Photovoltaic panel feature detection method and device, electronic equipment and storage medium

Abstract

The invention discloses a photovoltaic panel feature detection method and device, electronic equipment and a storage medium. The method comprises the following steps: acquiring a photovoltaic panel image collected by cleaning equipment at the current moment based on a binocular depth camera carried on the cleaning equipment; the cleaning equipment is used for cleaning the photovoltaic panel according to a preset route; performing feature recognition on a target object in the photovoltaic panel image to obtain parameter information of the target object; the parameter information comprises an object type, a target area and a first confidence coefficient; based on the object type of the target object, matching a feature detection algorithm for the target object, processing the feature information of the target area based on the feature detection algorithm, and determining a second confidence coefficient of the target object; and determining a target confidence coefficient based on the first confidence coefficient and the second confidence coefficient, and determining the precision of the target object in the target area based on the target confidence coefficient. According to the method, the effectiveness of feature extraction and the overall detection accuracy are remarkably improved.

Description

Technical Field

[0001] This invention relates to the field of image processing technology, and in particular to a method, apparatus, electronic device, and storage medium for detecting features of photovoltaic panels. Background Technology

[0002] Assembling multiple photovoltaic modules into a photovoltaic panel can significantly increase power generation. However, with the continuous expansion of photovoltaic power generation scale and the diversification of deployment areas, the efficiency loss caused by dust accumulation on the photovoltaic panel surface is becoming increasingly prominent. Therefore, cleaning photovoltaic panels is essential for improving the energy efficiency of photovoltaic systems.

[0003] Currently, fully automated cleaning equipment is commonly used to clean photovoltaic panels. During the cleaning process, the edges of the photovoltaic panels and the gaps between the panels can affect the path taken by the fully automated cleaning equipment. Therefore, the fully automated cleaning equipment needs to accurately identify these features to ensure that it can adjust its direction of travel and working posture in a timely manner.

[0004] Current photovoltaic panel feature recognition methods typically identify target features by collecting images, determining the edges of photovoltaic panels and gaps between panels. However, target feature recognition suffers from limited accuracy and frequent false detections. Summary of the Invention

[0005] This invention provides a photovoltaic panel feature detection method, device, electronic device, and storage medium to significantly improve the effectiveness of feature extraction and overall detection accuracy.

[0006] According to one aspect of the present invention, a method for detecting features of a photovoltaic panel is provided, the method comprising: The cleaning equipment uses a binocular depth camera to acquire images of the photovoltaic panels at the current moment; the cleaning equipment is used to clean the photovoltaic panels according to a preset route. Feature recognition is performed on the target object in the photovoltaic panel image to obtain parameter information of the target object; the parameter information includes object type, target region, and first confidence level. Based on the object type of the target object, a feature detection algorithm for the target object is matched, and the feature information of the target region is processed based on the feature detection algorithm to determine the second confidence level of the target object; A target confidence level is determined based on the first confidence level and the second confidence level, and the accuracy of the target object in the target region is determined based on the target confidence level; the accuracy is used to characterize the probability that the target object is real.

[0007] According to another aspect of the present invention, a photovoltaic panel feature detection device is provided, the device comprising: The image acquisition module is used to acquire images of the photovoltaic panels collected by the cleaning equipment at the current moment based on the binocular depth camera mounted on the cleaning equipment; the cleaning equipment is used to clean the photovoltaic panels according to a preset route; The parameter determination module is used to perform feature recognition on the target object in the photovoltaic panel image to obtain parameter information of the target object; the parameter information includes object type, target region and first confidence level; The feature processing module is used to match a feature detection algorithm for the target object based on the object type of the target object, process the feature information of the target region based on the feature detection algorithm, and determine the second confidence level of the target object; The analysis module is used to determine a target confidence level based on the first confidence level and the second confidence level, and to determine the accuracy of the target object in the target region based on the target confidence level; the accuracy is used to characterize the probability that the target object is real.

[0008] According to another aspect of the present invention, an electronic device is provided, the electronic device comprising: At least one processor; and A memory communicatively connected to the at least one processor; wherein, The memory stores a computer program that can be executed by the at least one processor, which enables the at least one processor to perform the photovoltaic panel feature detection method according to any embodiment of the present invention.

[0009] According to another aspect of the present invention, a computer-readable storage medium is provided, the computer-readable storage medium storing computer instructions for causing a processor to execute and implement the photovoltaic panel feature detection method according to any embodiment of the present invention.

[0010] The technical solution of this invention, based on a binocular depth camera mounted on a cleaning device, acquires images of photovoltaic panels collected by the cleaning device at the current moment. This enables real-time acquisition of images of photovoltaic panels ahead while cleaning the panels. Preliminary feature identification is performed on target objects in the photovoltaic panel images to obtain the object type, target region, and first confidence level. Further, based on the object type, a feature detection algorithm is matched to the target object. Then, based on the feature detection algorithm, the feature information of the target region is processed to determine the second confidence level of the target object. This allows for a more precise analysis of the features of the target region, thereby determining the target confidence level based on the first and second confidence levels. The target confidence level accurately reflects the accuracy of the target object in the target region, significantly improving the effectiveness of feature extraction and overall detection accuracy.

[0011] It should be understood that the description in this section is not intended to identify key or essential features of the embodiments of the present invention, nor is it intended to limit the scope of the invention. Other features of the invention will become readily apparent from the following description. Attached Figure Description

[0012] To more clearly illustrate the technical solutions in the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0013] Figure 1 This is a flowchart of a photovoltaic panel feature detection method provided according to an embodiment of the present invention; Figure 2 This is a flowchart of another photovoltaic panel feature detection method provided according to an embodiment of the present invention; Figure 3 This is a flowchart of another photovoltaic panel feature detection method provided according to an embodiment of the present invention; Figure 4 This is a schematic diagram of the structure of a photovoltaic panel feature detection device according to an embodiment of the present invention; Figure 5 This is a schematic diagram of the structure of an electronic device that implements the photovoltaic panel feature detection method according to an embodiment of the present invention. Detailed Implementation

[0014] To enable those skilled in the art to better understand the present invention, the technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments of the present invention. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort should fall within the scope of protection of the present invention.

