Photovoltaic panel assembly segmentation method, system, device and storage medium

Abstract

The embodiment of the application discloses a photovoltaic panel assembly segmentation method, system, device and storage medium, relates to the photovoltaic power generation technical field, and deep learning semantic segmentation can accurately identify all photovoltaic panel regions, and map the photovoltaic panel regions to photovoltaic panel infrared thermal images to obtain infrared photovoltaic panel regions; after edge detection and straight line detection processing are performed on each infrared photovoltaic panel region, straight line segments in the region are obtained; by comparing the first slope of the straight line segments with the second slope of the photovoltaic panel wide side, it is ensured that only the straight line segments similar to the slope of the photovoltaic panel wide side are selected as target straight line segments, so as to exclude the straight lines obviously not belonging to the photovoltaic panel assembly boundary; the target straight line segments are used for region segmentation, it is ensured that the segmented photovoltaic panel assembly is not connected, and therefore the photovoltaic panel assembly is clearly segmented, and the accuracy of photovoltaic panel assembly identification is improved.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of photovoltaic power generation, and particularly to a photovoltaic panel assembly segmentation method, system, device and storage medium. BACKGROUND

[0002] In the prior art, a segmentation positioning method for a photovoltaic panel assembly is based on deep learning to segment the photovoltaic panel image to obtain a segmented image of the photovoltaic panel assembly, and then contour detection is performed on the segmented image to determine the coordinate parameters of the photovoltaic panel assembly. However, in actual application, since a single photovoltaic panel assembly is very small and arranged very closely, the actual segmented photovoltaic panel assembly is connected together simply by relying on the image segmentation algorithm, and the photovoltaic panel assembly cannot be clearly segmented. SUMMARY

[0003] Therefore, the present application provides a photovoltaic panel assembly segmentation method, system, device and storage medium.

[0004] The specific technical scheme of the first embodiment of the present application is as follows: a photovoltaic panel assembly segmentation method, the method comprising: performing deep learning semantic segmentation on an infrared thermal image of a photovoltaic panel to obtain all photovoltaic panel regions in the infrared thermal image of the photovoltaic panel; mapping each photovoltaic panel region to the infrared thermal image of the photovoltaic panel to obtain an infrared photovoltaic panel region corresponding to each photovoltaic panel region; performing edge detection and straight line detection on each infrared photovoltaic panel region to obtain a straight line segment in each infrared photovoltaic panel region; obtaining a first slope of the straight line segment and a second slope of a wide side of the photovoltaic panel in the infrared photovoltaic panel region; the wide side is a short side; obtaining a target straight line segment according to the first slope and the second slope; the target straight line segment is a straight line segment whose difference between the first slope and the second slope is within a first preset threshold; and segmenting the infrared photovoltaic panel region by using the target straight line segment to obtain a photovoltaic panel assembly in the infrared photovoltaic panel region.

[0005] Preferably, after obtaining all photovoltaic panel regions in the photovoltaic panel infrared thermal image, the method further comprises: obtaining an outer contour of a first photovoltaic panel region and a circumscribed rectangle of the outer contour; obtaining a convex hull line segment relative to the circumscribed rectangle in the outer contour; the convex hull line segment is at least one; determining whether a photovoltaic panel region division line exists in the first photovoltaic panel region according to two end points of the convex hull line segment, a farthest point in the convex hull line segment relative to the circumscribed rectangle, a long side of the circumscribed rectangle, and a wide side of the circumscribed rectangle; the convex hull line segment is a line segment protruding relative to the circumscribed rectangle; if the photovoltaic panel region division line exists, the first photovoltaic panel region is divided by using the photovoltaic panel region division line to obtain a first photovoltaic panel division region and a second photovoltaic panel division region, and the first photovoltaic panel division region and the second photovoltaic panel division region are both taken as the optimized photovoltaic panel region.

