Intelligent energy-saving light supplementing method and system based on ship multi-target video guidance

Abstract

The invention relates to the technical field of energy-saving light supplement, and particularly discloses an intelligent energy-saving light supplement method and system based on ship multi-target video guidance, and the method comprises the steps: obtaining a top view of a target water area, and determining the basic risk degree of each position based on a still life contour in the top view; acquiring a ship position in real time, querying a basic risk degree at the ship position, and determining a basic light supplementing frequency according to the basic risk degree; acquiring a night vision image based on a night vision camera, and synchronously determining an image range based on detection parameters of the night vision camera; identifying the night vision image, and determining the floating light supplementing frequency at each position in the image range; and performing light supplement on each position based on the basic light supplement frequency and the floating light supplement frequency. According to the intelligent light supplementing system provided by the invention, the light supplementing process is controlled as required under the condition of black light and low illumination at night, and the energy consumption of the light supplementing system can be reduced to the maximum extent, so that the energy-saving effect is achieved.

Description

Technical Field

[0001] The invention relates to the technical field of energy-saving supplementary lighting, and in particular to an intelligent energy-saving supplementary lighting method and system based on multi-target video guidance of a ship. Background Art

[0002] In order to meet the requirements of static images of ships captured by electronic bayonet on water, and to be applicable to the combined recognition of multiple external features of the ship itself, such as ship name, color and texture distribution, spatial structure, etc., the captured images need to have sufficient clarity and resolution, and also need to retain color information, i.e. color images. The images captured at night through infrared and laser fill light can only be black and white, which cannot meet the requirements of high image recognition. Therefore, under extremely low illumination or black light conditions, white light fill light is required for dynamic sailing ships.

[0003] The existing general waterborne electronic bayonet white light fill-in system is generally controlled by a time switch. The time switch presets the time period for the fill-in system to be turned on at night and off during the day, and the fill-in lighting is continuously performed during the time period when the fill-in lighting is turned on; while the night capture system only needs the fill-in system to work when a ship target passes by, this asymmetric supply and demand situation causes a large waste of energy; how to reduce the energy consumption in the fill-in process is the technical problem that the technical solution of the present invention intends to solve. Summary of the invention

[0004] The object of the present invention is to provide an intelligent energy-saving fill lighting method and system based on multi-target video guidance of ships to solve the problems raised in the above background technology.

[0005] To achieve the above object, the present invention provides the following technical solutions:

[0006] An intelligent energy-saving fill-light method based on multi-target video guidance of a ship, the method comprising:

[0007] Obtain a bird's-eye view of the target waters, and determine the basic risk level at each location based on the static object outline in the bird's-eye view;

[0008] Acquire the ship's position in real time, query the basic risk level at the ship's position, and determine the basic fill light frequency according to the basic risk level;

[0009] Acquire a night vision image based on a night vision camera, and synchronously determine the image range based on detection parameters of the night vision camera; the night vision camera includes a static camera installed on the shoreline and a dynamic camera installed on the ship;

[0010] Identify night vision images and determine the floating fill light frequency at each position within the image range;

[0011] Fill light is applied to each position based on the basic fill light frequency and the floating fill light frequency.

[0012] As a further solution of the present invention: the step of obtaining a top view of the target water area and determining the basic risk at each position based on the static object contour in the top view includes:

[0013] Regularly acquire remote sensing data of the target water area, and extract a bird's-eye view of the target water area based on the remote sensing data;

[0014] Performing contour recognition on the top view to determine the contour of the non-water body;

[0015] According to the coordinates of the non-water body contour, the contour body and risk level are queried in the preset record library. When the query fails, the risk level of the corresponding non-water body contour is set to the highest value;

[0016] Determine the basic risk at each location based on the risk of all non-water body contours;

[0017] The relationship between the basic risk at each location and the risk of all non-water body contours is:

[0018] In the formula, J(x,y) represents the basic risk at the position (x,y), (i k ,j k ) represents the coordinates of the intersection of the line connecting the position (x, y) and the center point of the k-th contour body and the boundary of the k-th contour body, σ is a preset parameter, and F k is the risk of the kth contour subject; K is the total number of non-water body contours.

