JP6454579B2 - Aerial image processing system and aerial image processing method - Google Patents

Description

  The present invention relates to a large-scale photovoltaic power generation system (hereinafter referred to as “mega solar”) having a solar panel output of 1 megawatt or more formed by each panel for generating power with sunlight (hereinafter referred to as “solar panel”). The present invention relates to an aerial image processing system and an aerial image processing method for performing image processing on aerial image data obtained by aerial imaging.

  In recent years, the interest in renewable energy has been increasing along with the changes in the situation surrounding energy, and the construction of mega solar has accelerated because it is expected to secure profitability under the feed-in tariff system for renewable energy. ing.

  Here, when a natural disaster such as lightning occurs, the deterioration of each solar panel forming the mega solar progresses at a stretch, and early inspection is necessary. In the past, each solar panel was manually inspected one by one, but since a huge number of solar panels are installed in mega solar, all solar panels were manually inspected. The load is excessive.

  For this reason, a technique for automatically detecting an abnormal power generation state of each solar panel forming the mega solar is known. For example, Patent Document 1 has a plurality of solar cell strings in which a plurality of solar cell modules (cells) forming a solar panel are connected in series, and a backflow prevention diode is connected to each power output terminal of the solar cell string. In a solar cell power generation system abnormality detection device, a technique is disclosed that includes a measurement means for measuring a current flowing in a backflow prevention diode, and that the measurement means is configured to be supplied with power from both ends of the backflow prevention diode. Yes.

JP 2010-263027 A

  However, since the thing of this patent document 1 detects the abnormal electric power generation state of each solar panel using an electric circuit, effective maintenance inspection cannot be performed only by this detection result. For example, even if a foreign object (eg, a grass leaf) is temporarily placed on a solar panel and the amount of current is reduced due to this, an abnormal power generation state of the solar panel is detected. Therefore, it is not possible to specify whether or not this solar panel should be subject to maintenance.

  For this reason, a method to take aerial image of mega solar image data with a small unmanned aerial vehicle called drone (hereinafter simply referred to as “aircraft”) and identify the solar panel to be maintained using the aerial image data. Can be considered. Specifically, the flying object is made to fly at an altitude that can detect the presence or absence of abnormalities on each solar panel that forms the mega solar, and the presence or absence of abnormalities is detected using aerial image data captured by this flying object. Will do.

  However, when flying an aircraft at an altitude that can detect anomalies on each solar panel, even if the presence of anomalies on each solar panel can be detected, There is a problem that the location can not be specified. This is because the mega solar is formed from a large number of solar panels having the same appearance. On the other hand, if the flying object is caused to fly at a high altitude that can identify the position of the solar panel, it becomes difficult to detect the presence or absence of an abnormality on each solar panel. In this way, there is a trade-off relationship between the point of detecting the presence or absence of abnormality on each solar panel by aerial photography and the point of specifying the location where the solar panel is located. The actual situation is that inspections are not widespread.

  Therefore, when detecting the presence or absence of abnormalities on each solar panel that forms a mega solar by aerial photography with a flying object, how to identify the location of each solar panel is an important issue. ing.

  The present invention has been made to solve the above-described problems of the prior art, and each solar panel is used to detect the presence or absence of abnormality on each solar panel forming a mega solar by aerial photography using a flying object. It is an object of the present invention to provide an aerial image processing system and an aerial image processing method capable of efficiently specifying the position where the camera is located.

  In order to solve the above-described problems and achieve the object, the invention described in claim 1 is a method of flying video data including a plurality of solar panels forming part of the mega solar while flying over the mega solar. An aerial image processing system comprising: an air vehicle equipped with an imaging device for taking an image; and an aerial image processing device for performing image processing on aerial image data included in video data taken by the aerial vehicle. The aerial image processing apparatus is cut out by the cutting unit and a cutting unit that cuts out a plurality of aerial image data from the video data so that a common solar panel is included between adjacent aerial image data. A feature point extraction unit that extracts feature points that are present in common in the adjacent aerial image data, and the cutout unit extracts the feature points based on the feature points extracted by the feature point extraction unit. Identification means for specifying a common solar panel located in each adjacent aerial image data, and a plurality of aerial images cut out by the cutting means based on the common solar panel specified by the specifying means It is characterized by comprising identification information giving means for giving identification information for uniquely identifying each solar panel included in the data.