[0006] Preferably, the determining whether the photovoltaic panel region division line exists in the first photovoltaic panel region according to the two end points of the convex hull line segment, the farthest point in the convex hull line segment relative to the circumscribed rectangle, the long side of the circumscribed rectangle, and the wide side of the circumscribed rectangle comprises: connecting and extending the two end points of the convex hull line segment to obtain an end point connecting straight line; the end point connecting straight line intersects a target long side of the circumscribed rectangle; the target long side is a long side of the circumscribed rectangle closest to the farthest point of the convex hull line segment; obtaining a shortest distance between the farthest point and the end point connecting straight line; obtaining a first included angle between the end point connecting straight line and the target long side of the circumscribed rectangle, and obtaining a second included angle between the end point connecting line segment and a target wide side of the circumscribed rectangle; the target wide side is a wide side of the circumscribed rectangle closest to the farthest point of the convex hull line segment; comparing the first included angle and the second included angle; if the first included angle is smaller than the second included angle and the shortest distance is greater than a second preset threshold value, a first photovoltaic panel region division line exists in the first photovoltaic panel region; the first photovoltaic panel region division line is a straight line passing through the farthest point and parallel to the wide side of the photovoltaic panel; if the first included angle is greater than the second included angle and the shortest distance is greater than a third preset threshold value, a second photovoltaic panel region division line exists in the first photovoltaic panel region; the second photovoltaic panel region division line is a straight line passing through the farthest point and along the long side of the photovoltaic panel; the first photovoltaic panel region division line and the second photovoltaic panel region division line are both the photovoltaic panel region division line.

[0007] Preferably, the second preset threshold value is smaller than a first length of the wide side of the circumscribed rectangle; and the third preset threshold value is smaller than a second length of the long side of the circumscribed rectangle.

[0008] Preferably, after obtaining all photovoltaic panel regions in the photovoltaic panel infrared thermal image, the method further comprises: obtaining an outer contour of a second photovoltaic panel region and a contour area of the outer contour; determining whether the contour area is less than a fourth preset threshold; and if the contour area is less than the fourth preset threshold, deleting the second photovoltaic panel region.

[0009] Preferably, after obtaining the target straight line segments according to the first slope and the second slope, the method further comprises: sorting all the target straight line segments from top to bottom to obtain a target straight line segment sequence; obtaining a first target straight line segment with a median number; determining whether a difference between a third length of the first target straight line segment and a fourth length of the wide side of the photovoltaic panel is greater than a fifth preset threshold; and if the difference is greater than the fifth preset threshold, deleting the first target straight line segment.

[0010] Preferably, after obtaining the first target straight line segment with the median number, the method further comprises: obtaining a line segment distance between a second target straight line segment adjacent to the first target straight line segment; and if the line segment distance is less than a sixth preset threshold, deleting the second target straight line segment.

[0011] A second embodiment of the present application provides a photovoltaic panel assembly segmentation system, which comprises a photovoltaic region segmentation module, a mapping module, a straight line segment detection module, a slope acquisition module, a target straight line segment acquisition module, and a segmentation module. The photovoltaic region segmentation module is configured to perform deep learning semantic segmentation on a photovoltaic panel infrared thermal image to obtain all photovoltaic panel regions in the photovoltaic panel infrared thermal image. The mapping module is configured to map each photovoltaic panel region to the photovoltaic panel infrared thermal image to obtain an infrared photovoltaic panel region corresponding to each photovoltaic panel region. The straight line segment detection module is configured to perform edge detection and straight line detection on each infrared photovoltaic panel region to obtain a straight line segment in each infrared photovoltaic panel region. The slope acquisition module is configured to acquire a first slope of the straight line segment and a second slope of a wide side of a photovoltaic panel in the infrared photovoltaic panel region. The target straight line segment acquisition module is configured to acquire a target straight line segment according to the first slope and the second slope. The target straight line segment is a straight line segment whose difference between the first slope and the second slope is within a first preset threshold. The segmentation module is configured to segment the infrared photovoltaic panel region using the target straight line segment to obtain a photovoltaic panel assembly in the infrared photovoltaic panel region.

[0012] A third embodiment of the present application provides a photovoltaic panel assembly segmentation device, which comprises a memory and a processor. The memory stores a computer program. When the computer program is executed by the processor, the processor performs the steps of the method according to any one of the first embodiment of the present application.

[0013] A specific technical scheme of the fourth embodiment of the present application is: a computer readable storage medium, which stores a computer program, the computer program is executed by a processor, so that the processor executes the steps of the method according to any one of the first embodiment of the present application.

[0014] The embodiments of the present application have the following beneficial effects:

[0015] The deep learning semantic segmentation in the present application can accurately identify all photovoltaic panel regions and map the photovoltaic panel regions to the photovoltaic panel infrared thermal image to obtain an infrared photovoltaic panel region. After edge detection and straight line detection are performed on each infrared photovoltaic panel region, straight line segments in the region are obtained. By comparing the first slope of the straight line segments with the second slope of the photovoltaic panel wide side, only the straight line segments similar to the slope of the photovoltaic panel wide side are selected as target straight line segments, so as to exclude the straight lines obviously not belonging to the photovoltaic panel component boundary. The region is segmented by using the target straight line segments, so as to ensure that the segmented photovoltaic panel components are not connected, thereby clearly segmenting the photovoltaic panel components and improving the accuracy of photovoltaic panel component identification. BRIEF DESCRIPTION OF DRAWINGS

[0016] In order to more clearly illustrate the technical solutions in the embodiments of the present application or the prior art, the drawings needed in the embodiments or the prior art description will be briefly introduced. Obviously, the drawings in the following description are only some embodiments of the present application, and other drawings can be obtained by those skilled in the art without creative labor.