[0015] It should be noted that the terms "first," "second," etc., in the specification, claims, and accompanying drawings of this invention are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate so that the embodiments of the invention described herein can be implemented in orders other than those illustrated or described herein. Furthermore, the terms "comprising" and "having," and any variations thereof, are intended to cover a non-exclusive inclusion; for example, a process, method, system, product, or apparatus that comprises a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to such processes, methods, products, or apparatus.

[0016] Example 1 Figure 1 This is a flowchart illustrating a photovoltaic panel feature detection method provided in an embodiment of the present invention. This embodiment is applicable to the identification of edge and gap features of photovoltaic panels. The method can be executed by a photovoltaic panel feature detection device, which can be implemented in hardware and / or software. This photovoltaic panel feature detection device can be configured in any electronic device with network communication capabilities. Figure 1 As shown, the photovoltaic panel feature detection method of the present invention may include the following process: S110. Based on the binocular depth camera mounted on the cleaning equipment, the image of the photovoltaic panel collected by the cleaning equipment at the current moment is obtained; the cleaning equipment is used to clean the photovoltaic panel according to a preset route.

[0017] The preset route can be understood as the movement route configured for the cleaning equipment during the cleaning process of the photovoltaic panel. It can be adjusted at any time during the subsequent cleaning process. For example, it can be adaptively adjusted according to various information in the photovoltaic panel image acquired in real time according to the present invention, so as to ensure that the cleaning equipment can clean the photovoltaic panel safely and efficiently.

[0018] S120. Perform feature recognition on the target object in the photovoltaic panel image to obtain the parameter information of the target object; the parameter information includes the object type, target area and first confidence level.

[0019] A photovoltaic (PV) panel can be composed of multiple identical photovoltaic modules. Each PV panel typically consists of an even number of photovoltaic modules, forming a rectangular PV panel. Gaps exist between the connected photovoltaic modules; these are called PV panel gaps.

[0020] Specifically, to ensure the cleaning equipment can safely clean photovoltaic panels, it is necessary to detect the edges of target objects and / or gaps in the photovoltaic panels. This allows the cleaning equipment to be guided safely based on the feature results in the detected photovoltaic panel images. Correspondingly, feature recognition of target objects in the photovoltaic panel images can determine the object types included in the images and mark the corresponding target areas. This allows for further feature analysis in those areas, ensuring the accuracy of the analysis. The target edge is the edge of the photovoltaic panel directly opposite the direction of the cleaning equipment's movement. Based on the above embodiments, optionally, performing feature recognition on the target object in the photovoltaic panel image to obtain the parameter information of the target object may include steps A1-A2: Step A1: Based on the object recognition model, perform feature recognition on the target object in the photovoltaic panel image to obtain the object feature data and feature confidence level corresponding to the target object.

[0021] Step A2: If the feature confidence level is greater than the first preset confidence level, then the object feature data is labeled with a data frame to determine the target area of ​​the target object, and the feature confidence level greater than the first preset confidence level is determined as the first confidence level; based on the object feature data, the object type of the target object is determined.

[0022] The first pre-set confidence level involves setting different confidence levels for different target objects to effectively distinguish between target objects of different types. The target region can be understood as the region formed by labeling the minimum bounding rectangle of the identified target objects.

[0023] Specifically, the object recognition model can be a model capable of identifying target objects. For example, if the object recognition model is based on the YOLOv5 algorithm, the input can be a photovoltaic panel image. It extracts the object feature data corresponding to the target object and calculates the feature confidence score. If the feature confidence score is greater than a first preset confidence score, the object feature data is labeled with a data frame to determine the target region of the target object, thus outputting the target region of the target object. If the feature confidence score is less than the first preset confidence score for that object type, it indicates that the target region of the target object is unreliable, and no output is made. Further, feature analysis is performed on the object feature data of the target region to obtain the object type of the target object.

[0024] In this embodiment of the invention, feature recognition of target objects in a target image is performed based on an object recognition model to obtain object feature data corresponding to the target object, so as to accurately identify the feature information of the target object. At the same time, a first feature confidence level of the object feature data is determined. If the feature confidence level is greater than the first preset confidence level, the object feature data is labeled with a data frame to determine the target region of the target object, thereby improving the accuracy of the target region determination of the target object. Based on the analysis of object feature data, the object type of the target object is determined more accurately.

[0025] S130. Based on the object type of the target object, match the feature detection algorithm for the target object, process the feature information of the target region based on the feature detection algorithm, and determine the second confidence level of the target object.

[0026] Feature detection algorithms can be understood as algorithms that accurately identify features of different object types in order to accurately detect key points in images that are unique and identifiable.

[0027] Specifically, for target objects of different object types, in order to ensure accurate determination of the object types included in the photovoltaic panel image, feature detection algorithms are matched to the target objects according to different object types.

[0028] Furthermore, after determining the second confidence level of the target object, if the second confidence level is greater than the second preset confidence level, it is determined that the target region contains more target region features and can be further analyzed and processed. If the second confidence level is less than or equal to the second preset confidence level, it indicates that the target region is unreliable, so the target region is deleted and subsequent operations are not performed.

[0029] S140. Determine the target confidence level based on the first confidence level and the second confidence level, and determine the accuracy of the target object in the target area based on the target confidence level; the accuracy level is used to characterize the probability that the target object is real.

[0030] Specifically, the first and second confidence levels are weighted to determine the target confidence level. If the target confidence level is greater than the third preset confidence level, the target area is determined to be valid, meaning the target area includes the target object. The dual detection of the first and second confidence levels ensures accurate detection of the target object in the target area and is more conducive to adjusting the cleaning route of the cleaning equipment based on the location information of the target object, ensuring a more accurate cleaning route.

