Illumination method based on direct current lighting system

Abstract

The application relates to the technical field of direct-current lighting, in particular to a lighting method based on a direct-current lighting system, which comprises the following steps: collecting lighting area images of each street lamp at each moment and environmental light intensity values at each moment in a rainy day; identifying umbrellas in the lighting area images to obtain each umbrella area in the lighting area images; extracting each key point of each umbrella area based on the shape features of the umbrellas; determining the detail richness and pattern contrast of each key point; constructing a feature descriptor of each key point to determine a pedestrian flow evaluation value of each street lamp at each moment; determining the light reflection intensity of each key point; dividing the lighting area images at each moment into image blocks to determine the possibility of rain filaments of each image block; screening feature image blocks in all image blocks at each moment; determining a rain intensity evaluation strength at each moment; obtaining a lighting surface brightness coefficient at each moment; and adjusting the brightness of the street lamp by using a PWM dimming technology. The application improves the flexibility of street lamp lighting brightness adjustment.

Description

Technical Field

[0001] This application relates to the field of DC lighting technology, and more specifically to lighting methods based on DC lighting systems. Background Technology

[0002] The DC lighting system converts AC power to DC power through a centralized DC drive power supply cabinet, providing a stable DC power supply for the entire lighting system. Each DC LED street light is connected in parallel, and the DC power supply provides a more stable voltage, extending the lifespan of the LED lights. It has the advantages of centralized power supply, easy management and maintenance, and energy efficiency.

[0003] Traditional lighting methods are not very flexible and cannot be effectively adjusted according to changes in the environment and pedestrian activity. Especially in rainy weather, raindrops are suspended in the air and interact with light, causing light scattering and refraction, which affects the lighting effect or causes energy waste. Summary of the Invention

[0004] To address the aforementioned technical problems, this application provides a lighting method based on a DC lighting system to solve the existing issues.

[0005] The lighting method based on a DC lighting system in this application adopts the following technical solution:

[0006] One embodiment of this application provides a lighting method based on a DC lighting system, the method comprising the following steps:

[0007] Collect images of the illuminated areas of each street light at different times during rainy weather, as well as the ambient light intensity values ​​at different times;

[0008] Umbrellas in the illuminated area image are identified to obtain each umbrella region in the illuminated area image; key points of each umbrella region are extracted based on the shape features of the umbrellas; the distribution of corner points in the local neighborhood of each key point is analyzed to determine the detail richness of each key point;

[0009] Obtain the peak points in the grayscale histogram of all pixels in the local neighborhood of each key point, and determine the pattern contrast of each key point based on the positional distribution relationship between any two peak points in the local neighborhood of each key point and the grayscale difference.

[0010] Based on the detail richness, pattern contrast, and gradient magnitude of each key point, the feature descriptor of each key point is determined. Combined with the illumination area images at adjacent time points, the movement direction of each umbrella is obtained, and the pedestrian flow assessment value of each street light at each time point is determined.

[0011] Analyze the grayscale differences between pixels in the local neighborhood of each key point and all pixels in the umbrella area, as well as the differences in pattern contrast between key points, to determine the reflectivity of each key point; obtain the optical flow direction and optical flow intensity of each pixel in the image of the illuminated area at each time moment, divide the image of the illuminated area at each time moment into image blocks, and determine the probability of rain streaks in each image block based on the distribution of optical flow direction and optical flow intensity of pixels in each image block.

[0012] Threshold segmentation is performed based on the probability of rain streaks to determine the feature image blocks in all image blocks at each time. The distribution of optical flow intensity of pixels in all feature image blocks is analyzed, and combined with the reflected light intensity, the rain intensity assessment intensity at each time is determined. The illumination brightness coefficient at each time is determined by combining the pedestrian flow assessment value, the rain intensity assessment intensity, and the ambient light intensity value at each time. The brightness of the streetlights is adjusted using PWM dimming technology.

[0013] In one embodiment, the extraction of key points for each umbrella region based on the shape features of the umbrella includes:

[0014] Obtain the perpendicular bisectors of each boundary line of the umbrella area. The pixel point where the perpendicular bisectors of the umbrella area intersect the most is determined as the umbrella tail feature point. The center pixel point of the line connecting the center pixel point of each boundary line and the umbrella tail feature point is determined as the umbrella surface feature point. The center pixel point connecting the endpoint of each boundary line and the umbrella tail pixel point is determined as the umbrella support feature point. The endpoint of each boundary line is determined as the bead tail feature point.

[0015] The umbrella tail feature point, umbrella surface feature point, umbrella support feature point, and bead tail feature point in the umbrella area are all recorded as key points.

[0016] In one embodiment, the richness of detail is the ratio of the number of corner points in the local neighborhood of each key point to the maximum number of corner points in the local neighborhood of all key points in the corresponding umbrella area.