[0019] As a further solution of the present invention: the relationship between the basic risk and the basic fill light frequency is:

[0020] f(x,y)=αJ(x,y)+βdJ(x,y); where f(x,y) represents the basic fill light frequency at the position (x,y), α and β are the preset correction coefficients, dJ(x,y) represents the difference of the basic risk at the position (x,y), and the four-neighborhood calculation scheme is adopted;

[0021] The step of synchronously determining the image range based on the detection parameters of the night vision camera includes:

[0022] The direction vector and wide angle of the night vision camera are obtained, the viewing cone is determined according to the direction vector and the wide angle, and the image range is determined according to the viewing cone and a preset interface; the interface includes a water surface and a plane at a preset distance.

[0023] As a further solution of the present invention: the step of identifying the night vision image and determining the floating fill light frequency at each position within the image range comprises:

[0024] Extract water color values ​​in night vision images;

[0025] Extract the histogram of the night vision image and obtain the pixel percentage corresponding to each grayscale value;

[0026] Taking the pixel percentage corresponding to the water body color value as the mean, calculating the class standard deviation of all pixel percentages;

[0027] The floating fill light frequency is determined according to the class standard deviation.

[0028] As a further solution of the present invention: the step of performing fill light on each position based on the basic fill light frequency and the floating fill light frequency includes:

[0029] Get the working range of the fill light device;

[0030] Get the basic fill light frequency and floating fill light frequency at each position within the working range;

[0031] Determine the fill light cycle according to the basic fill light frequency and the floating fill light frequency;

[0032] After each fill light cycle, a fill light instruction pointing to the position is generated;

[0033] Send the fill light command to the fill light device.

[0034] As a further solution of the present invention: the method further comprises:

[0035] At every preset time period, read the basic fill light frequency at all positions and the floating fill light frequency at the most recent moment;

[0036] Add the basic fill light frequency and the most recent floating fill light frequency, and determine the image acquisition frequency at each position according to the sum;

[0037] The image acquisition frequency at each location is broadcast to all ships.

[0038] The technical solution of the present invention also provides an intelligent energy-saving fill light system based on multi-target video guidance of ships, the system comprising:

[0039] A location assessment module is used to obtain a bird's-eye view of the target waters and determine the basic risk level at each location based on the static object outline in the bird's-eye view;

[0040] A basic frequency determination module is used to obtain the ship position in real time, query the basic risk level at the ship position, and determine the basic fill light frequency according to the basic risk level;

[0041] An image acquisition module, used for acquiring night vision images based on a night vision camera, and synchronously determining the image range based on the detection parameters of the night vision camera; the night vision camera includes a static camera installed on the shoreline and a dynamic camera installed on the ship;

[0042] An image recognition module is used to recognize night vision images and determine the floating fill light frequency at each position within the image range;

[0043] The fill light instruction generation module is used to perform fill light on each position based on the basic fill light frequency and the floating fill light frequency.

[0044] As a further solution of the present invention: the position assessment module includes:

[0045] A bird's-eye view acquisition unit is used to periodically acquire remote sensing data of a target water area and extract a bird's-eye view of the target water area based on the remote sensing data;

[0046] A contour recognition unit, used to perform contour recognition on the top view to determine the contour of the non-water body;

[0047] A risk query unit is used to query the contour body and risk in a preset record library according to the coordinates of the non-water body contour. When the query fails, the risk of the corresponding non-water body contour is set to the highest value;

[0048] A risk application unit, used to determine the basic risk at each location according to the risk of all non-water body contours;

[0049] The relationship between the basic risk at each location and the risk of all non-water body contours is:

[0050] In the formula, J(x,y) represents the basic risk at the position (x,y), (i k ,j k ) represents the coordinates of the intersection of the line connecting the position (x, y) and the center point of the k-th contour body and the boundary of the k-th contour body, σ is a preset parameter, and F k is the risk of the kth contour subject; K is the total number of non-water body contours.

[0051] As a further solution of the present invention: the relationship between the basic risk and the basic fill light frequency is:

[0052] f(x,y)=αJ(x,y)+βdJ(x,y); where f(x,y) represents the basic fill light frequency at the position (x,y), α and β are the preset correction coefficients, dJ(x,y) represents the difference of the basic risk at the position (x,y), and the four-neighborhood calculation scheme is adopted;

[0053] The content of synchronously determining the image range based on the detection parameters of the night vision camera includes:

[0054] The direction vector and wide angle of the night vision camera are obtained, the viewing cone is determined according to the direction vector and the wide angle, and the image range is determined according to the viewing cone and a preset interface; the interface includes a water surface and a plane at a preset distance.