  According to a second aspect of the present invention, in the first aspect of the present invention, the imaging device images infrared light image data and visible light image data, respectively, and the cutting means A plurality of infrared aerial image data is cut out from the infrared light video data, and a plurality of visible aerial image data is cut out from the visible light video data. A feature point is extracted from at least one of the aerial image data of light and the aerial image data of visible light, and the specifying unit includes the aerial image data of visible light or between the aerial image data of infrared light. A common solar panel is identified among the solar panels included in the aerial image data of the plurality of infrared light based on the common solar panel identified by the identification unit. Each The identification information for uniquely identifying the solar panel is given, and the identification information for uniquely identifying the solar panel is given to each of the solar panels included in the plurality of aerial image data of visible light And

  The invention according to claim 3 is the invention according to claim 2, wherein the feature point extracting means extracts the feature points from a luminance distribution in the aerial image data of the infrared light. And

  The invention according to claim 4 is the invention according to claim 1, 2 or 3, further comprising a marker member installed in the vicinity of the solar panel, wherein the feature point extracting means is the aerial image. The feature point is extracted using an image of a marker member included in the data.

  According to a fifth aspect of the present invention, in the invention according to any one of the first to fourth aspects, the cutting means includes a flying speed of the flying object, a flying altitude of the flying object, and the imaging device. The interval at which the aerial image data is cut out from the video data is calculated based on the angle of view.

  The invention according to claim 6 is the invention according to any one of claims 1 to 5, wherein the specifying means determines a positional relationship among a plurality of solar panels included in one aerial image data. The identification information providing means includes a position of each solar panel included in the plurality of aerial image data based on the common solar panel and a positional relationship between the plurality of solar panels included in the aerial image data. The relationship is determined and the identification information is given to each solar panel.

  The invention according to claim 7 is a vehicle equipped with an imaging device for aerial imaging of video data including a plurality of solar panels forming part of the mega solar while flying over the mega solar, An aerial image processing method by an aerial image processing system including an aerial image processing apparatus that performs image processing on aerial image data included in video data captured aerial by the imaging device of the flying object, A cut-out step of cutting out a plurality of aerial image data from the video data so that a common solar panel is included between the captured image data, and adjacent aerial image data cut out by the cut-out step. A feature point extracting step for extracting feature points, and adjacent aerial images cut out by the cutout step based on the feature points extracted by the feature point extraction step A specific step of identifying a common solar panel located in each of the data, and each solar included in the plurality of aerial image data cut out by the cutting step based on the common solar panel specified by the specifying step An identification information providing step for providing identification information for uniquely identifying each solar panel to each of the panels.

  According to the present invention, an image of a plurality of solar panels forming part of the mega solar while the flying object flies over the mega solar is used, and a solar common to adjacent aerial image data is used. Cut out multiple aerial image data from video data to include panels, extract feature points that are common to adjacent aerial image data, and install a common solar panel located in each adjacent aerial image data Since identification information that uniquely identifies the solar panel is assigned to each of the solar panels included in the plurality of aerial image data based on the common solar panel based on the characteristic points, When detecting the presence or absence of abnormality on each solar panel that forms a mega solar by using, to efficiently identify the location where each solar panel is located Door can be.

FIG. 1 is an explanatory diagram of aerial image processing according to the embodiment. FIG. 2 is a configuration diagram showing a system configuration of the aerial image processing system. FIG. 3 is an explanatory diagram of data stored in the storage unit of the aerial image processing apparatus. FIG. 4 is an explanatory diagram of the image data extraction interval by the extraction unit. FIG. 5 is a flowchart showing image data extraction processing by the extraction unit. FIG. 6 is a flowchart showing a processing procedure for joining image data. FIG. 7 is an explanatory diagram of mutual use of features in infrared light and visible light image data. FIG. 8 is an explanatory diagram of a specific example of giving identification information. FIG. 9 is an explanatory diagram regarding the installation of the marker member.

  Exemplary embodiments of an aerial image processing system and an aerial image processing method according to the present invention will be described below in detail with reference to the accompanying drawings.

  FIG. 1 is an explanatory diagram of aerial image processing according to the embodiment. The aerial image processing system according to the embodiment aerially captures mega solar image data using a flying object, and identifies a solar panel to be maintained using the aerial image data.

  Specifically, as shown in FIG. 1A, the flying object 30 flies over the mega solar and aerial images of video data including a plurality of solar panels 40 forming a part of the mega solar are taken. . At this time, the flying body 30 flies at an altitude that can detect the presence or absence of abnormality in the solar panel 40.