[0017] In one embodiment, the process of determining the pattern contrast is as follows:

[0018] Within the local neighborhood of each key point, for any gray value a corresponding to any peak point and the gray value b corresponding to any remaining peak point, the probability of a pixel with gray value b in the neighborhood of a pixel with gray value a is calculated. The average of the probabilities of all pixels with gray value a in the local neighborhood of each key point is taken as the co-occurrence probability of gray values ​​a and b.

[0019] The difference between grayscale value a and grayscale value b is calculated and denoted as the first difference. The product of the first difference and the co-occurrence probability is calculated. The sum of the products of the peak point corresponding to grayscale value a and all remaining peak points is calculated. The sum of the sums of all peak points in the local neighborhood of each key point is normalized to obtain the pattern contrast.

[0020] In one embodiment, the process for determining the pedestrian flow assessment value is as follows:

[0021] Based on the feature descriptors of each key point, the matching algorithm is used to obtain the same umbrella region in the illumination area image at adjacent time points, thereby determining the movement direction of each umbrella at each time point; the illumination area image at each time point is divided into a preset number of region blocks, and the umbrellas in the region blocks near the boundary of the illumination area image whose movement direction is away from the illumination area image and biased towards the adjacent street lamp are determined as the pedestrian flow prediction value of the adjacent street lamp.

[0022] The sum of the number of umbrella areas in the illumination area image of each street light at each time and the predicted pedestrian flow value is used as the pedestrian flow assessment value.

[0023] In one embodiment, the reflectivity is calculated as follows:

[0024] In the formula, E i Let C be the reflectance intensity of the i-th key point. max C is the maximum pattern contrast of all keypoints in the image of the illumination region containing the i-th keypoint. i For the pattern contrast of the i-th key point, l i Let be the average gray value of all pixels in the local neighborhood of the i-th keypoint. Let be the average grayscale value of all pixels within the umbrella region corresponding to the i-th key point.

[0025] In one embodiment, the rain filament probability is the ratio of the normalized value of the average level of optical flow intensity of all pixels in each image block to the normalized value of the dispersion of optical flow direction of all pixels.

[0026] In one embodiment, the process for determining the intensity of the rainfall assessment is as follows:

[0027] For the illumination area images at each time point, the rain intensity assessment is the product of the mean optical flow intensity of all pixels within all feature image blocks and the mean reflectance intensity of all key points within all umbrella areas.

[0028] In one embodiment, the process of determining the illumination brightness coefficient is as follows:

[0029] For each street light, the product of the pedestrian flow assessment value and the rain intensity assessment value at each time is calculated, and the ratio of the product value to the ambient light intensity value is determined as the illumination brightness coefficient at each time.

[0030] In one embodiment, the method of adjusting the brightness of streetlights using PWM dimming technology includes:

[0031] The PWM signal correction duty cycle at each moment is determined based on the aforementioned lighting brightness coefficient, and the calculation method is as follows:

[0032] In the formula, R g，j To correct the duty cycle of the PWM signal of the g-th street light at time j, R g，j-1 To correct the duty cycle of the PWM signal of the g-th street light at time j-1, S g，j Let S be the illumination luminance coefficient of the g-th street light at time j. g，j-1 Let g be the illumination luminance coefficient of the g-th street light at time j-1.

[0033] This application has at least the following beneficial effects:

[0034] This application acquires images of the illuminated areas of streetlights at various times during rainy weather, along with ambient light intensity values. It then identifies umbrellas within these illuminated areas, obtaining individual umbrella regions. Based on the shape features of the umbrellas, it extracts key points from each umbrella region. The distribution of corner points within the local neighborhood of each key point is analyzed to determine its detail richness. Detail richness reflects the pattern complexity of the corresponding umbrella within its local neighborhood, and this feature improves the accuracy of identifying the same target umbrella at different times. Finally, it obtains the grayscale histogram of all pixels within the local neighborhood of each key point. Based on the positional distribution relationship between any two peak points in the local neighborhood of each key point, and the grayscale difference, the pattern contrast of each key point is determined. Pattern contrast reflects the brightness difference in the local neighborhood of each key point, demonstrating the clarity of details within that neighborhood and improving the accuracy of image detail extraction. Based on the detail richness, pattern contrast, and gradient magnitude of each key point, feature descriptors are determined. Combined with the illumination area images at adjacent time points, the movement direction of each umbrella is obtained, and the pedestrian flow assessment value of each street light at each time point is determined. Pedestrian flow assessment... The values ​​reflect the pedestrian traffic prediction results of each street light at each time moment, improving the reliability of street light brightness adjustment; the grayscale difference between pixels in the local neighborhood of each key point and all pixels in the umbrella area, as well as the difference between the pattern contrast of the key points, are analyzed to determine the reflectance intensity of each key point; the optical flow direction and optical flow intensity of each pixel in the illumination area image at each time moment are obtained, and the illumination area image at each time moment is divided into image blocks. Based on the distribution of optical flow direction and optical flow intensity of pixels in each image block, the probability of rain streaks in each image block is determined; both reflectance intensity and rain streak probability reflect the rainfall intensity of each street light at each time moment. The size provides a reliable reference standard for adjusting the brightness of streetlights; threshold segmentation is performed based on the probability of rain streaks to determine the feature image blocks in all image blocks at each time; the distribution of optical flow intensity of pixels within all feature image blocks is analyzed, and combined with the reflected light intensity, the rain intensity assessment intensity at each time is determined; the illumination brightness coefficient at each time is determined by combining the pedestrian flow assessment value, the rain intensity assessment intensity, and the ambient light intensity value at each time, and the brightness of the streetlights is adjusted using PWM dimming technology; while ensuring the safety of pedestrians walking in the park on rainy days, unnecessary lighting losses are reduced, resulting in energy conservation and environmental protection. Attached Figure Description