[0055] As a further solution of the present invention: the image recognition module includes:

[0056] A water color value extraction unit, used to extract water color values ​​from night vision images;

[0057] A ratio acquisition unit is used to extract the histogram of the night vision image and obtain the pixel ratio corresponding to each grayscale value;

[0058] A class standard deviation calculation unit, used to calculate the class standard deviation of all pixel proportions by taking the pixel proportion corresponding to the water body color value as the mean;

[0059] The class standard deviation application unit is used to determine the floating fill light frequency according to the class standard deviation.

[0060] Compared with the prior art, the beneficial effects of the present invention are: the present invention provides an intelligent fill light system, which controls the fill light process on demand under dark light and low illumination conditions at night, and can greatly reduce the energy consumption of the fill light system, thereby achieving energy-saving effects. BRIEF DESCRIPTION OF THE DRAWINGS

[0061] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings in the following description are only some embodiments of the present invention.

[0062] Figure 1 This is a flowchart of an intelligent energy-saving fill lighting method based on ship multi-target video guidance.

[0063] Figure 2 This is the first sub-process flowchart of the intelligent energy-saving fill lighting method based on ship multi-target video guidance.

[0064] Figure 3 This is the second sub-process flowchart of the intelligent energy-saving fill lighting method based on ship multi-target video guidance.

[0065] Figure 4 This is the third sub-process flowchart of the intelligent energy-saving fill lighting method based on ship multi-target video guidance.

[0066] Figure 5 This is a structural block diagram of the intelligent energy-saving fill lighting system based on ship multi-target video guidance. DETAILED DESCRIPTION

[0067] In order to make the technical problems, technical solutions and beneficial effects to be solved by the present invention more clearly understood, the present invention is further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not intended to limit the present invention.

[0068] Figure 1 The flowchart of the intelligent energy-saving fill-light method based on multi-target video guidance of a ship is shown in the figure. In an embodiment of the present invention, an intelligent energy-saving fill-light method based on multi-target video guidance of a ship comprises:

[0069] Step S100: obtaining a bird's-eye view of the target water area, and determining a basic risk at each position based on the outline of a static object in the bird's-eye view;

[0070] The application scenario of the technical solution of the present invention is the water area inspection scenario. For some water areas such as lakes, staff need to patrol at night. The patrol equipment is a ship. Cameras and supplementary lights are installed on the ship to obtain and record data during the patrol process. For a water area, a bird's-eye view of the water area can be obtained with the help of existing technology. By identifying the bird's-eye view, the outline of still life can be obtained. The still life outline is some anchoring points in the water area, such as small islands. Under normal conditions, the water area presents a stable state, but for ships, the closer to these still life outlines, the higher the probability of risk. Therefore, the basic risk level at each position in the water area is determined based on the still life outline, and the basic risk level indicates the probability of risk occurring at the corresponding position of the ship.

[0071] Step S200: acquiring the ship position in real time, querying the basic risk level at the ship position, and determining the basic fill light frequency according to the basic risk level;

[0072] When applied, the ship's position is obtained in real time, and the basic risk level is queried based on the ship's position. The basic fill light frequency is determined by the basic risk level. The higher the basic risk level, the higher the basic fill light frequency, and more resources are consumed. When the basic risk level is low, the basic fill light frequency is low, and fewer resources are consumed. This constructs a differentiated fill light solution. The closer to the still object, the more resources the fill light consumes, and the farther away from the still object, the fewer resources the fill light consumes, thereby reducing resource consumption in a safe state and achieving energy-saving function.

[0073] Step S300: acquiring a night vision image based on a night vision camera, and synchronously determining an image range based on detection parameters of the night vision camera; the night vision camera includes a static camera installed on the shoreline and a dynamic camera installed on the ship;

[0074] The working monitoring screen of the night vision camera is black and white at night, and in color during the day. The night vision image is obtained based on the night vision camera. If there is fill light, the image obtained is equivalent to the image in the daytime, which is a color image. If there is no fill light, it is a black and white image. When acquiring the night vision image, the detection parameters of the night vision camera are obtained synchronously to determine the image range it collects. The image range indicates the positions of the images collected by the night vision camera, which is generally related to the centerline direction and wide angle of the night vision camera.

