Three-dimensional modeling method for point cloud and image data of distribution network non-power-cut operation site

Abstract

The invention belongs to the technical field of distribution networks, and particularly relates to a distribution network uninterruptible power operation site point cloud and image data three-dimensional modeling method, which comprises the following specific steps of: acquiring an image and point cloud data thereof acquired by an unmanned aerial vehicle through an image acquisition module; the point cloud data in the image acquisition module is acquired through the point cloud acquisition module, and after acquisition, a three-dimensional model is constructed through the first construction module according to the point cloud data; according to the invention, the point cloud data and the image data are respectively utilized to construct a three-dimensional model, and the three-dimensional models are subjected to coincidence detection, so that the accuracy of the three-dimensional model can be greatly improved, the distribution network is simulated by establishing the three-dimensional model, and the accuracy of the distribution network is improved. Therefore, the technicians can observe the working environment in real time, so that the technicians can accurately understand the working environment, and the non-power-cut working scheme design and the on-site implementation safety strategy are optimized.

Description

Technical Field

[0001] The present invention relates to the technical field of distribution network, and in particular to a method for three-dimensional modeling of point cloud and image data at a distribution network uninterrupted power supply operation site. Background Art

[0002] The distribution network refers to the power grid that receives electric energy from the transmission network or regional power plants and distributes it locally or step by step according to voltage to various users through distribution facilities. It is composed of overhead lines, cables, towers, distribution transformers, disconnectors, reactive power compensators and some ancillary facilities, and plays an important role in distributing electric energy in the power grid.

[0003] At present, in the process of power distribution, it is difficult for technicians to accurately understand the working environment, which makes it difficult for technicians to effectively design non-stop power supply operation plans and implement safety strategies on site. There are limitations. The reason is that it is difficult for technicians to observe the working environment in real time. Therefore, a three-dimensional modeling method of point cloud and image data of non-stop power supply operation site of power distribution network is invented. Summary of the invention

[0004] In view of the above-mentioned and / or existing problems in the existing methods for three-dimensional modeling of point cloud and image data at the site of non-stop power distribution network operation, the present invention is proposed.

[0005] Therefore, the purpose of the present invention is to provide a three-dimensional modeling method of point cloud and image data at the distribution network non-stop operation site, which can solve the above-mentioned existing problems.

[0006] To solve the above technical problems, according to one aspect of the present invention, the present invention provides the following technical solutions:

[0007] A method for three-dimensional modeling of point cloud and image data of a distribution network non-stop operation site includes the following specific steps:

[0008] Step 1: Obtain the images and point cloud data collected by the drone through the image acquisition module;

[0009] Step 2: The point cloud data in the image acquisition module is acquired through the point cloud acquisition module. After acquisition, a three-dimensional model is constructed according to the point cloud data through the first construction module;

[0010] Step 3: The image collected by the image acquisition module is processed by the image preprocessing module to adjust the light and dark parts in the image so that the objects in the image become clear. After processing, the processed image is cut by the image cutting module to cut the processed image into several independent images. After cutting, the independent images cut out are subjected to feature extraction by the feature extraction module. After extraction, the extracted feature shape is judged by the shape judgment module. After judgment, the extracted feature size is calculated by the size calculation module to obtain the feature size. After that, the three-dimensional model is constructed according to the features, shape and size of the image by the second construction module.

[0011] Step 4: The 3D model constructed by the first 3D module is fused with the 3D model constructed by the second 3D module through the fusion module. After the fusion, the fused 3D model is analyzed by the analysis module to find out whether there are abnormal points. If there are abnormal points, it means that there are differences between the two sets of 3D models. At this time, it is necessary to start the construction again;

[0012] Step 5: If there are no abnormal points, the color of the image in the image acquisition module will be analyzed by the color analysis module, and the texture in the image acquisition module will be analyzed by the texture analysis module. After that, the fused three-dimensional model will be optimized based on the analyzed color and texture through the fourth construction module. At this point, the technicians can accurately understand the working environment, optimize the design of non-stop operation plans and implement safety strategies on site.

