchlight output execution module, a target real-time tracking module and an image verification feedback module, each module is integrated in the searchlight controller, in addition, the system has two prerequisite steps of core threshold value and model preset, through a large number of laboratory simulation experiments and actual field tests, a theoretical best shooting reflectivity threshold value is determined, which makes the target have the best identifiable in the night water surface image; in addition, by collecting a sample data set containing a large number of calm water surface and wave water surface images, the gray variance and edge density of these sample images are calculated and statistically analyzed, so as to determine the gray variance threshold value and the edge density threshold value which can effectively distinguish the calm water surface and the wave water surface, finally, the pre-trained power control model is implanted into the system, the power control model can be a model based on back propagation neural network, the network structure of which includes an input layer, at least one hidden layer and an output layer, or a multivariate linear regression model, the input and output relationship of the model has been verified through a large amount of simulation data and actual experimental data, for example, under a specific input parameter combination, the model can output the accurate power distribution scheme of each lamp core.

[0021] After the unmanned aerial vehicle takes off, it enters the continuous data acquisition mode, wherein the multi-source data acquisition module is responsible for synchronously acquiring the water surface image shot by the visual camera and the unmanned aerial vehicle flight data information, wherein the unmanned aerial vehicle flight data information includes the flight state parameters and the environmental parameters of the unmanned aerial vehicle, the flight state parameters include the flight height of the unmanned aerial vehicle and the attitude of the unmanned aerial vehicle, the height of the light outlet of the searchlight from the water surface can be deduced through the flight height of the unmanned aerial vehicle, the vertical height of the searchlight from the water surface can be measured by the millimeter wave radar carried by the unmanned aerial vehicle, in addition, it can also be estimated through the water surface image shot by the visual camera, similarly, the two can be combined to judge to obtain the accurate and reliable height; The environmental parameters include water height and ambient light intensity, wherein the water height can be obtained by visual or laser sensor cooperation, the continuous water surface image sequence captured by the visual camera is analyzed by the optical flow method or the background difference method to estimate the relative height of the water wave, the optical flow method mainly calculates the motion vector of the pixels between adjacent frames to identify the vertical displacement of the water surface texture, and the laser wave height sensor carried by the unmanned aerial vehicle directly obtains the distance change of the water surface relative to the unmanned aerial vehicle, and the accurate wave height data is obtained after data processing, the data of the direct measurement method is preferentially used, and the visual estimation result is used as auxiliary verification; in addition, the ambient light intensity can be obtained in real time by the high-precision photosensitive sensor built-in the visual camera, which provides a reference for the dynamic adjustment of the subsequent searchlight to adapt to different night environments such as moonlight, moonless night, urban light pollution and other illumination requirements.

[0022] After the unmanned aerial vehicle collects various data, the system automatically enters the water surface type analysis, as shown in the figure, Figure 3 The water surface type analysis module analyzes the water surface image in real time to determine whether the current water surface is in a calm water surface state or a wave water surface state, specifically, after the water surface image is preprocessed by denoising, distortion correction and other steps, a region located in the center of the image is selected as an analysis region, the analysis region can be all regions within the illumination range or part of the square region, in the analysis region, the gray value variance of all pixel points is calculated, the gray value variance reflects the dispersion degree of the pixel gray value, the pixel gray value of the calm water surface is distributed concentratedly due to uniform reflection of light, the variance is small, and the wave water surface has large pixel gray value fluctuation due to light scattering and shadow formation, so the variance is also large. At the same time, the Canny edge detection algorithm is applied to the analysis region to count the number of edge pixels detected in the region, and the number of edge pixels is divided by the total number of pixels in the analysis region to obtain the edge density, the calm water surface usually has fewer edges and lower edge density, and the wave water surface has more edges and higher edge density, in the type determination logic, the calculated gray value variance and edge density are combined with the preset threshold to determine the water surface type, if the calculated gray value variance is less than or equal to the preset calm water surface gray value variance threshold, and the edge density is less than or equal to the preset edge density threshold, the system determines that the current water surface is a calm water surface, and triggers the subsequent theoretical sub-strategy light control mode, otherwise, if the calculated gray value variance is greater than the preset calm water surface gray value variance threshold, or the edge density is greater than the preset edge density threshold, the system determines that the current water surface is a wave water surface, and triggers the subsequent regulation and control sub-strategy light control mode, in order to avoid misjudgment caused by single frame image noise, temporary water surface disturbance or occasional sensor error, the system introduces a continuous 3-frame confirmation mechanism, only when the continuous 3 frames of images all meet the same type determination condition, the system finally confirms the water surface type, and the light control strategy is switched accordingly.