[0031] The technical solution of this invention, based on a binocular depth camera mounted on a cleaning device, acquires images of photovoltaic panels collected by the cleaning device at the current moment. This enables real-time acquisition of images of photovoltaic panels ahead while cleaning the panels. Preliminary feature identification is performed on the target objects in the photovoltaic panel images to obtain the object type, target region, and first confidence level. Further, based on the object type, a feature detection algorithm is matched to the target object. Then, the feature information of the target region is processed based on the feature detection algorithm to determine the second confidence level of the target object. This allows for a more precise analysis of the features of the target region, thereby determining the target confidence level based on the first and second confidence levels. The target confidence level accurately reflects the accuracy of the target object in the target region, significantly improving the effectiveness of feature extraction and overall detection accuracy.

[0032] Example 2 Figure 2 This is a flowchart of another photovoltaic panel feature detection method provided by an embodiment of the present invention. The technical solution of this embodiment further optimizes the process of S130 in the aforementioned embodiments based on the above embodiments. This embodiment can be combined with various optional solutions in one or more of the above embodiments. Figure 2 As shown, the photovoltaic panel feature detection method includes: S210. Based on the binocular depth camera mounted on the cleaning equipment, the image of the photovoltaic panel collected by the cleaning equipment at the current moment is obtained; the cleaning equipment is used to clean the photovoltaic panel according to a preset route.

[0033] S220. Perform feature recognition on the target object in the photovoltaic panel image to obtain the parameter information of the target object. Based on the object type of the target object, match the feature detection algorithm for the target object. The parameter information includes the object type, target region and first confidence level.

[0034] S230. If the target object is the target edge, the feature information of the target region is processed based on the feature detection algorithm to determine the second confidence of the target object. The target confidence is determined based on the first confidence and the second confidence. The accuracy of the target object in the target region is determined based on the target confidence. The feature information of the target region includes depth information.

[0035] Specifically, the central region of the target area is determined according to the preset feature extraction conditions; the preset feature extraction conditions are used to describe the rule information for extracting the central region; the edge pixels of the target edge are determined based on the depth information of the central region; and the second confidence level of the target object is determined based on the number of edge pixels and the total number of pixels in the central region.

[0036] Generally, the target edge of a photovoltaic panel is made of a fixed material, such as aluminum strips. Therefore, the target area will contain many aluminum strip feature points. Since the density of points in the central area is higher, the central area of ​​the target area can be determined according to preset feature extraction conditions, so that the central area contains more feature information of the target edge. Furthermore, since the features of the target edge are significantly different from those of other areas, the depth information will be significantly different. Therefore, the depth information of the central area can be analyzed to better determine the edge pixels of the target edge. Then, the proportion of the number of edge pixels to the total number of pixels in the central area is used to determine the second confidence level of the target object. That is, the higher the proportion of the number of edge pixels to the total number of pixels in the central area, the higher the second confidence level, and vice versa.

[0037] Optionally, determining the central region of the target region according to preset feature extraction conditions may include: determining the width and height information of the target region; and determining the first coordinate information of the upper left corner and the second coordinate information of the lower right corner of the target region; and determining the central region of the target region based on the width information, height information, first coordinate information, and second coordinate information; the central region may be represented by the following formula: ; in, This is the first coordinate information; This is the second coordinate information; W box and H box These are the width and height information of the target area, respectively; the width information of the target area is the difference in the horizontal coordinate between the second coordinate information and the first coordinate information; the height information of the target area is the difference in the vertical coordinate between the second coordinate information and the first coordinate information; that is... ; .

[0038] Optionally, in this embodiment of the invention, determining the edge pixels of the target edge based on the depth information of the central region may include: performing gradient calculation on the depth information of the central region to determine multiple gradient magnitudes of the central region; if the gradient magnitude is greater than a preset magnitude, then the central region corresponding to the gradient magnitude greater than the preset magnitude is determined as a depth abrupt change region; and the pixels of each depth abrupt change region are determined as edge pixels of the target edge. The gradient change in depth information can reflect the changes on the surface of the photovoltaic panel, and the edge of the photovoltaic panel is generally made of aluminum strip material, which is significantly different from other features. Therefore, performing gradient calculation on the depth information of the central region can accurately distinguish the target edge of the photovoltaic panel, achieving rapid and accurate positioning of the target edge.

[0039] Furthermore, the ratio of the number of edge pixels to the total number of pixels in the central region is used to determine the proportion of edge pixels. Based on this proportion, a second confidence level for the target object is determined. Then, the first and second confidence levels are combined to determine the target confidence level, ensuring that the target confidence level can more accurately reflect the target edges contained in the target region. In other words, the accuracy of determining the target object in the target region based on the target confidence level is achieved.

[0040] Optionally, after determining the central region corresponding to the gradient magnitude greater than a preset magnitude as the depth mutation region, the method further includes: performing morphological processing on each depth mutation region to obtain an updated depth mutation region, thereby enhancing each region and ensuring the continuity of each edge, improving the accuracy of the depth mutation region, and ensuring the accuracy of the extracted pixels in each depth mutation region.

[0041] S240. If the target object is a gap in a photovoltaic panel, the target area is expanded according to a preset expansion ratio to obtain an expanded area; based on the height information of the expanded area, the second confidence level of the target object is determined; based on the first confidence level and the second confidence level, the target confidence level is determined; based on the target confidence level, the accuracy of the target object in the target area is determined; the feature information of the target area includes depth information.

[0042] The preset expansion ratio can be understood as the number of pixels by which the boundary of the target area is expanded; for example, the preset expansion ratio can be 10 pixels. There is a preset correlation between the height information of the expanded area and the second confidence level of the target object.