[0013] As a preferred solution of the method for three-dimensional modeling of point cloud and image data of distribution network uninterrupted power supply operation site described in the present invention, it also includes a three-dimensional modeling system, and the three-dimensional modeling system includes:

[0014] Image acquisition module, used to obtain images and point cloud data collected by drones;

[0015] The first 3D module builds a 3D model based on point cloud data;

[0016] The second 3D module builds a 3D model based on the image;

[0017] A fusion module, used to fuse the three-dimensional model constructed by the first three-dimensional module with the three-dimensional model constructed by the second three-dimensional module;

[0018] An analysis module is used to analyze the fused three-dimensional model to determine whether there are any abnormal points;

[0019] The optimization module is used to optimize the fused 3D model.

[0020] As a preferred scheme of the three-dimensional modeling method of point cloud and image data at the distribution network non-stop operation site described in the present invention, wherein: the image acquisition module is connected to the first three-dimensional module, the image acquisition module is connected to the second three-dimensional module, the first three-dimensional module is connected to the fusion module, the second three-dimensional module is connected to the fusion module, the fusion module is connected to the analysis module, and the analysis module is connected to the optimization module.

[0021] As a preferred solution of the method for three-dimensional modeling of point cloud and image data of distribution network non-stop operation site of the present invention, wherein: the first three-dimensional module includes:

[0022] A point cloud acquisition module is used to acquire point cloud data in the image acquisition module;

[0023] The first construction module is used to construct a three-dimensional model according to point cloud data.

[0024] As a preferred solution of the method for three-dimensional modeling of point cloud and image data at the distribution network non-stop operation site described in the present invention, wherein: the image acquisition module is connected to the point cloud acquisition module, the point cloud acquisition module is connected to the first construction module, and the first construction module is connected to the fusion module.

[0025] As a preferred solution of the method for three-dimensional modeling of point cloud and image data of distribution network non-stop operation site of the present invention, the second three-dimensional module includes:

[0026] An image preprocessing module, used for processing the images collected by the image acquisition module;

[0027] An image cutting module is used to cut the processed image into a plurality of independent images;

[0028] A feature extraction module, used for extracting features from the cut independent images;

[0029] A shape judgment module is used to judge the extracted characteristic shape;

[0030] A size calculation module is used to calculate the extracted feature size to obtain the feature size;

[0031] The second construction module is used to construct a three-dimensional model according to the characteristics, shape and size of the image.

[0032] As a preferred solution of the method for three-dimensional modeling of point cloud and image data at the distribution network non-stop operation site described in the present invention, wherein: the image acquisition module is connected to the image preprocessing module, the image preprocessing module is connected to the image cutting module, the image cutting module is connected to the feature extraction module, the feature extraction module is connected to the shape judgment module, the shape judgment module is connected to the size calculation module, the size calculation module is connected to the second construction module, and the second construction module is connected to the fusion module.

[0033] As a preferred solution of the method for three-dimensional modeling of point cloud and image data of distribution network non-stop operation site described in the present invention, the optimization module includes:

[0034] A color analysis module, used to analyze the color of the image in the image acquisition module;

[0035] A texture analysis module, used for analyzing the texture in the image acquisition module;

[0036] The fourth building module optimizes the fused three-dimensional model based on the analyzed color and texture.

[0037] As a preferred solution of the method for three-dimensional modeling of point cloud and image data of distribution network non-stop operation site described in the present invention, wherein: the analysis module is connected to the color analysis module, the color analysis module is connected to the texture analysis module, and the texture analysis module is connected to the fourth construction module.

[0038] As a preferred solution of the method for three-dimensional modeling of point cloud and image data of distribution network non-stop operation site described in the present invention, the specific workflow of the image preprocessing module is as follows:

[0039] Process 1: Input the scene image, and calculate the temporal fitness model of rods and cones respectively. The bilateral filtering method is used to simulate the interaction between photoreceptor cells and surrounding cells after receiving light stimulation.