[0035] To more clearly illustrate the technical solutions and advantages in the embodiments of this application or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0036] Figure 1 The flowchart of the lighting method based on a DC lighting system provided in this application;

[0037] Figure 2 A schematic diagram of key points in the umbrella area;

[0038] Figure 3 A schematic diagram showing the division of the illumination area image;

[0039] Figure 4 Flowchart for determining the duty cycle for PWM signal correction. Detailed Implementation

[0040] To further illustrate the technical means and effects adopted by this application to achieve the intended inventive purpose, the following, in conjunction with the accompanying drawings and preferred embodiments, details the specific implementation, structure, features, and effects of the lighting method based on a DC lighting system proposed in this application. In the following description, different "one embodiment" or "another embodiment" do not necessarily refer to the same embodiment. Furthermore, specific features, structures, or characteristics in one or more embodiments can be combined in any suitable form.

[0041] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this application pertains.

[0042] The specific scheme of the lighting method based on the DC lighting system provided in this application is described in detail below with reference to the accompanying drawings.

[0043] This application provides a lighting method based on a DC lighting system in one embodiment. Specifically, the following lighting method based on a DC lighting system is provided. Please refer to [link / reference]. Figure 1 The method includes the following steps:

[0044] Step S001: Collect images of the illuminated areas of each street light at each time point during rainy weather, as well as the ambient light intensity values ​​at each time point, and perform preprocessing.

[0045] In this embodiment, a data acquisition device is installed on each LED street light. The data acquisition device includes a photosensitive sensor and a camera. The photosensitive sensor collects the ambient light intensity value at each moment and records it as the ambient light intensity value at each moment. The camera collects images of the illuminated area on the ground illuminated by the LED street light at each moment. At the same time, in order to reduce the energy consumption and computing resources of the data acquisition device during high-frequency operation, the data acquisition interval of the data acquisition device in this embodiment is set to 1 second. The implementer can set it according to the actual situation. This embodiment does not impose any restrictions.

[0046] This embodiment utilizes a bilateral filtering algorithm to denoise the acquired lighting area images, eliminating noise interference during camera acquisition. The denoised lighting area images are then subjected to histogram equalization to adjust the histogram distribution, thereby enhancing the contrast and improving the visual effect of the lighting area images. This facilitates subsequent analysis of rainfall conditions and pedestrian activity in the park. Both the bilateral filtering algorithm and histogram equalization are well-known techniques, and their specific processes will not be elaborated upon. Implementers can choose other feasible filtering and image enhancement algorithms based on their specific circumstances. Finally, the lighting area images at each time point are converted into grayscale images for subsequent analysis.

[0047] Step S002: Identify the umbrellas in the illuminated area image to obtain each umbrella region in the illuminated area image; extract each key point of each umbrella region based on the shape features of the umbrellas; analyze the distribution of corner points in the local neighborhood of each key point to determine the detail richness of each key point.

[0048] Park pedestrian traffic at night is often highly random, placing extremely high demands on the flexibility of the lighting system. In rainy park scenarios, traditional lighting methods cannot effectively and energy-efficiently control LED streetlights. However, it is essential to adaptively adjust the brightness of LED streetlights according to rainfall conditions and pedestrian activity in the park, which can reduce unnecessary lighting losses while ensuring the safety of pedestrians in the park on rainy days.

[0049] In park scenarios, rainy weather is common, and pedestrians often use umbrellas to visit the park. Cameras mounted on LED streetlights are difficult to use for direct visual detection of pedestrians due to the obstruction caused by umbrellas. Therefore, a pre-built recognition model is used to identify umbrellas in the illuminated area image and extract their bounding boxes. In this embodiment, the recognition model is Mask R-CNN, a well-known existing technology. Implementers can choose other feasible recognition models, such as Faster R-CNN or YOlOv5, depending on the specific circumstances; this embodiment does not impose any limitations.

[0050] In a rainy nighttime park scene, multiple umbrellas may appear simultaneously in an image of an illuminated area, meaning there are multiple umbrella bounding boxes. The movement trajectory of each umbrella can represent the movement trajectory of a pedestrian in the park. Therefore, it is necessary to analyze each umbrella in the image of the illuminated area.