[0023] Once the current water surface type is successfully determined, such as Figure 4 As shown, the searchlight output control module in the system makes decisions to control the output of the searchlights. Specifically, when the water surface is determined to be calm, since the water surface reflectivity distribution is relatively uniform and stable, no complex dynamic adjustments are needed. The searchlight output control module will trigger a theoretical sub-strategy, simply outputting the total power of the searchlights based on a preset theoretical model and evenly distributing it to each lamp core to ensure that the water surface reflectivity reaches the optimal threshold. Specifically, this is based on the real-time collected data on the current height of the searchlights. and current ambient light intensity The system queries a preset mapping table, which contains a one-to-one correspondence of power, height, and light intensity. This table, established through experimental calibration, records the total power value of the searchlight that enables the water surface reflectivity to reach the system's preset optimal threshold under different searchlight heights and ambient light intensities. If a match cannot be found in the mapping table... and For index entries with perfectly matching values, the searchlight output control module uses a multidimensional linear interpolation algorithm to calculate the theoretical total power. This method ensures a smooth transition and accurate estimation between discrete data points in the mapping table. Since calm water surfaces have uniform reflection characteristics, the searchlight output control module adopts a uniform distribution strategy to distribute the calculated theoretical total power evenly to all the lamp cores of the searchlight.

[0024] When the current water surface is determined to be a wave surface, such as Figure 2 As shown, the water surface reflection characteristics are complex and dynamic. The searchlight output control module will trigger a sub-strategy to reliably identify targets through reflectivity analysis and multi-feature fusion. A pre-trained neural network model is used to achieve independent and adaptive power allocation for each lamp element of the searchlight. First, reflectivity analysis and reliable anomaly identification are performed. Traditional reflectivity analysis is easily affected by ambient light fluctuations and random wave reflections, leading to target misjudgment. This stage significantly improves target recognition accuracy through strategies such as environmental compensation, multi-feature fusion, and confidence screening. Based on the grayscale value and reflectivity conversion model calibrated during initialization, a real-time ambient light compensation coefficient is introduced to eliminate the potential impact of ambient light intensity fluctuations on reflectivity calculation. First, the current ambient light intensity is acquired in real-time through the photosensor built into the vision camera. At the same time, the preset standard ambient light intensity is retrieved during initialization. Calculate the ambient light compensation coefficient , If the actual ambient light intensity is higher than the standard value, then If the grayscale value is too bright, the reflectance needs to be reduced to correct the overly bright environment; conversely, it should be increased. For water surface images that have undergone preprocessing such as noise reduction and distortion correction, the reflectance is calculated pixel by pixel. ,in, is the pixel coordinate, is the gray value of the pixel, is the gray value obtained by standard gray scale calibration at initialization and the reflectivity basis function, and finally outputs the environment-compensated water surface reflectivity distribution map, which ensures that the calculation results of the reflectivity of the same physical object under different environmental lighting conditions have consistency and comparability.

[0025] To avoid misjudging the spray as the target by only judging the continuity of reflectivity or a single threshold, this stage combines reflectivity gradient, texture features, and shape features for three-element fusion to determine abnormal points. The Sobel operator is used to calculate the gradient in the horizontal and vertical directions on the environment-compensated water surface reflectivity distribution map. The Sobel operator approximates the gradient of the image by calculating the gray difference between the pixel point and its neighborhood pixels. The gradient module value of the horizontal gradient and the vertical gradient is: The gradient module value reflects the steepness of the reflectivity change. If the gradient module value of a certain pixel is greater than the preset gradient threshold , it is considered that the reflectivity at this pixel has changed abruptly, and it is preliminarily marked as a candidate abnormal point.