[0043] Specifically, expanding the target area according to a preset expansion ratio to obtain the expanded area can be achieved by: using the boundary of the target area as a reference, expanding each boundary outwards by a preset expansion ratio to form an expanded area, ensuring that it contains more information and guaranteeing the accuracy of subsequent detection. Then, the height information of the expanded area is determined. Based on the height information of the expanded area and a preset correlation, a second confidence level of the target object is determined. Based on the first and second confidence levels, the target confidence level is determined. Furthermore, based on the target confidence level, the accuracy of the target object in the target area is determined.

[0044] Optionally, determining the second confidence level of the target object based on the height information of the extended region may include: determining the second confidence level using the following formula. This enables the quantitative determination of the second confidence level. ; in, α represents the height information; h is the average height of the photovoltaic panel gap; α is the height tolerance coefficient.

[0045] The technical solution of this invention uses a binocular depth camera mounted on a cleaning device to acquire images of photovoltaic panels collected by the cleaning device at the current moment. Feature recognition is performed on target objects in the photovoltaic panel images to obtain parameter information of the target objects. Based on the object type, a feature detection algorithm is matched for the target object. The parameter information includes object type, target region, and a first confidence level. Then, when the target object is a target edge, the central region of the target area is determined according to preset feature extraction conditions. The central region contains more feature information of the target edge, enabling more accurate extraction of pixels with edge features for subsequent edge feature analysis. Further, the edge pixels of the target edge are accurately determined based on the depth information of the central region, so that the second confidence level of the target object is accurately reflected by the number of edge pixels and the total number of pixels in the central region. Simultaneously, when the target object is a gap in the photovoltaic panel, the target area is expanded according to a preset expansion ratio to obtain an expanded area, ensuring that the area contains more information and avoiding the omission of gap information, which would affect the accuracy of subsequent detection. Then, the second confidence level of the target object is determined based on the height information of the expanded area. This method enables targeted feature detection for different object types, ensuring the accuracy of the second confidence level for each target object. Finally, the target confidence level is determined based on the first and second confidence levels, and the accuracy of the target object in the target region is determined based on the target confidence level. This significantly improves the effectiveness of feature extraction and the overall detection accuracy.

[0046] Example 3 Figure 3 This is a flowchart of another photovoltaic panel feature detection method provided by an embodiment of the present invention. Based on the above embodiments, the technical solution of this embodiment describes the process of determining whether the photovoltaic panel image is valid if the target object is determined to be a target edge, and if valid, then processing the feature information of the target area based on a feature detection algorithm to determine the second confidence level of the target object. This embodiment can be combined with various optional solutions in one or more of the above embodiments. Figure 3 As shown, the photovoltaic panel feature detection method includes: S310: Based on the binocular depth camera mounted on the cleaning equipment, acquire the photovoltaic panel image collected by the cleaning equipment at the current moment; the cleaning equipment is used to clean the photovoltaic panel according to a preset route.

[0047] S320. Perform feature recognition on the target object in the photovoltaic panel image to obtain the parameter information of the target object. If the target object is a target edge, match the feature detection algorithm for the target object based on the object type of the target object. The parameter information includes the object type, target region and first confidence level. The photovoltaic panel image includes depth information, which is the depth value corresponding to each pixel.

[0048] S330. The area above the bottom edge of the photovoltaic panel image at a preset height is defined as the core area, and the core area is divided into a preset number of grid areas. Among the pixels in the grid area, the pixels with a depth value greater than a first depth threshold are defined as first pixels, and the number of first pixels and the proportion of first pixels in the total number of pixels in the grid area are determined.

[0049] In this image, the positions of each object displayed correspond to the relative positions of the objects captured by the cleaning equipment as it moves forward. The preset quantity is generally at least three.

[0050] Specifically, the first depth threshold is used to distinguish whether a pixel is the maximum depth value of an edge point. That is, if the depth value of each pixel in the grid area is greater than the first depth threshold, it means that it is less representative of the feature information of the target edge.

[0051] S340. If the proportion of the first pixel is greater than or equal to the target preset proportion, then the grid area is determined to be in the first state; the first state is used to describe the feature information that the grid area does not have the target edge.

[0052] The target preset ratio can be understood as the threshold percentage of pixels in the grid area that do not contain feature information of the target edge. If the percentage exceeds this threshold, the grid area is considered invalid. The target preset ratio can be set according to actual needs, and is generally set to 0.8.

[0053] S350. If the proportion of the first pixel is less than the target preset proportion, then the pixels with depth values ​​greater than the second depth threshold in the grid area are determined as the second pixels, and the number of the second pixels is determined as the proportion of the second pixels in the total number of pixels in the grid area; the first depth threshold is less than the second depth threshold.

[0054] Specifically, if the proportion of the first pixel is less than the target preset proportion, it means that the grid area needs to be analyzed again to accurately determine whether the grid area contains enough feature information of the target edge.

[0055] S360. Based on the proportion of the second pixel and the average depth value of the grid region, determine the state information of the grid region; based on the state information of all grid regions, determine whether to process the feature information of the target region based on the feature detection algorithm, and determine the second confidence of the target object; the state information is used to describe the feature proportion of the feature information of the target edge in the grid region.

[0056] Specifically, when the proportion of the second pixel is greater than the preset proportion threshold and the average depth value is greater than or equal to the third depth threshold, the state information of the grid region is determined to be the second state; the feature proportion of the second state used to describe the feature information of the target edge in the grid region is the first feature proportion.

[0057] If the average depth value is less than the third depth threshold, the state information of the grid region is determined to be the third state; the feature proportion of the third state used to describe the feature information of the target edge in the grid region is the second feature proportion; the first feature proportion is greater than the second feature proportion; If the proportion of the second pixel is greater than the preset proportion threshold, and the average depth value is greater than or equal to the fourth depth threshold, then the state information of the grid area is determined to be the first state; the fourth depth threshold is greater than the third depth threshold.