[0017] Figure 1 A step flow chart for photovoltaic panel component segmentation method;

[0018] Figure 2 A schematic diagram of a photovoltaic panel infrared thermal image;

[0019] Figure 3 A schematic diagram of a photovoltaic panel region;

[0020] Figure 4 A schematic diagram of an infrared photovoltaic panel region;

[0021] Figure 5 A schematic diagram of an infrared photovoltaic panel region after edge detection;

[0022] Figure 6 A schematic diagram of a straight line segment;

[0023] Figure 7 A schematic diagram of noise and misconnection region;

[0024] Figure 8Fig. 2 is a schematic diagram of a photovoltaic panel region after small noise regions are deleted;

[0025] Figure 9 Fig. 3 is a flow chart of a step of identifying whether a split line exists in a photovoltaic panel region;

[0026] Figure 10 Fig. 4 is a schematic diagram of an outer rectangle;

[0027] Figure 11 Fig. 5 is a schematic diagram of a convex hull line segment;

[0028] Figure 12 Fig. 6 is a schematic diagram of a misconnection of a photovoltaic panel region;

[0029] Figure 13 Fig. 7 is a schematic diagram of a position of a convex hull line segment and an outer rectangle;

[0030] Figure 14 Fig. 8 is a schematic diagram of a split line;

[0031] Figure 15 Fig. 9 is a schematic diagram of a photovoltaic panel region after a split line is split;

[0032] Figure 16 Fig. 10 is a schematic diagram of an infrared photovoltaic panel region after a split line is split;

[0033] Figure 17 Fig. 11 is a schematic diagram of a split line of a photovoltaic assembly;

[0034] Figure 18 Fig. 12 is a schematic diagram of a target straight line segment;

[0035] Figure 19 Fig. 13 is a schematic diagram of all photovoltaic assemblies after being split;

[0036] Figure 20 Fig. 14 is a schematic diagram of a structure of a photovoltaic panel assembly splitting system;

[0037] In the figure, 301 is a photovoltaic region splitting module, 302 is a mapping module, 303 is a straight line segment detection module, 304 is a slope acquisition module, 305 is a target straight line segment acquisition module, and 306 is a splitting module. DETAILED DESCRIPTION

[0038] The technical solutions in the embodiments of the present application will be described clearly and completely below with reference to the drawings in the embodiments of the present application. Obviously, the described embodiments are only some of the embodiments of the present application, but not all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative labor fall within the scope of the present application.

[0039] The terms "first", "second", and the like in the description and in the claims of the present application and in the drawings refer to different objects and not to a particular sequence. Furthermore, the terms "comprises", "comprising", "has", "having", "includes", "including", and the like are to be construed open-ended, allowing for instances where there are equivalents to the specified features, steps or components as set forth in any of the claims. For example, processes, methods, systems, products, or apparatuses that comprise, have, include or the like a series of steps or modules are not to be construed as necessarily limited to the specific order of steps or modules, unless specifically stated as such in the claims.

[0040] Reference herein to "an embodiment" means that a particular feature, structure, or characteristic described in connection with the embodiment can be included in at least one embodiment of the application. The appearances of the phrase in various places in the specification are not necessarily all referring to the same embodiment, nor are they necessarily mutually exclusive of one another. It is expressly understood that the embodiments described herein are merely examples from a potentially infinite number of embodiments that serve the same, a similar, or other purposes, as will be apparent to one of ordinary skill in the art.