[0075] Step S400: Identify the night vision image and determine the floating fill light frequency at each position within the image range;

[0076] The basic fill light frequency determined based on the bird's-eye view actually means the basic risk. On this basis, each ship will obtain night vision images in real time. By identifying the night vision images, the real-time risk can be determined. The floating fill light frequency at each position can be determined based on the real-time risk. That is, the floating fill light frequency reflects the dynamic real-time risk level.

[0077] It is worth mentioning that the floating fill light frequency can also be determined in conjunction with radar. Radar can also be used to detect whether there are any unnecessary objects around, and then determine the dynamic and real-time risk level.

[0078] Step S500: performing fill light on each position based on the basic fill light frequency and the floating fill light frequency;

[0079] After the floating fill light frequency is determined, for the bird's-eye view, each position has a basic fill light frequency and multiple floating fill light frequencies (the number is the same as the number of ships). By adding them together, a comprehensive frequency can be obtained. There is a definite relationship between frequency and period. The period determined by the comprehensive frequency is called the fill light period.

[0080] Finally, a fill light instruction is generated after each fill light cycle. Specifically, when generating the fill light instruction, the night vision image needs to be recognized again. If there are some abnormal contours, it is necessary to determine a longer fill light duration and a higher fill light intensity in the corresponding direction.

[0081] Figure 2 The first sub-process flowchart of the intelligent energy-saving fill lighting method based on multi-target video guidance of ships is shown in the figure. The steps of obtaining a bird's-eye view of the target waters and determining the basic risk at each position based on the still object outline in the bird's-eye view include:

[0082] Step S101: acquiring remote sensing data of a target water area at a regular interval, and extracting a bird's-eye view of the target water area based on the remote sensing data;

[0083] Step S102: performing contour recognition on the top view to determine the contour of the non-water body;

[0084] Step S103: querying the contour body and risk in a preset record library according to the coordinates of the non-water body contour. When the query fails, the risk of the corresponding non-water body contour is set to the highest value;

[0085] Step S104: determining the basic risk at each location according to the risk of all non-water body contours;

[0086] The above content explains the process of determining the basic risk level. For the target water area, its area is generally large. The way to obtain the bird's-eye view can be through a drone or a remote sensing satellite. The remote sensing satellite data can be obtained with the help of existing map services. Therefore, this application uses remote sensing data; remote sensing data is an image from a bird's-eye view. It is very simple to extract the bird's-eye view, which is an image simplification process. The simplest way is to extract the contour in the remote sensing image with the help of a contour recognition algorithm; the contour recognition process is used more than once in the technical solution of the present invention. Since it is a very conventional image processing method, the present invention will not repeat it.

[0087] Specifically, in the overhead view of the target water area, except for the water body, other contours can be regarded as static contours, that is, contour recognition is performed on the overhead view, and the water body part is eliminated from all contours, and the non-water body contour obtained is the target of analysis; then, the coordinates of each non-water body contour are obtained, and the contour subject and risk degree are queried in the preset record library according to the coordinates of the non-water body contour. The data in the record library indicate which locations have which static objects, such as some anchorages or pumice. Since the water area is in a stable state most of the time, most of the coordinates of the non-water body contours obtained are in the record library. If not, it means that there are extra objects. At this time, the risk degree is set to the highest value, indicating that the contour is an abnormal contour. The closer to it, the higher the possibility of risk.

[0088] Finally, the basic risk at each location can be determined based on the risk of all non-water body contours. The relationship between the basic risk at each location and the risk of all non-water body contours is:

[0089] In the formula, J(x,y) represents the basic risk at the position (x,y), (i k ,j k ) represents the coordinates of the intersection of the line connecting the position (x, y) and the center point of the k-th contour body and the boundary of the k-th contour body, σ is a preset parameter, and F k is the risk of the kth contour subject; K is the total number of non-water body contours.

[0090] The above content provides a certain calculation process, the principle of which is that, for a certain position, it is necessary to calculate the degree of influence of each non-water body contour at the position, and then superimpose them; for the degree of influence of each non-water body contour at the position, first calculate the weight of the position relative to the contour, and then multiply it by the risk. The weight adopts Gaussian weight, and the distance item in the Gaussian weight is calculated using the position and the intersection point, and the intersection point is the intersection of the line connecting the position and the center of the contour and the contour.