[0058] Furthermore, if the number of second states is greater than or equal to a preset number, then the feature information of the target region is processed based on the feature detection algorithm to determine the second confidence level of the target object; otherwise, the target region is deleted and no further analysis is performed on the target region.

[0059] Optionally, in an embodiment of the present invention, determining the state information of a grid region based on the proportion of second pixels and the average depth value of the grid region may include: determining a third confidence level of the grid region based on the proportion of second pixels; the third confidence level is used to reflect the feature proportion of the feature information of the target edge in the grid region; if the third confidence level is greater than a preset confidence threshold and the average depth value is greater than or equal to the third depth threshold, then the state information of the grid region is determined to be a second state; the second state is used to describe the feature proportion of the feature information of the target edge in the grid region as a first feature proportion; if the average depth value is less than the third depth threshold, then the state information of the grid region is determined to be a third state; the third state is used to describe the feature proportion of the feature information of the target edge in the grid region as a second feature proportion; the first feature proportion is greater than the second feature proportion; if the third confidence level is greater than a preset confidence threshold and the average depth value is greater than or equal to a fourth depth threshold, then the state information of the grid region is determined to be a first state; the fourth depth threshold is greater than the third depth threshold.

[0060] The third confidence level, which determines the grid region based on the proportion of the second pixel, can be expressed by the following formula: ; in, The percentage of the second pixel. This represents the third confidence level.

[0061] In this embodiment of the invention, the third confidence level of the grid region is determined based on the proportion of the second pixel points. This allows the feature proportion of the target edge in the grid region to be reflected by the third confidence level. Furthermore, the average depth value of the grid region is combined to determine the state information of the grid region. Based on the state information of all grid regions, it is determined whether to process the feature information of the target region using a feature detection algorithm, thereby determining the second confidence level of the target object. This further filters the target region before determining the second confidence level, ensuring the effectiveness of subsequent processes.

[0062] S370. If it is determined that the process of determining the second confidence level should continue, the feature information of the target area is processed based on the feature detection algorithm to determine the second confidence level of the target object. The target confidence level is determined based on the first confidence level and the second confidence level. The accuracy of the target object in the target area is determined based on the target confidence level.

[0063] The technical solution of this invention is based on a binocular depth camera mounted on a cleaning device to acquire a photovoltaic panel image collected by the cleaning device at the current moment. Feature recognition is performed on the target object in the photovoltaic panel image to obtain the parameter information of the target object. If the target object is a target edge, a feature detection algorithm is matched based on the object type of the target object. Before determining whether to process the feature information of the target area based on the feature detection algorithm and determining the second confidence level of the target object, the area above the bottom edge of the photovoltaic panel image at a preset height is first determined as the core area, and the core area is divided into a preset number of grid areas. Pixels with depth values ​​greater than a first depth threshold in the grid areas are determined as first pixels. The number of first pixels and the proportion of first pixels in the total number of pixels in the grid areas are determined. The state of each grid area, used to describe the feature proportion of the target edge feature information in the grid area, is determined by the proportion of first pixels, the preset target ratio, and / or the average depth value of the grid area. Based on the state information of all grid regions, it determines whether to process the feature information of the target region using a feature detection algorithm, and determines the second confidence level of the target object. This allows for further screening of the target region before determining the second confidence level, ensuring the effectiveness of subsequent processes. If it is determined that the second confidence level determination process should continue, the feature information of the target region is processed using a feature detection algorithm to determine the second confidence level of the target object. The target confidence level is then determined based on the first and second confidence levels, and the accuracy of the target object in the target region is determined based on the target confidence level. This significantly improves the effectiveness of feature extraction and the overall detection accuracy.

[0064] Example 4 Figure 4This is a schematic diagram of a photovoltaic panel feature detection device provided in Embodiment 3 of the present invention. This embodiment is applicable to the identification of edge features and gap features of photovoltaic panels. The photovoltaic panel feature detection device can be implemented in hardware and / or software, and can be configured in any electronic device with network communication capabilities. Figure 4 As shown, the photovoltaic panel feature detection device of the present invention includes: The image acquisition module 410 is used to acquire images of the photovoltaic panels collected by the cleaning equipment at the current moment based on the binocular depth camera mounted on the cleaning equipment; the cleaning equipment is used to clean the photovoltaic panels according to a preset route; The parameter determination module 420 is used to perform feature recognition on the target object in the photovoltaic panel image to obtain parameter information of the target object; the parameter information includes object type, target region and first confidence level; Feature processing module 430 is used to match a feature detection algorithm for the target object based on the object type of the target object, process the feature information of the target region based on the feature detection algorithm, and determine the second confidence level of the target object; Analysis module 440 is used to determine a target confidence level based on the first confidence level and the second confidence level, and to determine the accuracy of the target object in the target region based on the target confidence level; the accuracy is used to characterize the probability that the target object is real.

[0065] Based on the above embodiments, optionally, the target object includes a target edge and / or a gap in the photovoltaic panel; the target edge is the edge of the photovoltaic panel directly corresponding to the forward direction of the cleaning equipment; correspondingly, the parameter determination module is used to: perform feature recognition on the target object of the photovoltaic panel image based on the object recognition model, and obtain object feature data and feature confidence level corresponding to the target object; if the feature confidence level is greater than a first preset confidence level, then label the object feature data with a data frame, determine the target area of ​​the target object, and determine the feature confidence level greater than the first preset confidence level as the first confidence level; and determine the object type of the target object based on the object feature data.

[0066] Based on the above embodiments, optionally, if the target object is a target edge, the feature information of the target region includes depth information; the feature processing module includes a first processing unit, which is used to: determine the central region of the target region according to preset feature extraction conditions; the preset feature extraction conditions are used to describe the rule information for extracting the central region; determine the edge pixels of the target edge based on the depth information of the central region; and determine the second confidence level of the target object based on the number of edge pixels and the total number of pixels in the central region.