[0041] Reference is made to Figure 1 , a flow chart of steps of a method for segmenting a photovoltaic panel assembly in a first embodiment of the application, to improve the accuracy of photovoltaic panel assembly recognition, the method comprising:

[0042] Step 101, performing deep learning semantic segmentation on the photovoltaic panel infrared thermal image to obtain all photovoltaic panel regions in the photovoltaic panel infrared thermal image;

[0043] Step 102, mapping each of the photovoltaic panel regions to the photovoltaic panel infrared thermal image to obtain an infrared photovoltaic panel region corresponding to each photovoltaic panel region;

[0044] Step 103, performing edge detection and straight line detection on each of the infrared photovoltaic panel regions to obtain straight line segments in each of the infrared photovoltaic panel regions;

[0045] Step 104, obtaining a first slope of the straight line segment and a second slope of a wide side of the photovoltaic panel in the infrared photovoltaic panel region; the wide side is a short side;

[0046] Step 105, obtaining a target straight line segment according to the first slope and the second slope; the target straight line segment is a straight line segment whose difference between the first slope and the second slope is within a first preset threshold;

[0047] Step 106, segmenting the infrared photovoltaic panel region using the target straight line segment to obtain a photovoltaic panel assembly in the infrared photovoltaic panel region.

[0048] Specifically, the photovoltaic panel infrared thermal image is collected by a drone, and a schematic diagram of the photovoltaic panel infrared thermal image is shown in Figure 2To perform deep learning semantic segmentation on the photovoltaic panel infrared thermal image, all photovoltaic panel regions in the photovoltaic panel infrared thermal image are obtained. Please refer to Figure 3 , Figure 3 A white region in the figure is a photovoltaic panel region. Each photovoltaic panel region is mapped to the photovoltaic panel infrared thermal image to obtain an infrared photovoltaic panel region corresponding to each photovoltaic panel region. Please refer to Figure 4 Since each unit component of the photovoltaic panel is separated by a clear edge dividing line, edge detection is performed on each region, and then, according to Hough line detection, a straight line segment can be screened out. Please refer to Figure 5 for a schematic diagram of the edge detection. Please refer to Figure 6 for a schematic diagram of the straight line segment. The slope of each straight line segment is calculated. If the slope is close to the slope of the photovoltaic panel wide side width, the straight line segment is considered to belong to the dividing line of the photovoltaic panel component, and the adhered photovoltaic panel components are divided using the dividing line.

[0049] The deep learning semantic segmentation in the method can accurately identify all photovoltaic panel regions and map the photovoltaic panel regions to the photovoltaic panel infrared thermal image to obtain infrared photovoltaic panel regions. After edge detection and line detection are performed on each infrared photovoltaic panel region, a straight line segment in the region is obtained. By comparing the first slope of the straight line segment with the second slope of the photovoltaic panel wide side, only the straight line segment with a slope close to that of the photovoltaic panel wide side is selected as the target straight line segment to exclude the straight lines that obviously do not belong to the boundary of the photovoltaic panel component. The target straight line segment is used for region division to ensure that the divided photovoltaic panel components are not connected, thereby clearly dividing the photovoltaic panel components and improving the accuracy of photovoltaic panel component recognition.

[0050] Specifically, the divided photovoltaic panel regions may have two problems: (1) a part of the noise region may be divided due to misdivision (for example, the red position in Figure 7 ), and (2) two photovoltaic panels may be misconnected (for example, the green position in Figure 7 ).

[0051] To solve the first problem, the contour detection is used to obtain each region of the entire image, and the regions with small areas are screened out. Specifically, after the all photovoltaic panel regions in the photovoltaic panel infrared thermal image are obtained, the method further includes: obtaining an outer contour of a second photovoltaic panel region and a contour area of the outer contour; determining whether the contour area is less than a fourth preset threshold; and if the contour area is less than the fourth preset threshold, deleting the second photovoltaic panel region. The regions with small areas are screened out as shown in Figure 8 .

[0052] To solve the second problem, it is necessary to determine which of the regions detected by the contouring are connected by two or more photovoltaic panels, and then find the misconnection points of the photovoltaic panels in these regions, and then separate them, so as to separate the misconnected photovoltaic panel regions.

[0053] Specifically, please refer to Figure 9 , after step 101 obtains all the photovoltaic panel regions in the infrared thermal image of the photovoltaic panel, it further includes:

[0054] Step 201, obtaining the outer contour of a first photovoltaic panel region and the circumscribed rectangle of the outer contour;

[0055] Step 202, obtaining a convex hull line segment in the outer contour relative to the circumscribed rectangle; the convex hull line segment is at least one;

[0056] Step 203, judging whether there is a photovoltaic panel region division line in the first photovoltaic panel region according to two end points of the convex hull line segment, the farthest point in the convex hull line segment relative to the circumscribed rectangle, the long side of the circumscribed rectangle and the width of the circumscribed rectangle; the convex hull line segment is a line segment protruding relative to the circumscribed rectangle;

[0057] Step 204, if there is the photovoltaic panel region division line, then the first photovoltaic panel region is divided by using the photovoltaic panel region division line to obtain a first photovoltaic panel division region and a second photovoltaic panel division region, and the first photovoltaic panel division region and the second photovoltaic panel division region are both regarded as the optimized photovoltaic panel region.