[0091] As a preferred embodiment of the technical solution of the present invention, some details are supplemented and defined as follows:

[0092] The relationship between the basic risk and the basic fill light frequency is:

[0093] f(x,y)=αJ(x,y)+βdJ(x,y); where f(x,y) represents the basic fill light frequency at the position (x,y), α and β are the preset correction coefficients, dJ(x,y) represents the difference of the basic risk at the position (x,y), and a four-neighborhood calculation scheme is adopted.

[0094] The above content explains the relationship between the basic risk and the basic fill light frequency. Specifically, the basic fill light frequency is related to the basic risk itself and the difference between the basic risk and the surrounding area. Among them, the difference between the basic risk and the surrounding area adopts four-neighborhood difference, that is, the difference between it and the four adjacent positions above, below, left and right is calculated, and then the average is obtained.

[0095] The step of synchronously determining the image range based on the detection parameters of the night vision camera includes:

[0096] The direction vector and wide angle of the night vision camera are obtained, the viewing cone is determined according to the direction vector and the wide angle, and the image range is determined according to the viewing cone and a preset interface; the interface includes a water surface and a plane at a preset distance.

[0097] The detection parameters are inherent parameters of the night vision camera. The detection parameters are limited to the direction vector and the wide angle. When using the night vision camera, these two parameters are built-in parameters of the night vision camera and can be read directly. The viewing cone is determined according to the direction vector and the wide angle, and the interface between the viewing cone and the preset interface is calculated to determine the image range. This process can be understood as dividing the viewing cone according to the preset interface to obtain the range corresponding to the captured image.

[0098] Figure 3 This is a second sub-flow chart of the intelligent energy-saving fill light method based on ship multi-target video guidance. The step of identifying the night vision image and determining the floating fill light frequency at each position within the image range includes:

[0099] Step S401: extracting water color values ​​from night vision images;

[0100] Step S402: extracting a histogram of the night vision image to obtain the percentage of pixels corresponding to each grayscale value;

[0101] Step S403: taking the pixel percentage corresponding to the water body color value as the mean, and calculating the class standard deviation of all pixel percentages;

[0102] Step S404: determining a floating fill light frequency according to the class standard deviation.

[0103] The above content describes the generation process of the floating fill light frequency. For the image acquisition process, most of the acquired images are water bodies. The parameters corresponding to the water bodies in the night vision images are obtained, which are called water body color values. Then, the histogram of the night vision image is extracted to obtain the number of pixels corresponding to each grayscale value. Since the number of pixels is large, data simplification is required. The simplification method is to divide the number of pixels by the total number of pixels. The obtained ratio is called the pixel ratio. At this time, the histogram is converted into an array. The serial number of each element in the array corresponds to a grayscale value, and the element value represents the pixel ratio of the corresponding grayscale value. Among them, the grayscale value range is an integer in [0,255].

[0104] For the obtained array, the proportion of pixels corresponding to the water body color value is taken as the mean, and the standard deviation is calculated. Since it is not really calculated based on the mean, it is called the class standard deviation. The class standard deviation reflects the dispersion of each value in the array. The lower the dispersion, the closer the value of each pixel in the night vision image is to the water body color value. At this time, it is more normal. The greater the dispersion, the farther the value of each pixel in the night vision image is from the water body color value. At this time, it is more abnormal. Correspondingly, the more abnormal it is, the more the floating fill light frequency should be increased.

[0105] Regarding the floating fill light frequency, it is proportional to the class standard deviation and preferably has a negative value. One solution is:

[0106] f ′ =C(B-B0); where f ′ represents the floating fill light frequency, B represents the class standard deviation, B0 represents the preset reference value, and C is the preset normal number.

[0107] Figure 4 The third sub-flow chart of the intelligent energy-saving fill light method based on multi-target video guidance of ships, wherein the step of performing fill light on each position based on the basic fill light frequency and the floating fill light frequency includes:

[0108] Step S501: obtaining the working range of the fill light device;

[0109] Step S502: obtaining a basic fill light frequency and a floating fill light frequency at each position within the working range;

[0110] Step S503: determining a fill light cycle according to the basic fill light frequency and the floating fill light frequency;

[0111] Step S504: after each fill light cycle, generating a fill light instruction pointing to the position;

[0112] Step S505: Send the fill light instruction to the fill light device.