[0026] Then, texture feature assisted screening is performed. For the local neighborhood of each candidate abnormal point, its local binary pattern texture feature is calculated. This feature describes the relative gray relationship between the neighborhood pixels and the center pixel, and can effectively capture the local texture pattern. The local texture feature of the water spray usually shows random distribution and no fixed texture pattern, and its local texture histogram distribution is relatively flat. The local texture feature of the rescue target such as the human body, clothing, and life buoy shows local texture consistency. For example, the fabric texture of clothing and the specific stripe texture of life buoy have specific peak values or patterns in the local texture histogram. The local texture feature histogram of the candidate abnormal point is matched with the templates in the preset target texture template library. The target texture template library contains local texture feature texture samples of common rescue targets such as the human body, life buoy, and life jacket. Through real data collection and labeling, if the matching degree is greater than or equal to the preset threshold, the candidate abnormal point is retained and upgraded to a suspected abnormal point.

[0027] Further, shape feature secondary screening is performed. For all suspected abnormal points, their circumscribed contours are obtained through a contour extraction algorithm. For each extracted contour, the area and the width-height ratio (the ratio of the long side to the short side) of its minimum circumscribed rectangle are calculated. If the area of the circumscribed rectangle is less than the preset area threshold If the aspect ratio is greater than the preset aspect ratio threshold, the suspected abnormal point is identified as an interference point and removed. The remaining points are confirmed as the final abnormal points. These points have three characteristics: abrupt change in reflectivity, target texture pattern, and reasonable target shape.

[0028] After identifying the final outliers, the control strategy continues with outlier clustering and target confidence assessment. The final outliers are clustered using the Density-Based Clustering (DBSCAN) algorithm, replacing the traditional K-means algorithm. This is because DBSCAN does not require pre-setting the number of clusters and is better suited to complex scenarios with no target, a single target, or multiple targets. In the DBSCAN algorithm, a neighborhood radius is set. and minimum points After clustering, the system generates one or more anomalous blocks, each of which is considered a suspected target requiring rescue. The system also calculates the core parameters of each anomalous block, including its center coordinates (i.e., the coordinates of all pixels within the block). coordinates and Average of coordinates , indicating the center position of the suspected target in the pixel coordinate system; area size, i.e., the total number of pixels contained in the anomaly block. Mean reflectance, which is the average reflectance of all pixels within the abnormal block. Based on this, target confidence is calculated, and a confidence function is constructed based on the area of ​​the anomalous block, the mean reflectance, and the texture matching degree: ,in, The preset maximum target area, For the theoretically optimal target reflectivity, For texture matching, this function comprehensively evaluates the confidence level of the target by weighted combination of multiple features, retaining only the confidence score. Larger anomalous blocks are considered high-confidence targets, and subsequent searchlight control model calculations are performed only on these high-confidence targets, further reducing the probability of misidentifying non-target interference objects as targets.

[0029] Once a high-confidence target is identified, the searchlight output control module will filter and extract key input parameters for the searchlight control model from the real-time acquired data, including the current height of the searchlight. Current height of the water waves , coordinates of the center of the anomaly block and the current ambient light intensity The center coordinates of the high-confidence targets obtained in the outlier block clustering step in the pixel coordinate system are... Using camera internal and external parameters, as well as the current altitude of the searchlight. Converted to target azimuth deviation in searchlight coordinate system (Yaw angle deviation) and pitch angle deviation For example, if the target is 100 pixels to the right of the image center, this pixel deviation will be converted into a specific yaw angle deviation value relative to the direction of the drone nose, and the pitch angle deviation is calculated in the same way.