[0058] First, the circumscribed rectangle of each region is obtained, and the schematic diagram of the circumscribed rectangle is shown in Figure 10 , the long side of the circumscribed rectangle is length and the width is width, the convex hull line segment relative to the circumscribed rectangle is obtained in the outer contour, the convex hull of each contour region is calculated, and then the convex hull line segment contains the following attributes: start point s, end point e, and farthest point f, wherein the farthest point f is the concave point of the contour (such as the red dot in Figure 11 ), and the misconnection point of two contours may exist in the concave point, whether there is a photovoltaic panel region division line in the first photovoltaic panel region is judged according to the start point s, the end point e, the farthest point f, the long side of the circumscribed rectangle and the width of the circumscribed rectangle, if there is a photovoltaic panel region division line, then the photovoltaic panel region is divided by using the photovoltaic panel region division line to obtain the divided photovoltaic panel region.

[0059] In a specific embodiment, the method of judging whether there is a photovoltaic panel area dividing line in the first photovoltaic panel area based on the two endpoints of the convex hull segment, the farthest point of the convex hull segment relative to the circumscribed rectangle, the long side of the circumscribed rectangle and the wide side of the circumscribed rectangle includes: connecting the two endpoints of the convex hull segment and extending it to obtain an endpoint connecting straight line; the endpoint connecting straight line intersects with the target long side of the circumscribed rectangle; the target long side is a long side of the circumscribed rectangle that is closest to the farthest point of the convex hull segment; obtaining the shortest distance between the farthest point and the endpoint connecting straight line; obtaining a first angle between the endpoint connecting straight line and the target long side of the circumscribed rectangle, and obtaining a second angle between the endpoint connecting line segment and the target wide side of the circumscribed rectangle; the target wide side is the target a wide side in the circumscribed rectangle that is closest to the farthest point of the convex hull segment; comparing the sizes of the first angle and the second angle; if the first angle is smaller than the second angle, and the shortest distance is greater than a second preset threshold, then there is a first photovoltaic panel area dividing line in the first photovoltaic panel area; the first photovoltaic panel area dividing line is a straight line passing through the farthest point and parallel to the wide side of the photovoltaic panel; if the first angle is greater than the second angle, and the shortest distance is greater than a third preset threshold, then there is a second photovoltaic panel area dividing line in the first photovoltaic panel area; the second photovoltaic panel area dividing line is a straight line passing through the farthest point and along the long side of the photovoltaic panel; the first photovoltaic panel area dividing line and the second photovoltaic panel area dividing line are both the photovoltaic panel area dividing lines.

[0060] Specifically, calculate the angles angle_length and angle_width between the line dis connecting the starting point s and the end point e of each convex hull segment and the long side length and wide side width of the outline circumscribed rectangle. <angle_width,则这个凸包先算的起点s和终点e的连线dis_se靠近长边length,如angle_width<angle_length,认为凸包线段的起点s和终点e的连线dis_se靠近宽边width(如 Figure 13 Calculate the distance dis from the farthest point f of the convex hull segment to the line connecting the starting point s and the end point e (as shown in Figure 11 The following two cases are then divided into:

[0061] a: If the dis_se just calculated is close to the long side length, compare the distance of dis with the wide side width. If dis>width / n, where n>1, then point f is considered to be the incorrect connection point of the photovoltaic panel. Draw a dividing line through point f along the direction of the wide side width to separate the two photovoltaic panel areas.

[0062] b: If the just calculated dis_se is close to the width, compare the distance dis with the length of the long side, if dis>length / n, where n>1, then the f point is considered as a misconnection point of the photovoltaic panel, and the dividing line is along the direction of the long side length, i.e. the two photovoltaic panel regions are divided. As shown in the green point of Figure 12 , the dividing line is the blue line segment of Figure 14 , and the result of the division is as shown in Figure 15 .

[0063] In specific embodiments, the second preset threshold is less than a first length of the width of the circumscribed rectangle; and the third preset threshold is less than a second length of the length of the circumscribed rectangle. Specifically, if the just calculated dis_se is close to the long side length, compare the distance dis with the width, if dis>width / 3, then the f point is considered as a misconnection point of the photovoltaic panel; if the just calculated dis_se is close to the width, compare the distance dis with the length of the long side, if dis>length / 3, then the f point is considered as a misconnection point of the photovoltaic panel.