[0113] In an example of the technical solution of the present invention, the fill light device is generally installed on the shoreline, and its working process is as follows:

[0114] The water bayonet capture system obtains the ship's AIS (AIS is an automatic tracking system installed on the ship, which can simultaneously obtain the ship's (GPS) position, speed, change of course rate and course and other ship dynamics and ship static information such as ship name, call sign, draft and dangerous goods) position information, visible light video, or infrared thermal imaging video information, screens the information in order, and performs analysis and detection processing. According to the processed ship target position information, it automatically triggers the fill light system to generate a work control instruction. The instruction parameters include the number of channels opened and the brightness value of the fill light system to dynamically fill light the ship target; when the capture system completes the capture task of the ship target, the system automatically generates a fill light system work end instruction, the fill light system ends its work, enters the standby state, and waits for the next control instruction.

[0115] In the technical solution of the present invention, since the risk level of each position is determined based on the real-time collected data, and then the basic fill light frequency and floating fill light frequency of each position are determined, an additional fill light solution can be provided. First, the ranges for which the fill light device can provide fill light are obtained. For the positions within the range, the basic fill light frequency and the floating fill light frequency are read. The fill light frequency is determined based on the basic fill light frequency and the floating fill light frequency. Once the time has passed, a fill light instruction pointing to the position is sent to the fill light device.

[0116] The above two solutions are actually complementary to each other. In practical applications, if the water bayonet capture system is not activated, the fill-in light equipment can also fill-in light the water area with lower energy consumption. This is mainly to deal with some situations that cannot be identified, such as the failure of ship AI S. When the water bayonet capture system works normally, the water area can be filled-in light according to the logic of the previous solution. The two are not in conflict, and compared with the existing control scheme based on time-controlled switches, the energy consumption is relatively low.

[0117] As a preferred embodiment of the technical solution of the present invention, the method further includes:

[0118] At every preset time period, read the basic fill light frequency at all positions and the floating fill light frequency at the most recent moment;

[0119] Add the basic fill light frequency and the most recent floating fill light frequency, and determine the image acquisition frequency at each position according to the sum;

[0120] The image acquisition frequency at each location is broadcast to all ships.

[0121] In an example of the technical solution of the present invention, the image acquisition process is also specifically limited. The image acquisition process is completed by a night vision camera. Every preset time period, the basic fill light frequency at all positions and the floating fill light frequency at the most recent moment are read, and the sum is added to obtain the final frequency at each position in the map. The image acquisition frequency is determined according to the direct proportion of the final frequency. The image acquisition frequency is how many frames of images are collected per second, which indicates how many frames of images the ship needs to collect per second at each position. This can be represented by a matrix. The row and column positions in the matrix represent the actual positions, and the values ​​of the elements in the matrix represent the image acquisition frequency, which is broadcast to all ships working in the waters. Each ship can adjust the image acquisition frequency according to its real-time position, thereby acquiring images at a dynamic frequency, which is more energy-efficient than the traditional fixed frequency.

[0122] Figure 5 The structure diagram of the intelligent energy-saving fill-light system based on multi-target video guidance of ships is shown in FIG. In an embodiment of the present invention, an intelligent energy-saving fill-light system based on multi-target video guidance of ships is provided. The system 10 includes:

[0123] A location assessment module 11 is used to obtain a bird's-eye view of the target water area and determine a basic risk level at each location based on the static object outline in the bird's-eye view;

[0124] A basic frequency determination module 12 is used to obtain the ship position in real time, query the basic risk level at the ship position, and determine the basic fill light frequency according to the basic risk level;

[0125] An image acquisition module 13 is used to acquire night vision images based on a night vision camera and synchronously determine the image range based on the detection parameters of the night vision camera; the night vision camera includes a static camera installed on the shoreline and a dynamic camera installed on the ship;

[0126] An image recognition module 14 is used to recognize night vision images and determine the floating fill light frequency at each position within the image range;

[0127] The fill light instruction generating module 15 is used to perform fill light on each position based on the basic fill light frequency and the floating fill light frequency.