[0040] Process 2: Calculate the time-varying perceptual response values ​​of rods and cones in the process;

[0041] Process 3: Use the inverse responsivity transformation module to transform and map its high dynamic image onto a low dynamic range observation display device;

[0042] Process 4: By changing the color adaptation matrix, the corrected scene images perceived by the human eye at different times are obtained.

[0043] Compared with existing technologies:

[0044] By using point cloud data and image data to construct a three-dimensional model respectively and performing overlap detection, the accuracy of the three-dimensional model can be greatly improved. By building a three-dimensional model to simulate the distribution network, technical personnel can observe the working environment in real time, thereby enabling technical personnel to accurately understand the working environment, optimize the design of non-stop operation plans and implement safety strategies on site. BRIEF DESCRIPTION OF THE DRAWINGS

[0045] Figure 1 It is a schematic diagram of the process of the present invention. DETAILED DESCRIPTION

[0046] In order to make the objectives, technical solutions and advantages of the present invention more clear, the embodiments of the present invention will be further described in detail below with reference to the accompanying drawings.

[0047] The present invention provides a method for three-dimensional modeling of point cloud and image data at the scene of distribution network non-stop operation. Figure 1 , including the following specific steps:

[0048] Step 1: Obtain the images and point cloud data collected by the drone through the image acquisition module;

[0049] Step 2: The point cloud data in the image acquisition module is acquired through the point cloud acquisition module. After acquisition, a three-dimensional model is constructed according to the point cloud data through the first construction module;

[0050] Step 3: The image collected by the image acquisition module is processed by the image preprocessing module to adjust the light and dark parts in the image so that the objects in the image become clear. After processing, the processed image is cut by the image cutting module to cut the processed image into several independent images. After cutting, the independent images cut out are subjected to feature extraction by the feature extraction module. After extraction, the extracted feature shape is judged by the shape judgment module. After judgment, the extracted feature size is calculated by the size calculation module to obtain the feature size. After that, the three-dimensional model is constructed according to the features, shape and size of the image by the second construction module.

[0051] Step 4: The 3D model constructed by the first 3D module is fused with the 3D model constructed by the second 3D module through the fusion module. After the fusion, the fused 3D model is analyzed by the analysis module to find out whether there are abnormal points. If there are abnormal points, it means that there are differences between the two sets of 3D models. At this time, it is necessary to start the construction again;

[0052] Step 5: If there are no abnormal points, the color of the image in the image acquisition module will be analyzed by the color analysis module, and the texture in the image acquisition module will be analyzed by the texture analysis module. After that, the fused three-dimensional model will be optimized based on the analyzed color and texture through the fourth construction module. At this point, the technicians can accurately understand the working environment, optimize the design of non-stop operation plans and implement safety strategies on site.

[0053] Also included is a three-dimensional modeling system;

[0054] The three-dimensional modeling system includes: an image acquisition module for acquiring images collected by the drone and point cloud data thereof, a first three-dimensional module for constructing a three-dimensional model based on the point cloud data, a second three-dimensional module for constructing a three-dimensional model based on the image, a fusion module for fusing the three-dimensional model constructed by the first three-dimensional module with the three-dimensional model constructed by the second three-dimensional module, an analysis module for analyzing the fused three-dimensional model to analyze whether there are abnormal points, and an optimization module for optimizing the fused three-dimensional model;

[0055] The image acquisition module is connected to the first three-dimensional module, the image acquisition module is connected to the second three-dimensional module, the first three-dimensional module is connected to the fusion module, the second three-dimensional module is connected to the fusion module, the fusion module is connected to the analysis module, and the analysis module is connected to the optimization module.

[0056] The first three-dimensional module includes: a point cloud acquisition module for acquiring point cloud data in the image acquisition module, and a first construction module for constructing a three-dimensional model according to the point cloud data;

[0057] The image acquisition module is connected to the point cloud acquisition module, the point cloud acquisition module is connected to the first construction module, and the first construction module is connected to the fusion module.