[0064] In specific embodiments, the detected straight line limit is composed of many noise line segments, which need to be further filtered. Specifically, the target straight line segment is obtained according to the first slope and the second slope, and then the method further comprises: sorting all the target straight line segments from top to bottom to obtain a target straight line segment sequence; obtaining a first target straight line segment with a median serial number; judging whether a difference between a third length of the first target straight line segment and a fourth length of the width of the photovoltaic panel is greater than a preset fifth threshold; and if the difference is greater than the preset fifth threshold, deleting the first target straight line segment.

[0065] Specifically, if the slope of the straight line segment is close to the slope of the width of the photovoltaic panel, the straight line is considered to belong to the dividing line of the photovoltaic panel assembly, and a schematic diagram of the dividing line is as shown in Figure 16 . As can be seen from Figure 16 , some straight line segments are arranged relatively closely, and there are still noise line segments, which need to be further screened and filtered to finally determine the final photovoltaic unit assembly dividing line. The specific method is: sorting the straight line segments (green line segments) of each photovoltaic region according to the arrangement mode of the photovoltaic region, from top to bottom along the width of the photovoltaic region, or from left to right. Take the middle line segment (each red box screens out a middle line segment, because the red box represents a photovoltaic panel, and each photovoltaic panel unit assembly needs to be divided, and the middle line segment refers to the line segment corresponding to the middle number after sorting) as the determinable unit photovoltaic assembly dividing line (as shown in Figure 17) as the middle straight line is less affected by the surrounding interference than the edge line segments, and can be determined as the partition line of the unit assembly. The straight line segments near the edge of the photovoltaic panel region can be affected by the partition region boundary recognition region, and are actually non-unit assembly partition line segments. After selecting the middle straight line, screening is performed along the two side straight lines, such as Figure 4 It can be known that the partition lines between the photovoltaic unit assemblies in the infrared image are relatively obvious, and after straight line detection, Figure 17 The detected green lines in the middle red frame certainly contain the photovoltaic unit partition lines, but also contain a small part of noise line segments. The condition for judging whether it is a partition line is as follows: Figure 18 As shown in the figure, if the line segment AD is a partition line, according to the structure of the photovoltaic panel assembly, AD is also the length of the photovoltaic assembly, that is, should be close to the width EF of the red frame. If the length difference between AD and EF is large, delete the partition line AD.

[0066] In specific embodiments, after the first target straight line segment with the median number is obtained, the method further includes: obtaining a line segment distance between a second target straight line segment adjacent to the first target straight line segment; and if the line segment distance is less than a preset sixth threshold, deleting the second target straight line segment.

[0067] Specifically, screening is performed along both sides of the straight line C. If it is similar to GH in the figure, it is certainly not a photovoltaic unit partition line because GH is too close to line C. It is possible that AC is the width of the photovoltaic panel line segment. Therefore, the condition is set that the distance from C to the next partition line (such as AC) satisfies at least half of the width EF of the photovoltaic panel region, and it is considered to be a photovoltaic panel assembly partition line. By analogy, all unit assemblies in a photovoltaic panel region are finally partitioned out (such as Figure 19 Here, there is a situation that may exist a missing detection problem. After detecting all the photovoltaic panel partition lines, the distance between the partition lines is calculated. For a wide distance, there may be a missing partition line in the middle, and the partition line can be supplemented.

[0068] In specific embodiments, please refer to Figure 20Fig. 2 is a structural schematic diagram of a photovoltaic panel component segmentation system according to a second embodiment of the present application. The system includes a photovoltaic area segmentation module 301, a mapping module 302, a straight line segment detection module 303, a slope acquisition module 304, a target straight line segment acquisition module 305, and a segmentation module 306. The photovoltaic area segmentation module 301 is configured to perform deep learning semantic segmentation on an infrared thermal image of a photovoltaic panel to obtain all photovoltaic panel areas in the infrared thermal image of the photovoltaic panel. The mapping module 302 is configured to map each photovoltaic panel area to the infrared thermal image of the photovoltaic panel to obtain an infrared photovoltaic panel area corresponding to each photovoltaic panel area. The straight line segment detection module 303 is configured to perform edge detection and straight line detection on each infrared photovoltaic panel area to obtain straight line segments in each infrared photovoltaic panel area. The slope acquisition module 304 is configured to acquire a first slope of the straight line segments and a second slope of a wide side of a photovoltaic panel in the infrared photovoltaic panel area. The wide side is a short side. The target straight line segment acquisition module 305 is configured to acquire target straight line segments according to the first slope and the second slope. The target straight line segments are straight line segments whose difference between the first slope and the second slope is within a first preset threshold. The segmentation module 306 is configured to segment the infrared photovoltaic panel area using the target straight line segments to obtain photovoltaic panel components in the infrared photovoltaic panel area.