[0030] After the above parameter extraction is completed, the searchlight output distribution control module inputs the standardized input parameters into the preset searchlight control model for power distribution calculation. The searchlight control model is a pre-trained back propagation neural network model, wherein the training target of the back propagation neural network model is to output the power distribution parameters of each wick by using the current height of the searchlight, the current height of the wave, the target azimuth deviation, the pitch angle deviation, and the current light intensity of the environment as five key parameters, so as to ensure that the output power distribution can make the water surface reflectivity of the high-confidence target area reach the best target recognition reflectivity interval. The training data includes actual night water scene test and simulation data, and the data includes n groups of data sets, each of which includes the above-mentioned five key parameters and the optimal output power of the wick, is composed of label data, covers the scenes of calm water surface under different light intensities and the scenes of wave surface under different light intensities, and the label data can be realized by actual measurement calibration or simulation optimization, that is, by using a group of data sets for actual measurement, manually adjusting or simulating fine-tuning the output power of the wick to make the water surface reflectivity reach the best effect of visual image acquisition. The input layer of the back propagation neural network model includes five neurons corresponding to the standardized current height of the searchlight , the current height of the wave , the target azimuth deviation , the pitch angle deviation , and the current light intensity of the environment . The neural network model has at least one hidden layer, and the hidden layer uses a ReLU activation function to introduce nonlinearity. The output layer includes neurons corresponding to the number of searchlight wicks, that is, the output data of the control model of the searchlight is the target power value of each neuron corresponding to the wick. The model calculates the target power value of each wick through forward propagation, and the neural network model is aimed at the azimuth of the high-confidence target in the searchlight coordinate system , , to improve the power of the wick covering the target area, enhance the illumination of the target area, ensure that the target reflectivity reaches the best threshold, and improve the contrast; at the same time, for the non-target area, especially the strong reflection wave area detected, the model will reduce the output power of the corresponding wick to reduce the interference of the wave reflection and avoid image overexposure. In addition, the model will also adjust the output power of the wick according to the wave height Adjust the overall power compensation coefficient: the higher the water wave height, the stronger the water surface scattering and absorption effect, the model will appropriately increase the overall power to offset the light attenuation caused by the wave height, to ensure the effective illumination of the target area, after the completion of each frame image analysis, the above reflectivity analysis and abnormal point reliable identification, searchlight control model input parameter acquisition and model calculation and lamp core power output steps will be repeated to realize the real-time dynamic adjustment of the searchlight power, the adjustment frequency is consistent with the frame rate of the vision camera, to ensure that the system can quickly respond to the dynamic changes of the water surface and the target movement.

[0031] Whether the water surface type is determined to be calm water or wave water, the target real-time tracking module needs to calculate and adjust the projection angle of the searchlight in real time through the vision image to ensure that the center of the searchlight's aperture is always accurately aligned with the suspected target or the area of interest. In terms of target center positioning, when the water surface type is calm water, if the searchlight output sub-control module does not identify any abnormal points, i.e., there is no target to be rescued in the current field of view, the system defaults to the center point of the vision image as the temporary target center, ensuring that the searchlight illumination area is maintained in the center of the image. If the system identifies a high-confidence target, such as a life buoy or buoy floating on calm water, the abnormal block center of the high-confidence target in the pixel coordinate system is taken as the target center. When the water surface type is wave water, the system directly takes the abnormal block center of the high-confidence target as the target center. In terms of angle deviation calculation, the target real-time tracking module accurately compares the coordinates of the current target center point in the pixel coordinate system with the center coordinates of the vision image , calculates the pixel deviation value and , according to the setting parameters of the vision camera, including the intrinsic matrix: focal length , principal point and the current height of the searchlight , through accurate geometric relationship calculation, the target azimuth adjustment amount and the pitch angle adjustment amount , are calculated. Among them, the pixel size is the actual physical size of a single pixel, and the angle execution is the angle adjustment instruction sent by the system to the servo motor of the searchlight pan-tilt, ensuring that the center of the searchlight's aperture can always be accurately and stably aligned with the target.