[0128] Furthermore, the location assessment module 11 includes:

[0129] A bird's-eye view acquisition unit is used to periodically acquire remote sensing data of a target water area and extract a bird's-eye view of the target water area based on the remote sensing data;

[0130] A contour recognition unit, used to perform contour recognition on the top view to determine the contour of the non-water body;

[0131] A risk query unit is used to query the contour body and risk in a preset record library according to the coordinates of the non-water body contour. When the query fails, the risk of the corresponding non-water body contour is set to the highest value;

[0132] A risk application unit, used to determine the basic risk at each location according to the risk of all non-water body contours;

[0133] The relationship between the basic risk at each location and the risk of all non-water body contours is:.

[0134] Specifically, the relationship between the basic risk and the basic fill light frequency is:

[0135] The content of synchronously determining the image range based on the detection parameters of the night vision camera includes:

[0136] The direction vector and wide angle of the night vision camera are obtained, the viewing cone is determined according to the direction vector and the wide angle, and the image range is determined according to the viewing cone and a preset interface; the interface includes a water surface and a plane at a preset distance.

[0137] Furthermore, the image recognition module 14 includes:

[0138] A water color value extraction unit, used to extract water color values ​​from night vision images;

[0139] A ratio acquisition unit is used to extract the histogram of the night vision image and obtain the pixel ratio corresponding to each grayscale value;

[0140] A class standard deviation calculation unit, used to calculate the class standard deviation of all pixel proportions by taking the pixel proportion corresponding to the water body color value as the mean;

[0141] The class standard deviation application unit is used to determine the floating fill light frequency according to the class standard deviation.

[0142] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions and improvements made within the spirit and principles of the present invention should be included in the protection scope of the present invention.

Claims

1. An intelligent energy-saving fill-light method based on ship multi-target video guidance, characterized in that: The method comprises: Obtain a bird's-eye view of the target waters, and determine the basic risk level at each location based on the outline of static objects in the bird's-eye view; Acquire the ship's position in real time, query the basic risk level at the ship's position, and determine the basic fill light frequency according to the basic risk level; Acquire a night vision image based on a night vision camera, and synchronously determine the image range based on detection parameters of the night vision camera; the night vision camera includes a static camera installed on the shoreline and a dynamic camera installed on the ship; Identify night vision images and determine the floating fill light frequency at each position within the image range; Fill light is applied to each position based on the basic fill light frequency and the floating fill light frequency.

2. The intelligent energy-saving supplementary lighting method based on ship multi-target video guidance according to claim 1 is characterized in that: The step of obtaining a bird's-eye view of the target water area and determining the basic risk at each position based on the outline of the static object in the bird's-eye view comprises: Regularly acquire remote sensing data of the target water area, and extract a bird's-eye view of the target water area based on the remote sensing data; Performing contour recognition on the top view to determine the contour of the non-water body; According to the coordinates of the non-water body contour, the contour body and risk level are queried in the preset record library. When the query fails, the risk level of the corresponding non-water body contour is set to the highest value; Determine the basic risk at each location based on the risk of all non-water body contours; The relationship between the basic risk at each location and the risk of all non-water body contours is: In the formula, J(x,y) represents the basic risk at the position (x,y), (i k ,j k ) represents the coordinates of the intersection of the line connecting the position (x, y) and the center point of the k-th contour body and the boundary of the k-th contour body, σ is a preset parameter, and F k is the risk of the kth contour subject; K is the total number of non-water body contours.

3. The intelligent energy-saving supplementary lighting method based on ship multi-target video guidance according to claim 2 is characterized in that: The relationship between the basic risk and the basic fill light frequency is: f(x,y)=αJ(x,y)+βdJ(x,y); where f(x,y) represents the basic fill light frequency at the position (x,y), α and β are the preset correction coefficients, dJ(x,y) represents the difference of the basic risk at the position (x,y), and the four-neighborhood calculation scheme is adopted; The step of synchronously determining the image range based on the detection parameters of the night vision camera includes: Obtain the direction vector and wide angle of the night vision camera, determine the viewing cone according to the direction vector and the wide angle, and determine the image range according to the viewing cone and the preset interface; The interface includes a water surface and a plane at a preset distance.