[0069] The system according to the embodiment can accurately identify all photovoltaic panel areas through deep learning semantic segmentation, map the photovoltaic panel areas to the infrared thermal image of the photovoltaic panel to obtain infrared photovoltaic panel areas, obtain straight line segments in each infrared photovoltaic panel area through edge detection and straight line detection on each infrared photovoltaic panel area, compare the first slope of the straight line segments with the second slope of the wide side of the photovoltaic panel to ensure that only the straight line segments similar to the slope of the wide side of the photovoltaic panel are selected as target straight line segments to exclude straight lines obviously not belonging to the boundary of the photovoltaic panel components, and segment the areas using the target straight line segments to ensure that the segmented photovoltaic panel components are not connected, thereby clearly segmenting the photovoltaic panel components and improving the accuracy of photovoltaic panel component identification.

[0070] In specific embodiments, the third embodiment of the present application provides a photovoltaic panel component segmentation device, including a memory and a processor. The memory stores a computer program. When the computer program is executed by the processor, the processor executes the steps of the method according to any one of the first embodiment of the present application.

[0071] In specific embodiments, the fourth embodiment of the present application provides a computer readable storage medium storing a computer program. When the computer program is executed by a processor, the processor executes the steps of the method according to any one of the first embodiment of the present application.

[0072] The above embodiments only express several implementation ways of the present application, and the description is specific and detailed, but it should not be understood as a limitation to the patent scope of the present application. It should be noted that, for ordinary skilled in the art, without departing from the concept of the present application, several modifications and improvements can be made, which all belong to the protection scope of the present application. Therefore, the protection scope of the present application should be subject to the appended claims.

[0073] The above is only the preferred embodiment of the present application, and is not a limitation on other forms of the present application. Any skilled in the art can use the disclosed technical content to make changes or modifications to equivalent embodiments applied to other fields, but any simple modification, equivalent change and modification made to the above embodiments according to the technical essence of the present application without departing from the technical solution content of the present application still belongs to the protection scope of the technical solution of the present application.

Claims

1. A method for dividing a photovoltaic panel assembly, characterized in that: The method comprises: Performing deep learning semantic segmentation on the infrared thermal image of the photovoltaic panel to obtain the entire photovoltaic panel area in the infrared thermal image of the photovoltaic panel; Mapping each photovoltaic panel area to the photovoltaic panel infrared thermal image to obtain an infrared photovoltaic panel area corresponding to each photovoltaic panel area; Performing edge detection and straight line detection on each of the infrared photovoltaic panel areas to obtain straight line segments in each of the infrared photovoltaic panel areas; Obtaining a first slope of the straight line segment and a second slope of a broadside of a photovoltaic panel in the infrared photovoltaic panel region; Obtaining a target straight line segment according to the first slope and the second slope; the target straight line segment is a straight line segment whose difference between the first slope and the second slope is within a first preset threshold; Segmenting the infrared photovoltaic panel area using the target straight line segment to obtain photovoltaic panel components in the infrared photovoltaic panel area; After obtaining the entire photovoltaic panel area in the photovoltaic panel infrared thermal image, the method further includes: Obtaining an outer contour of a first photovoltaic panel region and a circumscribed rectangle of the outer contour; Obtaining a convex hull line segment relative to the circumscribed rectangle in the outer contour; the convex hull line segment is at least one; determining whether there is a photovoltaic panel area dividing line in the first photovoltaic panel area based on the two endpoints of the convex hull line segment, the farthest point of the convex hull line segment relative to the circumscribed rectangle, the long side of the circumscribed rectangle, and the wide side of the circumscribed rectangle; the convex hull line segment is a line segment that protrudes relative to the circumscribed rectangle; If the photovoltaic panel area dividing line exists, the first photovoltaic panel area is divided by the photovoltaic panel area dividing line to obtain a first photovoltaic panel dividing area and a second photovoltaic panel dividing area, and the first photovoltaic panel dividing area and the second photovoltaic panel dividing area are both used as the optimized photovoltaic panel area; Acquire a target straight line segment according to the first slope and the second slope, and then further include: Sort all the target straight line segments from top to bottom to obtain a target straight line segment sequence; Get the first target straight line segment whose serial number is the median; determining whether a difference between a third length of the first target straight line segment and a fourth length of the wide side of the photovoltaic panel is greater than a preset fifth threshold; If the difference is greater than the preset fifth threshold, the first target straight line segment is deleted.