[0067] Based on the above embodiments, optionally, the first processing unit includes an edge pixel point determination subunit, which is used to perform gradient calculation on the depth information of the central region to determine multiple gradient magnitudes of the central region; if the gradient magnitude is greater than a preset magnitude, the central region corresponding to the gradient magnitude greater than the preset magnitude is determined as a depth abrupt change region; and the pixels of each depth abrupt change region are determined as edge pixels of the target edge.

[0068] Based on the above embodiments, optionally, the edge pixel point determination subunit is further configured to, after determining the central region corresponding to the gradient magnitude greater than the preset magnitude as the depth abrupt change region, perform morphological processing on each depth abrupt change region to obtain the updated depth abrupt change region.

[0069] Based on the above embodiments, optionally, if the target object is a gap in a photovoltaic panel, the feature processing module includes a second processing unit, which is used to: expand the target area according to a preset expansion ratio to obtain an expanded area; and determine a second confidence level of the target object based on the height information of the expanded area.

[0070] Based on the above embodiments, optionally, determining the second confidence level of the target object based on the height information of the extended region includes: determining the second confidence level using the following formula. : ; in, α represents the height information; h is the average height of the photovoltaic panel gap; α is the height tolerance coefficient.

[0071] Based on the above embodiments, optionally, if the target object is a target edge, the photovoltaic panel image includes depth information, the depth information being the depth value corresponding to each pixel, and the photovoltaic panel feature detection device further includes a filtering module, which is used to: before processing the feature information of the target region based on the feature detection algorithm to determine the second confidence level of the target object, determine the region above the bottom edge of the photovoltaic panel image at a preset height as the core region, and divide the core region into a preset number of grid regions; determine the pixels with depth values ​​greater than a first depth threshold among the pixels in the grid regions as first pixels, and determine the first pixel ratio of the number of first pixels to the total number of pixels in the grid regions; if the first pixel ratio is greater than or equal to a target preset ratio, then determine the grid region as a first pixel. The grid region is in a first state; the first state is used to describe the absence of target edge feature information in the grid region; if the proportion of the first pixel is less than the target preset proportion, then the pixels with depth values ​​greater than a second depth threshold in the grid region are determined as second pixels, and the number of second pixels is determined to be the second pixel proportion of the total number of pixels in the grid region; the first depth threshold is less than the second depth threshold; based on the proportion of the second pixel and the average depth value of the grid region, the state information of the grid region is determined; the state information is used to describe the feature proportion of target edge feature information in the grid region; based on the state information of all the grid regions, it is determined whether to process the feature information of the target region based on the feature detection algorithm, and the second confidence level of the target object is determined.

[0072] Based on the above embodiments, optionally, the filtering module includes a state information determination unit, which is used to: determine a third confidence level of the grid region based on the proportion of the second pixel points; the third confidence level is used to reflect the feature proportion of the feature information of the target edge in the grid region; if the third confidence level is greater than a preset confidence threshold, and the average depth value is greater than or equal to the third depth threshold, then the state information of the grid region is determined to be a second state; the second state is used to describe the feature proportion of the feature information of the target edge in the grid region as a first feature proportion; if the average depth value is less than the third depth threshold, then the state information of the grid region is determined to be a third state; the third state is used to describe the feature proportion of the feature information of the target edge in the grid region as a second feature proportion; the first feature proportion is greater than the second feature proportion; if the third confidence level is greater than a preset confidence threshold, and the average depth value is greater than or equal to a fourth depth threshold, then the state information of the grid region is determined to be a first state; the fourth depth threshold is greater than the third depth threshold.

[0073] Based on the above embodiments, optionally, the filtering module includes a filtering unit, which is used to: if the number of second states is greater than or equal to a preset number, determine to process the feature information of the target region based on the feature detection algorithm to determine the second confidence level of the target object.

[0074] The photovoltaic panel feature detection device provided in this embodiment of the invention can execute the photovoltaic panel feature detection method provided in any embodiment of the invention, and has the corresponding functional modules and beneficial effects of the method.

[0075] Example 5 According to embodiments of this disclosure, this disclosure also provides an electronic device, a readable storage medium, and a computer program product.

[0076] Figure 5 A schematic diagram of an electronic device that can be used to implement the photovoltaic panel feature detection method of embodiments of the present invention is shown. The electronic device is intended to represent various forms of digital computers, such as laptop computers, desktop computers, workbenches, personal digital assistants, servers, blade servers, mainframe computers, and other suitable computers. The electronic device can also represent various forms of mobile devices, such as personal digital processors, cellular phones, smartphones, wearable devices (such as helmets, glasses, watches, etc.), and other similar computing devices. The components shown herein, their connections and relationships, and their functions are merely illustrative and are not intended to limit the implementation of the invention described and / or claimed herein.

[0077] like Figure 5 As shown, the electronic device 10 includes at least one processor 11 and a memory, such as a read-only memory (ROM) 12 or a random access memory (RAM) 13, communicatively connected to the at least one processor 11. The memory stores computer programs executable by the at least one processor. The processor 11 can perform various appropriate actions and processes based on the computer program stored in the ROM 12 or loaded from storage unit 18 into the RAM 13. The RAM 13 may also store various programs and data required for the operation of the electronic device 10. The processor 11, ROM 12, and RAM 13 are interconnected via a bus 14. An input / output (I / O) interface 15 is also connected to the bus 14.

[0078] Multiple components in electronic device 10 are connected to I / O interface 15, including: input unit 16, such as keyboard, mouse, etc.; output unit 17, such as various types of displays, speakers, etc.; storage unit 18, such as disk, optical disk, etc.; and communication unit 19, such as network card, modem, wireless transceiver, etc. Communication unit 19 allows electronic device 10 to exchange information / data with other devices through computer networks such as the Internet and / or various telecommunications networks.