4. The intelligent energy-saving supplementary lighting method based on ship multi-target video guidance according to claim 1 is characterized in that: The step of identifying the night vision image and determining the floating fill light frequency at each position within the image range comprises: Extract water color values ​​in night vision images; Extract the histogram of the night vision image and obtain the pixel percentage corresponding to each grayscale value; Taking the pixel percentage corresponding to the water body color value as the mean, calculating the class standard deviation of all pixel percentages; The floating fill light frequency is determined according to the class standard deviation.

5. The intelligent energy-saving supplementary lighting method based on ship multi-target video guidance according to claim 1 is characterized in that: The step of performing fill light on each position based on the basic fill light frequency and the floating fill light frequency comprises: Get the working range of the fill light device; Get the basic fill light frequency and floating fill light frequency at each position within the working range; Determine the fill light cycle according to the basic fill light frequency and the floating fill light frequency; After each fill light cycle, a fill light instruction pointing to the position is generated; Send the fill light command to the fill light device.

6. The intelligent energy-saving supplementary lighting method based on ship multi-target video guidance according to claim 1 is characterized in that: The method further comprises: At every preset time period, read the basic fill light frequency at all positions and the floating fill light frequency at the most recent moment; Add the basic fill light frequency and the most recent floating fill light frequency, and determine the image acquisition frequency at each position according to the sum; The image acquisition frequency at each location is broadcast to all ships.

7. An intelligent energy-saving fill-light system based on multi-target video guidance of ships, characterized in that: The system comprises: A location assessment module is used to obtain a bird's-eye view of the target waters and determine the basic risk level at each location based on the static object outline in the bird's-eye view; A basic frequency determination module is used to obtain the ship position in real time, query the basic risk level at the ship position, and determine the basic fill light frequency according to the basic risk level; An image acquisition module, used for acquiring night vision images based on a night vision camera, and synchronously determining the image range based on the detection parameters of the night vision camera; the night vision camera includes a static camera installed on the shoreline and a dynamic camera installed on the ship; An image recognition module is used to recognize night vision images and determine the floating fill light frequency at each position within the image range; The fill light instruction generation module is used to perform fill light on each position based on the basic fill light frequency and the floating fill light frequency.

8. The intelligent energy-saving supplementary lighting system based on ship multi-target video guidance according to claim 7 is characterized in that: The location assessment module includes: A bird's-eye view acquisition unit is used to periodically acquire remote sensing data of a target water area and extract a bird's-eye view of the target water area based on the remote sensing data; A contour recognition unit, used to perform contour recognition on the top view to determine the contour of the non-water body; A risk query unit is used to query the contour body and risk in a preset record library according to the coordinates of the non-water body contour. When the query fails, the risk of the corresponding non-water body contour is set to the highest value; A risk application unit, used to determine the basic risk at each location according to the risk of all non-water body contours; The relationship between the basic risk at each location and the risk of all non-water body contours is: In the formula, J(x,y) represents the basic risk at the position (x,y), (i k ,j k ) represents the coordinates of the intersection of the line connecting the position (x, y) and the center point of the k-th contour body and the boundary of the k-th contour body, σ is a preset parameter, and F k is the risk of the kth contour subject; K is the total number of non-water body contours.

9. The intelligent energy-saving supplementary lighting system based on ship multi-target video guidance according to claim 8 is characterized in that: The relationship between the basic risk and the basic fill light frequency is: f(x,y)=αJ(x,y)+βdJ(x,y); where f(x,y) represents the basic fill light frequency at the position (x,y), α and β are the preset correction coefficients, dJ(x,y) represents the difference of the basic risk at the position (x,y), and the four-neighborhood calculation scheme is adopted; The content of synchronously determining the image range based on the detection parameters of the night vision camera includes: Obtain the direction vector and wide angle of the night vision camera, determine the viewing cone according to the direction vector and the wide angle, and determine the image range according to the viewing cone and the preset interface; The interface includes a water surface and a plane at a preset distance.

10. The intelligent energy-saving supplementary lighting system based on ship multi-target video guidance according to claim 7, characterized in that: The image recognition module comprises: A water color value extraction unit, used to extract water color values ​​from night vision images; A ratio acquisition unit is used to extract the histogram of the night vision image and obtain the pixel ratio corresponding to each grayscale value; A class standard deviation calculation unit, used to calculate the class standard deviation of all pixel proportions by taking the pixel proportion corresponding to the water body color value as the mean; The class standard deviation application unit is used to determine the floating fill light frequency according to the class standard deviation.

Images