2. The photovoltaic panel assembly segmentation method according to claim 1, wherein: The determining whether there is a photovoltaic panel area dividing line in the first photovoltaic panel area according to the two endpoints of the convex hull line segment, the farthest point of the convex hull line segment relative to the circumscribed rectangle, the long side of the circumscribed rectangle, and the wide side of the circumscribed rectangle includes: Connecting the two endpoints of the convex hull segment and extending them to obtain an endpoint connecting line; the endpoint connecting line intersects with the target long side of the circumscribed rectangle; the target long side is the long side of the circumscribed rectangle that is closest to the farthest point of the convex hull segment; Obtaining the shortest distance between the farthest point and the straight line connecting the endpoints; Obtaining a first angle between the endpoint connecting line and a target long side of the circumscribed rectangle, and obtaining a second angle between the endpoint connecting line segment and a target wide side of the circumscribed rectangle; the target wide side is a wide side of the circumscribed rectangle that is closest to the farthest point of the convex hull segment; comparing the first angle and the second angle; If the first angle is smaller than the second angle, and the shortest distance is greater than a second preset threshold, then a first photovoltaic panel area dividing line exists in the first photovoltaic panel area; the first photovoltaic panel area dividing line is a straight line passing through the farthest point and parallel to the wide side of the photovoltaic panel; If the first angle is greater than the second angle, and the shortest distance is greater than a third preset threshold, then there is a second photovoltaic panel area dividing line in the first photovoltaic panel area; the second photovoltaic panel area dividing line is a straight line passing through the farthest point along the long side direction of the photovoltaic panel; the first photovoltaic panel area dividing line and the second photovoltaic panel area dividing line are both the photovoltaic panel area dividing lines.

3. The photovoltaic panel assembly segmentation method according to claim 2, wherein: The second preset threshold is smaller than a first length of a wide side of the circumscribed rectangle; and the third preset threshold is smaller than a second length of a long side of the circumscribed rectangle.

4. The photovoltaic panel assembly segmentation method according to claim 3, wherein: After obtaining the entire photovoltaic panel area in the photovoltaic panel infrared thermal image, the method further includes: Obtaining an outer contour of a second photovoltaic panel region and an outline area of ​​the outer contour; Determining whether the contour area is less than a fourth preset threshold; If the contour area is smaller than a fourth preset threshold, the second photovoltaic panel area is deleted.

5. The photovoltaic panel assembly segmentation method according to claim 1, wherein: After obtaining the first target straight line segment whose serial number is the median, the method further includes: Acquire a segment distance between second target straight line segments adjacent to the first target straight line segment; If the line segment distance is less than a preset sixth threshold, the second target straight line segment is deleted.

6. A photovoltaic panel component segmentation system, used to execute the photovoltaic panel component segmentation method according to claim 1, characterized in that: The system includes: a photovoltaic area segmentation module, a mapping module, a straight line segment detection module, a slope acquisition module, a target straight line segment acquisition module and a segmentation module; The photovoltaic area segmentation module is used to perform deep learning semantic segmentation on the photovoltaic panel infrared thermal image to obtain all photovoltaic panel areas in the photovoltaic panel infrared thermal image; The mapping module is used to map each photovoltaic panel area to the photovoltaic panel infrared thermal image to obtain the infrared photovoltaic panel area corresponding to each photovoltaic panel area; The straight line segment detection module is used to perform edge detection and straight line detection on each of the infrared photovoltaic panel areas to obtain straight line segments in each of the infrared photovoltaic panel areas; The slope acquisition module is used to acquire a first slope of the straight line segment and a second slope of a broadside of a photovoltaic panel in the infrared photovoltaic panel area; The target straight line segment acquisition module is used to acquire a target straight line segment according to the first slope and the second slope; the target straight line segment is a straight line segment whose difference between the first slope and the second slope is within a first preset threshold; The segmentation module is used to segment the infrared photovoltaic panel area using the target straight line segment to obtain photovoltaic panel components in the infrared photovoltaic panel area.

7. A photovoltaic panel assembly segmentation device, comprising a memory and a processor, characterized in that: The memory stores a computer program, and when the computer program is executed by the processor, the processor is caused to perform the steps of the method according to any one of claims 1 to 5.

8. A computer-readable storage medium storing a computer program, characterized in that: When the computer program is executed by a processor, the processor is caused to perform the steps of the method according to any one of claims 1 to 5.

Images