[0058] The second three-dimensional module includes: an image preprocessing module for processing the image collected by the image acquisition module, an image cutting module for cutting the processed image to cut the processed image into a plurality of independent images, a feature extraction module for extracting features from the cut independent images, a shape judgment module for judging the extracted feature shape, a size calculation module for calculating the extracted feature size to obtain the feature size, and a second construction module for constructing a three-dimensional model according to the features, shape and size of the image;

[0059] The image acquisition module is connected to the image preprocessing module, the image preprocessing module is connected to the image cutting module, the image cutting module is connected to the feature extraction module, the feature extraction module is connected to the shape judgment module, the shape judgment module is connected to the size calculation module, the size calculation module is connected to the second construction module, and the second construction module is connected to the fusion module.

[0060] The optimization module includes: a color analysis module for analyzing the color of the image in the image acquisition module, a texture analysis module for analyzing the texture in the image acquisition module, and a fourth construction module for optimizing the fused three-dimensional model based on the analyzed color and texture;

[0061] The analysis module is connected to the color analysis module, the color analysis module is connected to the texture analysis module, and the texture analysis module is connected to the fourth building module.

[0062] The specific workflow of the image preprocessing module is as follows:

[0063] Process 1: Input the scene image, and calculate the temporal fitness model of rods and cones respectively. The bilateral filtering method is used to simulate the interaction between photoreceptor cells and surrounding cells after receiving light stimulation.

[0064] Among them, the time adaptation of visual cells is calculated as follows:

[0065] The neural photosensitive processing process in vision is very fast. In rod cells, the adjustment can be completed within 200 seconds, and for cone cells, it is even shorter. The following exponential function can be used to simulate the photosensitive brightness L of the cell adaptation state a Changes:

[0066]

[0067] Where L0 is the initial brightness value before the change, L a，i and L a，i-1 are the light sensitivity at the i-th moment and the previous moment (i-1), τ b For cones and rods, the time interval is 80ms and 150ms respectively. By continuously updating L during the time change, a By using the values, we can get the changes in the light sensitivity of the rod and cone cells at different times.

[0068] When the human eye gradually adapts to the visible brightness in the scene, it can see the scene objects in a larger brightness range. At this time, the adaptation state of the photoreceptor cells at the current position is defined as L a, its value can be expressed as a function of the relationship between the brightness of other cells in the neighborhood. At this time, the local brightness value is calculated by the bilateral filtering method. Through this filter function, the joint effect of the surrounding cells can be approximately estimated, and the image can be free from the phenomenon of excessive halo. The calculation formula of the filter function is as follows:

[0069]

[0070] Where N(n) represents the neighborhood pixels, the filter window radius is usually 3 or 5, G(·) represents the two-dimensional Gaussian filter function, σ s is the weight attenuation factor in the spatial domain, σ g is the weight attenuation factor in the brightness domain, L m and L n Represent the light-sensitive brightness values ​​of the current and neighboring pixels respectively. Since the calculation of the weight coefficient in the bilateral filtering method is adaptively changed with the brightness, when the brightness difference is small, the weight coefficient is large, and when the brightness span is large, the weight coefficient is very small and tends to 0, which can effectively maintain the boundary and reduce the halo phenomenon in local filtering;

[0071] Process 2: Calculate the time-varying perceptual response values ​​of rods and cones in the process;

[0072] Among them, the response calculation model of rod cells and cone cells is as follows:

[0073] When the eyes are fully adapted to the ambient light, an S-shaped response curve can be obtained by measuring the light stimulation of neurons, which is as follows:

[0074]