[0051] In the structure of an umbrella, the canopy is typically supported by multiple ribs. The area between any two adjacent ribs is called the canopy sector, and the end of the canopy sector is the edge. In the illuminated area image, the edge is a smooth straight line close to the umbrella's bounding box. This embodiment extracts multiple umbrella regions from the illuminated area image based on the umbrella bounding box coordinates. For any umbrella region, the Canny edge detection algorithm is applied to obtain an edge binarization map, which is then used as input to a contour finding algorithm. Specifically, in this embodiment, the contour finding algorithm uses the `findContours` function in OpenCV to output the outer contour of the umbrella region. This outer contour is then used as input to the Hough line detection algorithm, which outputs each edge of the umbrella region, denoted as a boundary line. Both the Canny edge detection algorithm and the contour finding algorithm are existing known technologies and will not be described in detail here.

[0052] The umbrella area corresponds to the umbrella structure, including the canopy, ribs, tail, and beaded tail. The umbrella support is the joint point in the middle of the ribs that allows the umbrella to fold and open. This embodiment first obtains the perpendicular bisectors of each boundary line of the umbrella area. The pixel point where the perpendicular bisectors intersect most frequently is determined as the tail feature point. The center pixel point of the line connecting the center pixel of each boundary line to the tail feature point is determined as the canopy feature point. The center pixel point connecting the endpoints of each boundary line to the tail pixel point is determined as the umbrella support feature point. The endpoints of each boundary line are determined as the beaded tail feature points. All tail, canopy, support, and beaded tail feature points in each umbrella area are recorded as key points. A schematic diagram of the key points of the umbrella area is shown below. Figure 2 As shown, Figure 2 The pentagonal symbol represents the umbrella tail feature point, the rhombus symbol represents the umbrella surface feature point, the hexagonal symbol represents the umbrella support feature point, and the circular symbol represents the bead tail feature point.

[0053] Since the umbrellas used by pedestrians on rainy days are mostly different, and each umbrella surface often has a unique texture, making them highly distinguishable, and since the image area outside the outer contour of the umbrella area in the illuminated area image is usually the background part such as roads and park landscapes, which changes with the activities of pedestrians in the park and is not constant, this embodiment sets a 7*7 local window centered on each key point, and takes the image area within the outer contour of the umbrella area in the local window as the local neighborhood of each key point. The size of the local window can be set by the implementer according to the actual situation, and this embodiment does not impose any restrictions.

[0054] The Harris corner detection algorithm takes the illuminated area image as input and outputs all corners on the illuminated area image. Corners represent structural details of the image. The algorithm counts the number of corners in the local neighborhood of each keypoint within each umbrella region, obtains the maximum value of this corner count for all keypoints within each umbrella region, and uses the ratio of the number of corners in the local neighborhood of each keypoint within each umbrella region to this maximum value as the detail richness of each keypoint within that umbrella region. The Harris corner detection algorithm is a well-known existing technology, and its specific process will not be described in detail.

[0055] The more corner points there are, the richer the details of the umbrella surface pattern structure in the local neighborhood of the key points of the umbrella, and therefore, the greater the richness of detail.

[0056] Step S003: Obtain the peak points in the grayscale histogram of all pixels in the local neighborhood of each key point. Based on the positional distribution relationship between any two peak points in the local neighborhood of each key point and the grayscale difference, determine the pattern contrast of each key point.

[0057] The grayscale histograms of all pixels within the local neighborhood of each key point within each umbrella area are obtained. These grayscale histograms are then used as input to the Automatic multiscale-based peak detection (AMPD) algorithm, which outputs peak points in the grayscale histograms, each peak point corresponding to a grayscale value. In this embodiment, taking the grayscale values ​​a and b corresponding to peak points x and y as examples, the probability η of a pixel with grayscale value a having a grayscale value of b within its eight-neighborhood is calculated within the local neighborhood of the key point. a，b Each pixel with a grayscale value of 'a' has a probability η. a，b Calculate the probability η corresponding to all pixels with a gray value of a. a，b The mean of the values ​​is denoted as the co-occurrence probability of gray values ​​a and b. Wherein, probability η... a，b Specifically, it is the ratio of the number of pixels with a gray value of b in the eight-neighborhood to the total number of pixels in the eight-neighborhood. The AMPD algorithm is a well-known existing technology, and its specific process will not be described in detail.

[0058] The difference between grayscale value a and grayscale value b is calculated and denoted as the first difference. The product of the first difference and the co-occurrence probability is calculated. The sum of the products of peak point x and all remaining peak points in the local neighborhood is calculated. The sum of the sums of all peak points in the local neighborhood of each key point is normalized. The normalized result is used as the pattern contrast of each key point in each umbrella area.