  Then, a plurality of image data is cut out from the video data captured by the flying object 30. At this time, one frame of video data is cut out as one image data. Moreover, it cuts out so that the image of a common solar panel may be included between the image data adjacent in time series among several cut-out video data.

  Furthermore, feature points that exist in common between adjacent image data are extracted, solar panels are associated with the extracted feature points, and identification information for uniquely identifying the solar panels included in the video data is given. .

  FIG. 1B shows three image data F1 to F3 cut out from the video data. Here, the video data is infrared video data, and the image data F1 to F3 are infrared light image data.

  Even if the solar panel 40 has the same appearance in the visible light image data, there may be a difference in luminance in the infrared light image data. For example, when the output differs due to variations in the performance of the solar panel 40, a difference occurs in the temperature of the solar panel 40 due to the difference in output, and this temperature difference appears as a luminance difference in the image data of infrared light. Similarly, for the solar panel 40 to be maintained, an abnormality in its operating state appears in the temperature, resulting in a luminance difference in the image data of infrared light. In the image data F <b> 1 to F <b> 3 in FIG. 1B, the difference in luminance of the solar panel 40 is indicated by shading.

  The aerial image processing system according to the present embodiment uses the distribution pattern of the brightness of the solar panel 40 in the infrared image data as a feature, and superimposes the features common to the adjacent image data in time series. Connect multiple image data. FIG. 1C shows a state in which the image data F1 to F3 are connected so that the luminance distribution patterns of the solar panel 40 match.

  In this way, a plurality of image data is cut out from the mega solar video data captured by the flying object 30, features that exist in common in adjacent image data are extracted, and the image data is connected based on the extracted features. Thus, the positional relationship of the plurality of solar panels 40 included in the mega solar can be determined, and identification information for uniquely identifying each solar panel 40 can be given.

  And by providing identification information that uniquely identifies each solar panel in this way, when detecting the presence or absence of abnormality on each solar panel forming a mega solar, the location where each solar panel is located can be identified efficiently can do.

  Next, the configuration of the aerial image processing system will be described. FIG. 2 is a configuration diagram showing a system configuration of the aerial image processing system. As shown in FIG. 2, the aerial image processing system includes a flying object 30 and an aerial image processing apparatus 10.

  The flying object 30 includes an imaging unit 31, a GPS (Global Positioning System) unit 32, a wireless communication unit 33, a plurality of rotor blades 34, a storage unit 35, and a control unit 36.

  The imaging unit 31 includes an infrared light camera and a visible light camera. The infrared light camera and the visible light camera simultaneously image the same range. The infrared light image data captured by the infrared light camera and the visible light image data captured by the visible light camera are each output to the control unit 36.

  The GPS unit 32 is a unit that receives signals from a plurality of GPS satellites and identifies the position of the device itself. The wireless communication unit 33 is an interface unit for performing wireless communication with the aerial image processing apparatus 10. The plurality of rotor blades 34 are individually rotated under the control of the control unit 36 to realize attitude control and movement of the flying object 30.

  The storage unit 35 is a storage device such as a hard disk device or a nonvolatile memory, and stores flight history data 35a, infrared video data 35b, and visible video data 35c. The flight history data 35a is data in which a flight history of the flying object 30 is recorded together with time, and position information, flight altitude, flight direction, and flight speed at each time are indicated. The infrared video data 35b is video data of infrared light captured by the infrared camera of the imaging unit 31. The visible video data 35 c is visible light video data captured by the visible light camera of the imaging unit 31.

  The control unit 36 is a control unit that controls the entire aircraft 30 and includes a flight control unit 36a and an imaging control unit 36b. The flight control unit 36 a controls the flight by controlling the plurality of rotor blades 34 using a preset flight path and position information output from the GPS unit 32.

  The imaging control unit 36b causes the imaging unit 31 to image the solar panel 40, and stores the obtained infrared video data 35b and visible video data 35c in the storage unit 35 in association with the flight history data 35a. In addition, the imaging control unit 36b transmits the flight history data 35a, the infrared video data 35b, and the visible video data 35c to the aerial image processing apparatus 10. The flight history data 35a, the infrared video data 35b, and the visible video data 35c may be transmitted to the aerial image processing apparatus 10 in real time during the flight or collectively after the flight is completed.