[0032] In addition, the image verification feedback module is configured to continuously monitor and verify the spotlight control effect, and trigger system re-evaluation and adjustment instructions according to the verification result, to realize adaptive optimization of the system strategy. Specifically, the image verification feedback module includes an image sharpness index calculation and verification strategy and optimization instruction. The image verification feedback module calculates the image sharpness index of the water surface image after spotlight control every certain number of frames. The image sharpness index includes an edge intensity value, an image entropy value, and a contrast index. The edge intensity value is used to evaluate the edge richness and sharpness of the image using the Tenengrad gradient operator. The higher the edge intensity value, the richer the edge information of the image, and the higher the visual sharpness. The image entropy value reflects the richness of the image information and the complexity of the texture. The larger the entropy value, the more detailed information the image contains, and the higher the sharpness. The contrast index is used to evaluate the brightness difference between the target and the background. A unified image sharpness comprehensive index is obtained through weighted averaging or a specific fusion algorithm. In terms of verification strategy and optimization instruction, the image verification feedback module continuously compares the calculated current frame image sharpness comprehensive index with the preset minimum sharpness threshold at initialization. If the system detects that the calculated sharpness comprehensive index is lower than the minimum sharpness threshold for three consecutive times, the system triggers a re-evaluation and adjustment instruction. That is, the current lighting strategy of the spotlight may no longer be suitable for the current environment or target state, and needs to be optimized. After receiving the re-evaluation instruction, the system will start a comprehensive adaptive adjustment process. The water surface type analysis module will re-determine the water surface type to confirm whether the environment has changed significantly. The spotlight output control module will re-execute reflectivity analysis, multi-feature fusion recognition, abnormal block clustering, and target confidence evaluation based on the latest multi-source data and water surface type, to ensure that high-confidence targets are accurately identified. If it is a wave water surface, the spotlight output control module will re-calculate the power distribution parameters of each lamp core through the pre-trained neural network model. If it is a calm water surface, the total power is calculated and evenly distributed by querying the mapping table. The target real-time tracking module will re-calculate the projection angle adjustment amount of the spotlight based on the latest identified target position, and drive the pan-tilt to adjust. This ensures that the spotlight control system can continuously optimize the lighting effect in various complex night water environments, and maximizes the accuracy and efficiency of target recognition.

[0033] The above is only a preferred embodiment of the present application, and the protection scope of the present application is not limited to the above-mentioned embodiments. Any technical solutions falling within the scope of the present application should be considered within the protection scope of the present application. It should be noted that, for ordinary skilled persons in the art, some improvements and refinements without departing from the principles of the present application should also be considered within the protection scope of the present application.

Claims

1. A searchlight control system for an unmanned aerial vehicle (UAV), characterized in that: The application relates to a water surface illumination system based on unmanned aerial vehicle (UAV) and a method thereof. The application comprises: a multi-source data acquisition module which acquires water surface images captured by a vision camera carried by an unmanned aerial vehicle (UAV) and acquires flight data information of the unmanned aerial vehicle (UAV); a water surface type analysis module which judges a water surface type according to a gray variance and an edge density of the water surface images, wherein the water surface type comprises a calm water surface and a wave water surface; a searchlight output sub-control module which selects a light control strategy according to the water surface type, wherein the light control strategy comprises performing reflectivity analysis according to the water surface images, combining the flight data information of the unmanned aerial vehicle (UAV) to identify a high-confidence target through multi-feature fusion, determining power distribution parameters of each lampwick according to a position of the high-confidence target and the acquired parameters; a searchlight output execution module which drives multiple lampwicks of a searchlight to perform illumination according to the power distribution parameters; a target real-time tracking module which calculates a projection angle adjustment amount of the searchlight in real time according to the position of the high-confidence target or a center point of the water surface images so that the center point of the water surface images always coincides with the high-confidence target; 2. The searchlight control system of claim 1, wherein: an image verification feedback module which calculates an image definition index of the water surface images every certain number of frames and triggers system reevaluation and adjustment instructions according to the image definition index.

3. The searchlight control system of claim 2, wherein: The light control strategy comprises a theoretical sub-strategy and a regulation and control sub-strategy, the theoretical sub-strategy is matched with the calm water surface, the theoretical sub-strategy comprises determining a total power of the searchlight according to a preset mapping table and uniformly distributing the total power to each lampwick, and the regulation and control sub-strategy is matched with the wave water surface, the regulation and control sub-strategy comprises determining an output power of each lampwick of the searchlight according to a pre-trained neural network model.

4. The searchlight control system of claim 3, wherein: The theoretical sub-strategy comprises a preset mapping table, the mapping table comprises a one-to-one correspondence of a total power, a height and an ambient light intensity, the flight data information of the unmanned aerial vehicle (UAV) comprises a current height of the searchlight and a current ambient light intensity, the current height of the searchlight and the current ambient light intensity are used to index the theoretical total power in the mapping table, if there is no completely matched index entry in the mapping table, a multi-dimensional linear interpolation method is used to calculate the theoretical total power.