[0075] Where L is the brightness (in cd / m2m 2 ), V xam is the maximum value of the neural response of the photoreceptor cell, σ(L a ) is the brightness of the half-saturated state. Its physical meaning is that when its value is brightness L, V / V max =1 / 2 where V max and σ are affected by the current photoreceptor cell adaptation state L a Influence, V is the response to brightness L in the current adaptation state, n is a parameter used to control sensitivity. The larger the value of n, the steeper the curve, and the smaller the value of n, the flatter the curve. In this model, V / V maxThe relationship with L forms an S curve. From the S curve, we can see that in the middle part of the curve, the relationship between brightness and response is almost linear, indicating that the photoreceptor cells are relatively sensitive to changes in brightness. The upper and lower ends of the curve have a very narrow range of response. That is, when L is much smaller than or greater than σ, the photoreceptor cells are insensitive. At this time, if a stimulus response is to be generated, a larger brightness change is required. This situation is consistent with the response compression mechanism mentioned earlier in visual perception. The adaptation mechanism of the human eye can eliminate the effect of response compression and restore visibility. Using this response model, the corresponding perception processing process of the human eye can be approximately simulated.

[0076] To simplify the processing, V is converted into a relative response value, that is, V max =1, then the responsiveness of cones and rods at time i are:

[0077]

[0078] Where L i，rod and L i，cone represents the brightness of rods and cones in dark and bright vision at time i, respectively, σ(L a ) is the same as the current adaptation state brightness L a Related function, its value determines the magnitude of the responsiveness at the current brightness;

[0079] The key to establishing a time-dependent response model of visual cells is to find the half-saturation response value σ(L a ) changes with time t, firstly, according to the above formula (1), the light sensitivity L of the visual cell adaptation state is obtained a Changes with time t and after time adaptation calculation;

[0080] The process of vision generation begins with the chemical reaction of photosensitive pigments after photoreceptor cells receive light stimulation. This chemical reaction will lead to the bleaching of pigments, and the bleaching speed is related to the received light intensity. The greater the light intensity, the faster the bleaching speed. The other reverse process, pigment synthesis, is related to the amount of bleached pigments and the duration of the reaction. Since the speed of the chemical reaction is related to the current amount of pigments, and since the bleaching and synthesis processes are not completed instantly, the visual adaptation process depends on the current pigment concentration. Therefore, it can be obtained that in a stable state, B is taken as the total amount of unbleached pigments, and its relationship with the current brightness L is defined as:

[0081]

[0082] Where L represents the brightness value when the pigment concentration reaches half, L h is a constant, and 4logcd / m is used for cones and rods respectively. 2 and 4.5logcd / m2 , because in the psychological experiment, it was found that the pigment bleaching process satisfies the exponential change relationship, and because it is related to the current light intensity, it can be extended to the process of time change, and we get:

[0083]

[0084] When t→∞, it means that the current state has reached a stable state, and the above formula (6) can be obtained, where σ a For rods and cones, the time intervals were 400 and 120 s, respectively. h is the total amount of pigment at the time t=0. According to the amount of pigment in steady state, a curve can be obtained to show the relationship between the amount and brightness. Through this curve, it can be found that when the brightness is less than 3 logarithmic brightness levels, the total amount of pigment basically remains unchanged. When the brightness is expanded to 5-6 logarithmic brightness levels, the pigment is quickly consumed. When the brightness suddenly changes to a very high level, the number of photons absorbed by the photoreceptor cells will decrease, so people will feel dazzling.

[0085] The expression of the total amount of pigment changing with time is B = B(t) = B(L a (t)) and then find the half-saturation state response value σ(L a ) changes with time t. As the brightness gradually increases, the total amount of pigment in the photoreceptor cells decreases. When the brightness gradually decreases, rhodopsin will gradually synthesize, causing the total amount of pigment to increase. It can be assumed that the brightness and the total amount of pigment on the photoreceptor cells are inversely proportional, and their relationship can be expressed as:

[0086] σ b (L a )×B h (L a ) = const(8)

[0087] In this way, the change of the half-saturation state response value σ with time t can be obtained, that is, σ=σ(t)=σ(L a (t)), and combined with the change mechanism of cones and rods, the relationship between the semi-saturated state and the current adaptation state can be obtained, which can continuously simulate the perception process of dynamic visual changes during light and dark adaptation;

[0088] Process 3: Use the inverse responsivity transformation module to transform and map its high dynamic image onto a low dynamic range observation display device;

[0089] The inverse responsivity transformation is as follows:

[0090] After obtaining the variation law of the high dynamic range of the light-dark adaptation scene, how to transform it to the low dynamic range observation display device for effective display is exactly the purpose of tone mapping. Let V be the response value for the scene defined previously, and V′ be the response value calculated by the observation display device, so that: V = V′, then we can get:

[0091]

[0092] After the inverse transformation of the above formula, the relationship between the final brightness values ​​of cones and rods can be derived respectively;

[0093]

[0094] From the above formula, we can see that the key to obtaining the final brightness value is to calculate the unknown quantities σ′ and σ for cones and rods. In addition to being affected by the overall ambient brightness, people are mainly affected locally when observing, which is determined by the local adaptation characteristics of the human eye. In view of this phenomenon, the display device observer is also calculated using the local state solution method, which is improved:

[0095]

[0096] By using formula (11), the higher and lower brightness areas in the image can be compressed, so that most of the brightness areas can be stretched to a larger range. If the average brightness of the image is low, it means that most areas are dark, so the value of t should be reduced. Similarly, if it is brighter, the value of t should be increased. In the implementation process, δ is a very small value, and t is a parameter set by the user. According to the different values ​​of t, the obvious changes in image brightness can be seen;

[0097] Process 4: By changing the color adaptation matrix, the corrected scene images perceived by the human eye at different times are obtained;

[0098] Among them, the color adaptation matrix transformation is as follows:

[0099] Use the Y component of the brightness channel to scale the other channels in the same proportion to maintain the proportion of each component;

[0100]

[0101] where Y in (x, y) is the initial brightness, Y out(x, y) is the brightness after mapping, In(x, y) and Out(x, y) represent the initial and calculated X and Z channel components respectively. After this transformation, the new XYZ value is obtained, and then the value is converted to LMS space to obtain LMS1, where LMS1 represents the responsiveness of three different visual cells (S, M, L-Cones) respectively;

[0102]

[0103] Convert LMS1 to LMS2 value in the new environment, and finally convert it back to XYZ space to get the color-adapted value.

[0104] Although the present invention has been described above with reference to the embodiments, various modifications may be made thereto and parts thereof may be replaced by equivalents without departing from the scope of the present invention. In particular, as long as there is no structural conflict, the various features in the embodiments disclosed in the present invention may be used in combination with each other in any manner, and the fact that these combinations are not exhaustively described in this specification is only for the sake of omitting space and saving resources. Therefore, the present invention is not limited to the specific embodiments disclosed herein, but includes all technical solutions falling within the scope of the claims.

Claims

1. A three-dimensional modeling method of point cloud and image data of a distribution network non-stop operation site, characterized in that: The specific steps are as follows: Step 1: Obtain the images and point cloud data collected by the drone through the image acquisition module; Step 2: The point cloud data in the image acquisition module is acquired through the point cloud acquisition module. After acquisition, a three-dimensional model is constructed according to the point cloud data through the first construction module; Step 3: The image collected by the image acquisition module is processed by the image preprocessing module to adjust the light and dark parts in the image so that the objects in the image become clear. After processing, the processed image is cut by the image cutting module to cut the processed image into several independent images. After cutting, the independent images cut out are subjected to feature extraction by the feature extraction module. After extraction, the extracted feature shape is judged by the shape judgment module. After judgment, the extracted feature size is calculated by the size calculation module to obtain the feature size. After that, the three-dimensional model is constructed according to the features, shape and size of the image by the second construction module. Step 4: The 3D model constructed by the first 3D module is fused with the 3D model constructed by the second 3D module through the fusion module. After the fusion, the fused 3D model is analyzed by the analysis module to find out whether there are abnormal points. If there are abnormal points, it means that there are differences between the two sets of 3D models. At this time, it is necessary to start the construction again; Step 5: If there are no abnormal points, the color of the image in the image acquisition module will be analyzed by the color analysis module, and the texture in the image acquisition module will be analyzed by the texture analysis module. After that, the fused three-dimensional model will be optimized based on the analyzed color and texture through the fourth construction module. At this point, the technicians can accurately understand the working environment, optimize the design of non-stop operation plans and implement safety strategies on site.