[0079] Processor 11 can be a variety of general-purpose and / or special-purpose processing components with processing and computing capabilities. Some examples of processor 11 include, but are not limited to, a central processing unit (CPU), a graphics processing unit (GPU), various special-purpose artificial intelligence (AI) computing chips, various processors running machine learning model algorithms, a digital signal processor (DSP), and any suitable processor, controller, microcontroller, etc. Processor 11 performs the various methods and processes described above, such as photovoltaic panel feature detection methods.

[0080] In some embodiments, the photovoltaic panel feature detection method may be implemented as a computer program tangibly contained in a computer-readable storage medium, such as storage unit 18. In some embodiments, part or all of the computer program may be loaded and / or installed on electronic device 10 via read-only memory (ROM) 12 and / or communication unit 19. When the computer program is loaded into random access memory (RAM) 13 and executed by processor 11, one or more steps of the photovoltaic panel feature detection method described above may be performed. Alternatively, in other embodiments, processor 11 may be configured to perform the photovoltaic panel feature detection method by any other suitable means (e.g., by means of firmware).

[0081] Various embodiments of the systems and techniques described above herein can be implemented in digital electronic circuit systems, integrated circuit systems, field-programmable gate arrays (FPGAs), application-specific integrated circuits (ASICs), application-specific standard products (ASSPs), system-on-a-chip (SoCs), complex programmable logic devices (CPLDs), computer hardware, firmware, software, and / or combinations thereof. These various embodiments may include implementations in one or more computer programs that can be executed and / or interpreted on a programmable system including at least one programmable processor, which may be a dedicated or general-purpose programmable processor, capable of receiving data and instructions from a storage system, at least one input device, and at least one output device, and transmitting data and instructions to the storage system, the at least one input device, and the at least one output device.

[0082] Computer programs used to implement the methods of the present invention may be written in any combination of one or more programming languages. These computer programs may be provided to a processor of a general-purpose computer, a special-purpose computer, or other programmable data processing device, such that when executed by the processor, the computer programs cause the functions / operations specified in the flowcharts and / or block diagrams to be performed. The computer programs may be executed entirely on a machine, partially on a machine, or as a standalone software package, partially on a machine and partially on a remote machine, or entirely on a remote machine or server.

[0083] In the context of this invention, a computer-readable storage medium can be a tangible medium that may contain or store a computer program for use by or in conjunction with an instruction execution system, apparatus, or device. A computer-readable storage medium may include, but is not limited to, electronic, magnetic, optical, electromagnetic, infrared, or semiconductor systems, apparatus, or devices, or any suitable combination thereof. Alternatively, a computer-readable storage medium may be a machine-readable signal medium. More specific examples of machine-readable storage media include electrical connections based on one or more wires, portable computer disks, hard disks, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fibers, portable compact disk read-only memory (CD-ROM), optical storage devices, magnetic storage devices, or any suitable combination thereof.

[0084] To provide interaction with a user, the systems and techniques described herein can be implemented on an electronic device having: a display device (e.g., a CRT (cathode ray tube) or LCD (liquid crystal display) monitor) for displaying information to the user; and a keyboard and pointing device (e.g., a mouse or trackball) through which the user provides input to the electronic device. Other types of devices can also be used to provide interaction with the user; for example, feedback provided to the user can be any form of sensory feedback (e.g., visual feedback, auditory feedback, or tactile feedback); and input from the user can be received in any form (including sound input, voice input, or tactile input).

[0085] The systems and technologies described herein can be implemented in computing systems that include backend components (e.g., as data servers), or middleware components (e.g., application servers), or frontend components (e.g., user computers with graphical user interfaces or web browsers through which users can interact with implementations of the systems and technologies described herein), or any combination of such backend, middleware, or frontend components. The components of the system can be interconnected via digital data communication of any form or medium (e.g., communication networks). Examples of communication networks include local area networks (LANs), wide area networks (WANs), blockchain networks, and the Internet.

[0086] A computing system can include clients and servers. Clients and servers are generally located far apart and typically interact through communication networks. The client-server relationship is created by computer programs running on the respective computers and having a client-server relationship with each other. The server can be a cloud server, also known as a cloud computing server or cloud host, which is a hosting product within the cloud computing service system to address the shortcomings of traditional physical hosts and VPS services, such as high management difficulty and weak business scalability.

[0087] It should be understood that the various forms of processes shown above can be used, with steps reordered, added, or deleted. For example, the steps described in this invention can be executed in parallel, sequentially, or in different orders, as long as the desired result of the technical solution of this invention can be achieved, and this is not limited herein.

[0088] The specific embodiments described above do not constitute a limitation on the scope of protection of this invention. Those skilled in the art should understand that various modifications, combinations, sub-combinations, and substitutions can be made according to design requirements and other factors. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of this invention should be included within the scope of protection of this invention.

Claims

1. A method for detecting features of a photovoltaic panel, characterized in that, The method includes: The cleaning equipment uses a binocular depth camera to acquire images of the photovoltaic panels at the current moment; the cleaning equipment is used to clean the photovoltaic panels according to a preset route. Feature recognition is performed on the target object in the photovoltaic panel image to obtain parameter information of the target object; the parameter information includes object type, target region, and first confidence level. Based on the object type of the target object, a feature detection algorithm for the target object is matched, and the feature information of the target region is processed based on the feature detection algorithm to determine the second confidence level of the target object; A target confidence level is determined based on the first confidence level and the second confidence level, and the accuracy of the target object in the target region is determined based on the target confidence level. The accuracy is used to characterize the probability that the target object is real.

2. The method according to claim 1, characterized in that, The target object includes the target edge and / or the gap in the photovoltaic panel; the target edge is the edge of the photovoltaic panel that corresponds to the direction of travel of the cleaning equipment. Accordingly, feature recognition is performed on the target object in the photovoltaic panel image to obtain the parameter information of the target object, including: Based on the object recognition model, the target object in the photovoltaic panel image is identified by feature recognition, and the object feature data and feature confidence level corresponding to the target object are obtained. If the feature confidence level is greater than the first preset confidence level, then the object feature data is labeled with a data frame to determine the target region of the target object, and the feature confidence level that is greater than the first preset confidence level is determined as the first confidence level; Based on the object feature data, the object type of the target object is determined.