5. The searchlight control system of any one of claims 1-4, wherein: The regulation and control sub-strategy comprises converting pixel coordinates of the high-confidence target into a target azimuth angle deviation and a target elevation angle deviation, analyzing a current wave height of the water surface according to the water surface images, and inputting the current height of the searchlight, the current wave height of the water surface, the target azimuth angle deviation, the target elevation angle deviation and the current ambient light intensity into the pre-trained neural network model to calculate and output the power distribution parameters of each lampwick through forward propagation.

6. The searchlight control system of claim 5, wherein: The light control strategy comprises a reflectivity analysis step, the reflectivity analysis step comprises calling an initialized preset standard ambient light intensity, calculating an ambient light compensation coefficient by taking the standard ambient light intensity and the current ambient light intensity as input, and performing pixel-by-pixel reflectivity calculation on the water surface images based on the ambient light compensation coefficient to obtain an ambient-compensated water surface reflectivity distribution map. The light control strategy further comprises a reflectivity gradient judgment step, a texture feature auxiliary screening step and a shape feature secondary screening step. The reflectivity gradient determination step uses Sobel operators to calculate the gradients in the horizontal and vertical directions of the water surface reflectivity distribution map after environmental compensation, and synthesizes the gradient modulus values thereof. If the gradient modulus value of a certain pixel is greater than a preset gradient threshold value, the pixel is preliminarily marked as a candidate abnormal point. The texture feature auxiliary screening step calculates the local binary pattern texture feature of the local neighborhood in which the candidate abnormal point is located, and matches the texture feature with a template in a preset target texture template library. If the matching degree is greater than or equal to a preset threshold value, the candidate abnormal point is retained and upgraded to a suspected abnormal point. The shape feature secondary screening step obtains the circumscribed contour of the suspected abnormal point through a contour extraction algorithm, and calculates the area and the width-height ratio of the minimum circumscribed rectangle of each contour. If the area is less than a preset area threshold value, the suspected abnormal point is determined to be an interference point and is removed, and the remaining points are confirmed to be final abnormal points.

7. The searchlight control system of claim 6, wherein: The light control strategy further includes an abnormal block clustering step and a target confidence evaluation step. The abnormal block clustering step clusters the final abnormal points into clusters by setting a neighborhood radius and a minimum number of points, so that the abnormal points that are spatially closely connected are clustered into one or more abnormal blocks, and each abnormal block is regarded as a suspected target to be tracked. The target confidence evaluation step extracts the center coordinates, the area value and the reflectivity average value of each abnormal block, constructs a confidence function according to the matching degree of the texture, the area value and the reflectivity average value to calculate the target confidence, and determines whether the abnormal block is a high-confidence target according to the target confidence.

8. The searchlight control system of claim 7, wherein: The target real-time tracking module includes a searchlight tracking strategy. The searchlight tracking strategy includes comparing the center coordinates of the high-confidence target with the center coordinates of the visual image, calculating a pixel deviation value, and calculating the target azimuth adjustment amount and the target elevation adjustment amount of the searchlight according to the visual camera setting parameters and the current height of the searchlight through geometric relationship.

9. The searchlight control system of claim 1, wherein: The image verification feedback module includes a verification strategy. The verification strategy includes performing frame-by-frame processing on the water surface image after searchlight control, analyzing the current frame image definition index, comparing the image definition index with a preset minimum definition threshold value, and if the definition index calculated for three consecutive times is all lower than the minimum definition threshold value, triggering a system re-evaluation and adjustment instruction to optimize the strategies in each module.

10. The searchlight control system of claim 2, wherein: The water surface type analysis module includes extracting an analysis region from the water surface image, calculating the gray variance thereof, and obtaining the edge density of the analysis region through edge detection of the edge pixel number. The gray variance and the edge density are compared with preset gray variance and edge density threshold values, respectively. If the gray variance is less than or equal to the gray variance threshold value, and the edge density is less than or equal to the edge density threshold value, it is determined that the current water surface is a calm water surface, otherwise, it is determined that the current water surface is a wave water surface.

Images