[0059] It should be noted that the difference represents the degree of difference between two variables. Specifically, it can be calculated using the absolute value of the difference, the square of the difference, the ratio, etc. In this embodiment, the absolute value of the difference is used as the calculation method. In the process of calculating the pattern contrast, the normalization result is obtained by using the Sigmoid function. Implementers can choose other feasible normalization methods according to the actual situation. This embodiment does not impose any restrictions here.

[0060] It should be understood that the more times gray values ​​a and b appear adjacent to each other in the local neighborhood of a key point, i.e., the greater the co-occurrence probability, the more frequent the transition between gray values ​​a and b is, and the greater the first difference is, the more significant the details of the umbrella surface pattern in the local neighborhood of the key point of the umbrella, and the greater the pattern contrast.

[0061] Step S004: Based on the detail richness, pattern contrast, and gradient magnitude of each key point, determine the feature descriptor of each key point, combine the illumination area images at adjacent time points, obtain the movement direction of each umbrella, and determine the pedestrian flow assessment value of each street light at each time point.

[0062] This embodiment utilizes the Sobel operator to obtain the gradient magnitude of each keypoint. The detail richness, pattern contrast, and gradient magnitude of each keypoint within each umbrella region are sequentially combined into a vector, serving as the feature descriptor for each keypoint. The use of the Sobel operator to obtain the gradient magnitude is a well-known existing technique; implementers can choose other feasible gradient calculation algorithms as needed.

[0063] The illumination area image of the previous acquisition time is acquired at each time step. Using the same method, each umbrella region, each key point within the umbrella region, and the feature descriptor of each key point in the illumination area image of the previous acquisition time are obtained. The key points and their feature descriptors from the illumination area images of each time step and the previous time step are used as input to the FLANN (Fast Library for Approximate Nearest Neighbors) fast nearest neighbor search algorithm. The parameters `trees` and `checks` in the FLANN algorithm are set to a constant value of 5, and the output is the feature matching result of the key points. If the number of feature matching points for the umbrella region in the illumination area image of each time step and the illumination area image of the previous time step is greater than a first threshold, then the two umbrella regions are determined to be the same target umbrella at different times. The first threshold is 80% of the total number of key points in the corresponding umbrella region in the illumination area image of each time step. The implementer can set the first threshold according to the actual situation; this embodiment does not impose any restrictions. The FLANN algorithm is a well-known existing technology, and its specific process will not be described in detail.

[0064] In a rainy park scene, the greater the pedestrian traffic, the brighter the streetlights should be to ensure pedestrians can clearly see potential dangers and prevent accidents. This embodiment uses umbrella areas to represent pedestrians, recording the center coordinates of the same target umbrella within its bounding box at each moment and the previous moment's umbrella bounding box. Pointing the previous moment's center coordinates to the current moment's center coordinates yields the target umbrella's movement direction. It's important to note that if an umbrella area exists at a given moment but didn't have a corresponding umbrella area at the previous moment, the umbrella's movement direction is set to the horizontal direction relative to the image boundary. Following these steps, the movement direction of each umbrella area in the illuminated area image at each moment is obtained.

[0065] In assessing pedestrian traffic, in addition to the number of umbrella areas in the illuminated area image at each time point, it is also necessary to predictively consider whether new pedestrians or umbrellas will subsequently participate. This application divides the illuminated area image under each LED streetlight into three horizontally equal blocks: the left-hand pedestrian area, the middle pedestrian area, and the right-hand pedestrian area, from left to right. A schematic diagram of the illuminated area image division is shown below. Figure 3 As shown, Figure 3 In the diagram, 301 represents the left-hand area, 302 the middle area, and 303 the right-hand area. The triangular symbol indicates the center coordinates within the umbrella area's bounding box, and the straight arrow of the triangular symbol indicates the direction of movement for the umbrella area.

[0066] When the center coordinates of the umbrella bounding box of the g-th LED streetlight at each moment fall within the left-hand zone and the movement direction is also left, it indicates that the pedestrian is highly likely to move into the LED streetlight illumination area to the left of the g-th LED streetlight. If the movement direction is right, it indicates that the pedestrian is highly likely to remain within the current LED streetlight's area at the next moment. The number of umbrella areas with left-hand movement within the left-hand zone is counted and used as the predicted pedestrian flow value for the adjacent LED streetlight to the left of the g-th LED streetlight.

[0067] When the center coordinates within the umbrella's bounding box fall in the middle zone, and the movement direction is left or right, the pedestrian is highly likely to remain within the illumination area of ​​this LED streetlight for a short period. When the center coordinates within the umbrella's bounding box fall in the right-hand zone, and the movement direction is also right, it indicates that the pedestrian is highly likely to move into the illumination area of ​​the LED streetlight to the right of the g-th LED streetlight. If the movement direction is left, it indicates that the pedestrian is highly likely to remain within the area illuminated by this LED streetlight in the next moment. The number of umbrella zones with right-hand movement within the right-hand zone is counted and used as the predicted pedestrian flow value for the adjacent LED streetlight to the right of the g-th LED streetlight.