  The aerial image processing apparatus 10 includes a display unit 11, an input unit 12, a wireless communication unit 13, a storage unit 15, and a control unit 16. The display unit 11 is a liquid crystal panel, a display device, or the like. The input unit 12 is a keyboard, a mouse, or the like. The wireless communication unit 13 is an interface unit for performing wireless communication with the flying object 30.

  The storage unit 15 is a storage device such as a hard disk device or a nonvolatile memory, and stores panel installation data 15a, infrared image synthesis data 15b, and visible image synthesis data 15c. The panel installation data 15a is data indicating information regarding the installation position of the solar panel 40 and the manufacture of the solar panel 40.

  The infrared image synthesis data 15b is data obtained by connecting image data cut out from the infrared video data 35b and adding identification information to each solar panel 40. The visible image synthesis data 15c is data in which image data cut out from the visible video data 35c is connected and identification information is given to each solar panel 40.

  The control unit 16 is a control unit that controls the aerial image processing apparatus 10 as a whole, and includes a cutout unit 16a, a feature extraction unit 16b, a common panel specifying unit 16c, an identification information adding unit 16d, and an abnormality detection unit 16e. Actually, programs corresponding to these functional units are stored in a ROM or a non-volatile memory (not shown), and these programs are loaded into a CPU (Central Processing Unit) and executed, whereby the cutting unit 16a, Processes corresponding to the feature extraction unit 16b, the common panel specification unit 16c, the identification information addition unit 16d, and the abnormality detection unit 16e are executed.

  The cutout unit 16a is a processing unit that acquires the flight history data 35a, the infrared video data 35b, and the visible video data 35c from the flying object 30, and cuts out image data from each. As described above, the cutting unit 16a cuts out one frame of video data as one image data. Moreover, it cuts out so that the image of a common solar panel may be included between the image data adjacent in time series among several cut-out video data. Specifically, the flight altitude and the flight speed of the flying object 30 are acquired with reference to the flight history data 35a, and the image data cut-out interval is determined so that adjacent image data overlap with each other in a time series.

  The feature extraction unit 16b is a processing unit that extracts features that exist in common in adjacent image data. In infrared image data, the luminance distribution pattern of the solar panel 40 can be used as a feature. In the visible light image data, articles around the solar panel 40 can be used as features. An arbitrary algorithm may be used for extracting features from infrared light or visible light image data.

  The common panel specifying unit 16c is a processing unit that specifies the common solar panels 40 located in adjacent image data based on the features extracted by the feature extracting unit 16b. Specifically, the common panel specifying unit 16c performs image processing on each image data to recognize the solar panel 40, and determines the positional relationship of the solar panel 40 in each image data. Then, the correspondence between the image data is determined from the features that exist in common between the image data, and the common solar panel 40 is identified from the correspondence between the image data and the positional relationship of the solar panels 40 in the image data. To do.

  Image processing for recognizing the solar panel 40 includes, for example, extracting an edge, detecting a straight line, detecting a region surrounded by the straight line, and detecting the solar panel 40 when the size of the region is within a predetermined range. What is necessary is just to recognize that it is.

  The identification information adding unit 16d connects a plurality of image data based on the common solar panel 40 specified by the common panel specifying unit 16c, generates identification image data by adding identification information to each solar panel 40, and Store in the storage unit 15. Note that infrared image synthesis data 15b is generated from the infrared video data 35b, and visible image synthesis data 15c is generated from the visible video data 35c. When a plurality of image data are connected, image conversion is performed as necessary so that the shape and size of the common solar panel 40 match.

  The abnormality detection unit 16e is a processing unit that detects an abnormality using the infrared image composite data 15b and the visible image composite data 15c. A temperature abnormality can be detected from the infrared image synthesis data 15b. For example, when the temperature of the solar panel 40 is too high, it is indicated that an inappropriate resistance component is generated due to foreign matters on the solar panel 40. Further, when the temperature of the solar panel 40 is too low, it is indicated that the solar panel 40 is not operating normally. From the visible image composite data 15c, physical damage, dirt, foreign matter, etc. of the solar panel 40 can be detected.

  Further, by analyzing the infrared image composite data 15b and the visible image composite data 15c, an abnormality can be analyzed. For example, if an abnormally high temperature is detected from the infrared image composite data 15b and a foreign object is detected from the visible image composite data 15c, it can be analyzed that the temperature of the solar panel 40 has increased due to the foreign object. The abnormality detection unit 16e outputs the results of abnormality detection and analysis to the display unit 11 and the like. Even if the solar panel 40 has an abnormality detected in only one of the infrared image composite data 15b and the visible image composite data 15c, it is desirable to output the other corresponding image data together.