3. The method according to claim 2, characterized in that, If the target object is a target edge, then the feature information of the target region includes depth information; the step of processing the feature information of the target region based on the feature detection algorithm to determine the second confidence level of the target object includes: The central region of the target region is determined according to preset feature extraction conditions; the preset feature extraction conditions are used to describe the rule information for extracting the central region. Based on the depth information of the central region, the edge pixels of the target edge are determined; The second confidence level of the target object is determined based on the number of edge pixels and the total number of pixels in the central region.

4. The method according to claim 3, characterized in that, The step of determining the edge pixels of the target edge based on the depth information of the central region includes: Gradient calculation is performed on the depth information of the central region to determine multiple gradient magnitudes of the central region; If the gradient magnitude is greater than a preset magnitude, then the central region corresponding to the gradient magnitude that is greater than the preset magnitude is determined as a deep abrupt change region; The pixels in each depth abrupt change region are determined as the edge pixels of the target edge.

5. The method according to claim 4, characterized in that, After determining the central region corresponding to the gradient magnitude greater than a preset magnitude as the deep abrupt change region, the method further includes: Morphological processing was performed on each deep mutation region to obtain the updated deep mutation regions.

6. The method according to claim 2, characterized in that, If the target object is a gap in a photovoltaic panel, then the step of processing the feature information of the target area based on the feature detection algorithm to determine the second confidence level of the target object includes: The target area is expanded according to a preset expansion ratio to obtain an expanded area; Based on the height information of the extended region, a second confidence level for the target object is determined.

7. The method according to claim 6, characterized in that, Determining the second confidence level of the target object based on the height information of the extended region includes: The second confidence level is determined using the following formula. : ； in, α represents the height information; h is the average height of the photovoltaic panel gap; α is the height tolerance coefficient.

8. The method according to claim 2, characterized in that, If the target object is a target edge, the photovoltaic panel image includes depth information, wherein the depth information is the depth value corresponding to each pixel. Before processing the feature information of the target region based on the feature detection algorithm to determine the second confidence level of the target object, the method further includes: The area above the bottom edge of the photovoltaic panel image at a predetermined height is defined as the core area, and the core area is divided into a predetermined number of grid areas; Pixels whose depth values ​​in the grid region are greater than a first depth threshold are identified as first pixels, and the number of first pixels is determined as the proportion of first pixels to the total number of pixels in the grid region. If the proportion of the first pixel is greater than or equal to the target preset proportion, then the grid region is determined to be in the first state; the first state is used to describe the feature information that the grid region does not have the target edge. If the proportion of the first pixel is less than the target preset proportion, then the pixels in the grid area whose depth values ​​are greater than the second depth threshold are determined as the second pixel, and the proportion of the number of the second pixel to the total number of pixels in the grid area is determined; the first depth threshold is less than the second depth threshold. Based on the proportion of the second pixel and the average depth value of the grid region, the state information of the grid region is determined; the state information is used to describe the feature proportion of the feature information of the target edge in the grid region. Based on the state information of all the grid regions, determine whether to process the feature information of the target region based on the feature detection algorithm, and determine the second confidence level of the target object.

9. The method according to claim 8, characterized in that, Based on the proportion of the second pixel and the average depth value of the grid region, the state information of the grid region is determined, including: The third confidence level of the grid region is determined based on the proportion of the second pixel; the third confidence level is used to reflect the feature proportion of the target edge feature information in the grid region; If the third confidence level is greater than a preset confidence threshold, and the average depth value is greater than or equal to the third depth threshold, then the state information of the grid region is determined to be the second state; the second state is used to describe the feature proportion of the feature information of the target edge in the grid region as the first feature proportion; If the average depth value is less than the third depth threshold, the state information of the grid region is determined to be the third state; the third state is used to describe the feature proportion of the feature information of the target edge in the grid region as the second feature proportion; the first feature proportion is greater than the second feature proportion; If the third confidence level is greater than a preset confidence threshold, and the average depth value is greater than or equal to a fourth depth threshold, then the state information of the grid region is determined to be the first state; the fourth depth threshold is greater than the third depth threshold.

10. The method according to claim 9, characterized in that, Based on the state information of all the grid regions, determine whether to process the feature information of the target region based on the feature detection algorithm, and determine the second confidence level of the target object, including: If the number of second states is greater than or equal to a preset number, then the feature information of the target region is processed based on the feature detection algorithm to determine the second confidence level of the target object.

11. A photovoltaic panel feature detection device, characterized in that, The device includes: The image acquisition module is used to acquire images of the photovoltaic panels collected by the cleaning equipment at the current moment based on the binocular depth camera mounted on the cleaning equipment; the cleaning equipment is used to clean the photovoltaic panels according to a preset route; The parameter determination module is used to perform feature recognition on the target object in the photovoltaic panel image to obtain parameter information of the target object; the parameter information includes object type, target region and first confidence level; The feature processing module is used to match a feature detection algorithm for the target object based on the object type of the target object, process the feature information of the target region based on the feature detection algorithm, and determine the second confidence level of the target object; The analysis module is used to determine a target confidence level based on the first confidence level and the second confidence level, and to determine the accuracy of the target object in the target region based on the target confidence level; the accuracy is used to characterize the probability that the target object is real.

12. An electronic device, characterized in that, The electronic device includes: At least one processor; and A memory communicatively connected to the at least one processor; wherein, The memory stores a computer program that can be executed by the at least one processor, the computer program being executed by the at least one processor to enable the at least one processor to perform the photovoltaic panel feature detection method according to any one of claims 1-10.

13. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores computer instructions that are used to cause a processor to execute the photovoltaic panel feature detection method according to any one of claims 1-10.

Images