[0068] It should be noted that the DC lighting system can manage all the LED streetlights in the park and number them according to their position. Taking the g-th LED streetlight as an example, the g-1-th LED streetlight is the left adjacent LED streetlight of the g-th LED streetlight, and the g+1-th LED streetlight is the right adjacent LED streetlight of the g-th LED streetlight.

[0069] The number of umbrella areas in the illumination area image of each LED street light at each time moment is counted as the pedestrian flow value. The sum of the pedestrian flow value of each LED street light at each time moment and all the predicted pedestrian flow values ​​is calculated as the pedestrian flow assessment value of each LED street light at each time moment. The larger the pedestrian flow assessment value, the more pedestrians there are in the LED street light illumination area at each time moment and in subsequent times. For pedestrian safety, the brightness of the LED street lights should be increased.

[0070] Step S005: Analyze the grayscale difference between pixels in the local neighborhood of each key point and all pixels in the umbrella area, as well as the difference between the pattern contrast of the key points, to determine the reflectivity of each key point; obtain the optical flow direction and optical flow intensity of each pixel in the image of the illuminated area at each time, divide the image of the illuminated area at each time into image blocks, and determine the probability of rain streaks in each image block based on the distribution of the optical flow direction and optical flow intensity of the pixels in each image block.

[0071] In a rainy nighttime park scene, the intensity of the rain affects the lighting effect of LED streetlights. It's necessary to assess the rain intensity in the park and adjust the brightness of the LED streetlights accordingly. Some rainwater remains on the umbrella surface, forming water stains. Under the illumination of the LED streetlights, the smooth surface of the umbrella makes light reflect more easily, creating strong reflective areas. Within these reflective areas of the water stains, the texture blurring and brightness are high. The reflective intensity of key points within each umbrella area is calculated using the following method:

[0072] In the formula, E i Let C be the reflectance intensity of the i-th key point. max C is the maximum pattern contrast of all keypoints in the image of the illumination region containing the i-th keypoint. i For the pattern contrast of the i-th key point, l i Let be the average gray value of all pixels in the local neighborhood of the i-th keypoint. Let be the average grayscale value of all pixels within the umbrella region corresponding to the i-th key point.

[0073] It should be understood that when (C) max -C i The larger the value, the lower the saliency of the edge within the local neighborhood of the umbrella's key point, and the higher the texture blur. Meanwhile, when... The larger the value, the higher the brightness of the pixels in the local neighborhood of the key point on the umbrella. This means that the local neighborhood of the key point on the umbrella is more likely to be a reflective area of ​​water stains on the umbrella surface, and the greater the reflective intensity of the water stains on the umbrella surface.

[0074] When it rains lightly in a park at night, rainwater is more likely to remain on the umbrella surface, forming reflective areas of water stains. Therefore, the intensity of reflection can reflect the amount of rainfall when the rain is light. When the rain is heavy, raindrops are usually larger and denser, with a stronger ability to reflect and refract light, making it easier to observe the phenomenon of rain streaks with a significant movement trend. This embodiment acquires images of the illuminated areas of each street lamp at each time and the time before it, and uses the RAFT (Recurrent All-Pairs FieldTransforms for Optical Flow) model to estimate optical flow, outputting the optical flow direction and intensity of each pixel in the illuminated area image of each street lamp at each time. The illuminated area image is divided into 256 image blocks. The RAFT model is a well-known existing technology, and the specific process is not described in detail. The probability of rain streaks in each image block is calculated, and the specific calculation method is as follows:

[0075] In the formula, τ is the probability of rain streaks in each image patch, μ is the mean optical flow intensity of all pixels in each image patch, and σ is the mean value of the optical flow intensity of all pixels in each image patch. 2 This represents the degree of dispersion of the optical flow direction of all pixels within each image block, with Norm() being the normalization function. The degree of dispersion can be calculated using methods such as variance, standard deviation, and coefficient of variation. This embodiment uses variance as the method for calculating the degree of dispersion.

[0076] It should be understood that when μ is larger, it indicates that the image patch is not a static region in the illuminated area image, but has a significant motion trend, while when σ 2 The smaller the value, the more concentrated the movement direction of the pixels within the image block, the more likely the image block is to contain raindrop pixels, and the higher the probability of raindrops τ.

[0077] Step S006: Based on the probability of rain streaks, threshold segmentation is performed to determine the feature image blocks in all image blocks at each time; the distribution of optical flow intensity of pixels in all feature image blocks is analyzed, and combined with the reflected light intensity, the rain intensity assessment intensity at each time is determined; the illumination brightness coefficient at each time is determined by combining the pedestrian flow assessment value, the rain intensity assessment intensity, and the ambient light intensity value at each time, and the brightness of the streetlights is adjusted using PWM dimming technology.