2. The method for three-dimensional modeling of point cloud and image data of distribution network non-stop operation site according to claim 1 is characterized in that: Also included is a three-dimensional modeling system, the three-dimensional modeling system comprising: Image acquisition module, used to obtain images and point cloud data collected by the drone; The first 3D module builds a 3D model based on point cloud data; The second 3D module builds a 3D model based on the image; A fusion module, used to fuse the three-dimensional model constructed by the first three-dimensional module with the three-dimensional model constructed by the second three-dimensional module; An analysis module is used to analyze the fused three-dimensional model to determine whether there are any abnormal points; The optimization module is used to optimize the fused 3D model.

3. The method for three-dimensional modeling of point cloud and image data of distribution network non-stop operation site according to claim 2 is characterized in that: The image acquisition module is connected to the first three-dimensional module, the image acquisition module is connected to the second three-dimensional module, the first three-dimensional module is connected to the fusion module, the second three-dimensional module is connected to the fusion module, the fusion module is connected to the analysis module, and the analysis module is connected to the optimization module.

4. The method for three-dimensional modeling of point cloud and image data of distribution network non-stop operation site according to claim 3 is characterized in that: The first three-dimensional module includes: A point cloud acquisition module is used to acquire point cloud data in the image acquisition module; The first construction module is used to construct a three-dimensional model according to point cloud data.

5. The method for three-dimensional modeling of point cloud and image data of distribution network non-stop operation site according to claim 4 is characterized in that: The image acquisition module is connected to the point cloud acquisition module, the point cloud acquisition module is connected to the first construction module, and the first construction module is connected to the fusion module.

6. The method for three-dimensional modeling of point cloud and image data of distribution network non-stop operation site according to claim 3 is characterized in that: The second three-dimensional module includes: An image preprocessing module, used for processing the images collected by the image acquisition module; An image cutting module is used to cut the processed image into a plurality of independent images; A feature extraction module, used for extracting features from the cut independent images; A shape judgment module is used to judge the extracted characteristic shape; A size calculation module is used to calculate the extracted feature size to obtain the feature size; The second construction module is used to construct a three-dimensional model according to the characteristics, shape and size of the image.

7. The method for three-dimensional modeling of point cloud and image data of distribution network non-stop operation site according to claim 6, characterized in that: The image acquisition module is connected to the image preprocessing module, the image preprocessing module is connected to the image cutting module, the image cutting module is connected to the feature extraction module, the feature extraction module is connected to the shape judgment module, the shape judgment module is connected to the size calculation module, the size calculation module is connected to the second construction module, and the second construction module is connected to the fusion module.

8. The method for three-dimensional modeling of point cloud and image data of distribution network non-stop operation site according to claim 3, characterized in that: The optimization module includes: A color analysis module, used to analyze the color of the image in the image acquisition module; A texture analysis module, used for analyzing the texture in the image acquisition module; The fourth building module optimizes the fused three-dimensional model based on the analyzed color and texture.

9. The method for three-dimensional modeling of point cloud and image data of distribution network non-stop operation site according to claim 8, characterized in that: The analysis module is connected to the color analysis module, the color analysis module is connected to the texture analysis module, and the texture analysis module is connected to the fourth building module.

10. The method for three-dimensional modeling of point cloud and image data of distribution network non-stop operation site according to claim 6, characterized in that: The specific workflow of the image preprocessing module is as follows: Process 1: Input the scene image, and calculate the temporal fitness model of rods and cones respectively. The bilateral filtering method is used to simulate the interaction between photoreceptor cells and surrounding cells after receiving light stimulation. Process 2: Calculate the time-varying perceptual response values ​​of rods and cones in the process; Process 3: Use the inverse responsivity transformation module to transform and map its high dynamic image onto a low dynamic range observation display device; Process 4: By changing the color adaptation matrix, the corrected scene images perceived by the human eye at different times are obtained.

Images