[0078] The probability of rain streaks is obtained for all image patches in the illuminated area images of each street light at each time step, and used as input to the Ostu algorithm. The output is a segmentation threshold. Image patches with a rain streak probability greater than the segmentation threshold are recorded as feature image patches, which are highly likely to contain a large number of rain streak pixels. The rainfall intensity assessment for each street light at each time step is calculated as follows:

[0079] In the formula, P represents the estimated rainfall intensity for each street light at each time point. V is the mean optical flow intensity of all pixels within all feature image blocks in the illumination area image of each street light at each time, and V is the mean reflectance intensity of all key points within all umbrella areas in the illumination area image of each street light at each time.

[0080] It should be understood that in light rain, a larger V indicates a greater reflectivity of the water stains on the umbrella surface, suggesting a heavier rainfall. Conversely, in heavy rain, when... The larger the value, the more significant the movement trend of the raindrop pixels, the greater the speed of the raindrops, and the greater the rainfall. Therefore, the greater the rainfall intensity assessment P, the higher the brightness of the LED streetlights should be to ensure the safety of pedestrians in the park on rainy days.

[0081] Based on pedestrian traffic, rainfall, and ambient light information at various times within the LED streetlight illumination area, the illumination brightness coefficient of the LED streetlights is obtained. Specifically, for each streetlight, the product of the pedestrian traffic assessment value and the rainfall intensity assessment value at each time is calculated. The ratio of the product value to the ambient light intensity value is determined as the illumination brightness coefficient at each time. The larger the illumination brightness coefficient, the higher the brightness of the LED streetlights should be to ensure the safety of pedestrians in the park on rainy days.

[0082] LED streetlights use PWM (Pulse Width Modulation) dimming technology for brightness adjustment. The duty cycle of the PWM signal for each streetlight at each time moment is determined based on the illumination brightness coefficient of each streetlight at each time moment. The specific calculation method is as follows:

[0083] In the formula, R g，j To correct the duty cycle of the PWM signal of the g-th street light at time j, R g，j-1 To correct the duty cycle of the PWM signal of the g-th street light at time j-1, S g，j Let S be the illumination luminance coefficient of the g-th street light at time j. g，j-1 Let g be the illumination brightness coefficient of the g-th street light at time j-1. The flowchart for determining the duty cycle correction using the PWM signal is shown below. Figure 4 As shown. It should be noted that, to ensure the calculation of the duty cycle correction for the PWM signal, the brightness of the LED streetlights is not adjusted at the initial moment.

[0084] It should be understood that when Furthermore, the higher the value, the more intense the rainfall, the greater the pedestrian traffic, and the weaker the ambient light in the park. To ensure the safety of pedestrians in the park during rainy days, the brightness of the LED streetlights should be increased. Furthermore, the smaller the value, the lower the brightness of the LED street light should be to achieve energy-saving control of the LED street light by using a lower PWM signal duty cycle.

[0085] At this point, the PWM signal correction duty cycle of each LED street light is obtained at the current moment. Through the single-lamp controller, the PWM signal correction duty cycle is used as the duty cycle of PWM dimming technology to adjust the brightness of each LED street light. This reduces unnecessary lighting loss and saves energy while ensuring the safety of pedestrians in the park on rainy days.

[0086] It should be noted that the order of the embodiments described above is merely for descriptive purposes and does not represent the superiority or inferiority of the embodiments. Furthermore, specific embodiments of this specification have been described above. Additionally, the processes depicted in the accompanying drawings do not necessarily require a specific or sequential order to achieve the desired results. In some implementations, multitasking and parallel processing are possible or may be advantageous.

[0087] The various embodiments in this specification are described in a progressive manner. The same or similar parts between the various embodiments can be referred to each other. Each embodiment focuses on describing the differences from other embodiments.

[0088] The above-described embodiments are only used to illustrate the technical solutions of this application, and are not intended to limit them; modifications to the technical solutions described in the foregoing embodiments, or equivalent substitutions of some of the technical features, do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of this application, and should all be included within the protection scope of this application.

Claims

1. A lighting method based on a DC lighting system, characterized in that, The method includes the following steps: Collect images of the illuminated areas of each street light at different times during rainy weather, as well as the ambient light intensity values ​​at different times; Umbrellas in the illuminated area image are identified to obtain each umbrella region in the illuminated area image; key points of each umbrella region are extracted based on the shape features of the umbrellas; the distribution of corner points in the local neighborhood of each key point is analyzed to determine the detail richness of each key point; Obtain the peak points in the grayscale histogram of all pixels in the local neighborhood of each key point, and determine the pattern contrast of each key point based on the positional distribution relationship between any two peak points in the local neighborhood of each key point and the grayscale difference. Based on the detail richness, pattern contrast, and gradient magnitude of each key point, the feature descriptor of each key point is determined. Combined with the illumination area images at adjacent time points, the movement direction of each umbrella is obtained, and the pedestrian flow assessment value of each street light at each time point is determined. Analyze the grayscale differences between pixels in the local neighborhood of each key point and all pixels in the umbrella area, as well as the differences in pattern contrast between key points, to determine the reflectivity of each key point; obtain the optical flow direction and optical flow intensity of each pixel in the image of the illuminated area at each time moment, divide the image of the illuminated area at each time moment into image blocks, and determine the probability of rain streaks in each image block based on the distribution of optical flow direction and optical flow intensity of pixels in each image block. Threshold segmentation is performed based on the probability of rain streaks to determine the feature image blocks in all image blocks at each time. The distribution of optical flow intensity of pixels in all feature image blocks is analyzed, and combined with the reflective intensity, the rain intensity assessment intensity at each time is determined. The lighting brightness coefficient at each time is determined by combining the pedestrian flow assessment value, rain intensity assessment intensity, and ambient light intensity value at each time. The brightness of the street lights is adjusted using PWM dimming technology. The process of determining the lighting luminance coefficient is as follows: For each street light, calculate the product of the pedestrian flow assessment value and the rain intensity assessment value at each time, and determine the ratio of the product value to the ambient light intensity value as the lighting luminance coefficient at each time. The brightness of streetlights is adjusted using PWM dimming technology, including: determining the duty cycle of the PWM signal at each moment based on the lighting brightness coefficient, calculated as follows: In the formula, To correct the duty cycle of the PWM signal of the g-th street light at time j, To correct the duty cycle of the PWM signal of the g-th street light at time j-1, Let g be the illumination luminance coefficient of the g-th street light at time j. Let g be the illumination brightness coefficient of the g-th street light at time j-1; initially, the brightness of the LED street light is not adjusted. Based on the shape features of the umbrella, key points of each umbrella region are extracted, including: obtaining the perpendicular bisectors of each boundary line of the umbrella region; identifying the pixel point where the perpendicular bisectors of the umbrella region intersect the most as the umbrella tail feature point; identifying the center pixel of the line connecting the center pixel of each boundary line and the umbrella tail feature point as the umbrella surface feature point; identifying the center pixel of the line connecting the endpoint of each boundary line and the umbrella tail pixel as the umbrella support feature point; and identifying the endpoint of each boundary line as the bead tail feature point. The umbrella tail feature point, umbrella surface feature point, umbrella support feature point, and bead tail feature point of the umbrella region are all recorded as key points. The method for calculating reflectivity is as follows: In the formula, Let i be the reflectance of the i-th key point. Let be the maximum pattern contrast of all keypoints in the image of the illumination region containing the i-th keypoint. The pattern contrast at the i-th key point, Let be the average gray value of all pixels in the local neighborhood of the i-th keypoint. The average grayscale value of all pixels within the umbrella area corresponding to the i-th key point; The process for determining the intensity of rainstorm assessment is as follows: For the illuminated area image at each time point, the intensity of rainstorm assessment is the product of the mean optical flow intensity of all pixels in all feature image blocks and the mean reflectance intensity of all key points in all umbrella areas.

2. The lighting method based on a DC lighting system as described in claim 1, characterized in that, The detail richness is the ratio of the number of corner points in the local neighborhood of each key point to the maximum number of corner points in the local neighborhood of all key points in the corresponding umbrella area.

3. The lighting method based on a DC lighting system as described in claim 1, characterized in that, The process for determining pattern contrast is as follows: Within the local neighborhood of each key point, for any gray value a corresponding to any peak point and the gray value b corresponding to any remaining peak point, the probability of a pixel with gray value b in the neighborhood of a pixel with gray value a is calculated. The average of the probabilities of all pixels with gray value a in the local neighborhood of each key point is taken as the co-occurrence probability of gray values ​​a and b. The difference between grayscale value a and grayscale value b is calculated and denoted as the first difference. The product of the first difference and the co-occurrence probability is calculated. The sum of the products of the peak point corresponding to grayscale value a and all the remaining peak points is calculated. The sum of the sums of all peak points in the local neighborhood of each key point is normalized to obtain the pattern contrast.

4. The lighting method based on a DC lighting system as described in claim 1, characterized in that, The process for determining pedestrian flow assessment values ​​is as follows: Based on the feature descriptors of each key point, the matching algorithm is used to obtain the same umbrella region in the illumination area image at adjacent time points, thereby determining the movement direction of each umbrella at each time point; the illumination area image at each time point is divided into a preset number of region blocks, and the umbrellas in the region blocks near the boundary of the illumination area image whose movement direction is away from the illumination area image and biased towards the adjacent street lamp are determined as the pedestrian flow prediction value of the adjacent street lamp. The sum of the number of umbrella areas in the illumination area image of each street light at each time point and the predicted pedestrian flow value is used as the pedestrian flow assessment value.

5. The lighting method based on a DC lighting system as described in claim 1, characterized in that, The probability of rain streaks is the ratio of the normalized value of the average level of optical flow intensity of all pixels in each image block to the normalized value of the dispersion of optical flow direction of all pixels.

Images