  Next, data stored in the storage unit 15 of the aerial image processing apparatus 10 will be described. FIG. 3 is an explanatory diagram of data stored in the storage unit 15 of the aerial image processing apparatus 10. In the panel installation data 15a shown in FIG. 3A, an installation position, an installation date, a manufacturer, a model number, a manufacturing date, and the like are associated with a solar panel ID that is identification information of the solar panel 40. The installation position specifies a point where the solar panel 40 is installed. For example, latitude and longitude may be used for specifying the point.

  In FIG. 3A, the panel installation data 15a indicates that the installation position of the solar panel 40 that is the solar panel “P0001” is “35.670986, 139.7745882”, the installation date is “20133061”, and the manufacturer It is “malso”, the model number is “slp-07”, and the manufacturing date is “20130202”.

  The infrared image synthesis data 15b shown in FIG. 3B is data in which image data cut out from the infrared video data 35b is connected and identification information is given to each solar panel 40. The infrared image composition data 15b indicates the positional relationship of the solar panels 40 forming the mega solar, and numbers (1 to 69) are assigned to the solar panels 40 as identification information.

  The visible image composite data 15c shown in FIG. 3C is data in which image data cut out from the visible video data 35c is connected and identification information is given to each solar panel 40. The visible image composition data 15c indicates the positional relationship of the solar panels 40 forming the mega solar, and numbers (1-69) are assigned to the solar panels 40 as identification information.

となる。 Next, the image data cut-out interval by the cut-out unit 16a will be described. FIG. 4 is an explanatory diagram of the image data extraction interval by the extraction unit 16a. As shown in FIG. 4, assuming that the flight altitude from the solar panel 40 to the flying object 30 is h, the flight speed is v, and the angle of view of the imaging unit 31 is θ, the image data at time t1 and the image at time t2. The extraction interval T (t2−t1) at which the overlapping ratio with the data is R is

It becomes.

  The cutout unit 16a calculates a cutout interval T at which the overlapping rate is R based on the flight altitude and the flight speed of the flying object 30 indicated in the flight history data 35a. Then, image data is cut out at intervals of T ± th, with ± th as an error range.

  FIG. 5 is a flowchart showing image data extraction processing by the extraction unit 16a. The flowchart of FIG. 5 shows a processing procedure when the nth image data Fn is cut out from the video data, and is started after the (n−1) th image data is cut out. Note that the time when the (n−1) -th image data is captured is t (n−1).

  The cutout unit 16a first reads the next frame (step S101). At this time, the imaging time of the read frame is tn. The cutout unit 16a compares tn−t (n−1) with T + th (step S102).

  If tn−t (n−1) ≦ T + th (step S102; No), the cutout unit 16a compares tn−t (n−1) with T−th (step S103).

  If tn−t (n−1) ≦ T−th (Step S103; No), since sufficient time has not elapsed since the (n−1) th image data is cut out, the process proceeds to Step S101. Read the next frame.

  If tn-t (n-1)> T-th (step S103; Yes), the cutout unit 16a determines whether or not the state of the image of the current frame is good (step S104). If the image state of the current frame is not good (step S104; No), the cutout unit 16a moves to step S101 and reads the next frame.

  If the image state of the current frame is good (step S104; Yes), the cutout unit 16a samples the current frame as the image data Fn (step S105). Then, n is incremented (step S106), and the process ends.

  If tn−t (n−1)> T + th (step S102; Yes), since T + th is exceeded without obtaining good image data, n is not sampled without sampling the image data Fn. Increment (step S106), the process is terminated. In this case, t (n-2) + T is used as t (n-1) in the next process.

  Next, image data stitching will be described. FIG. 6 is a flowchart showing a processing procedure for joining image data. First, the feature extraction unit 16b acquires one piece of image data from a plurality of pieces of image data cut out from the same video data, and sets it as base image data (step S201). Then, the feature extraction unit 16b extracts the features of the base image data (Step S202).

  Thereafter, the feature extraction unit 16b acquires the image data adjacent to the base image data as image data to be added (step S203), and extracts the features of the image data to be added (step S204).

  The common panel specifying unit 16c associates the characteristics of the base image data with the characteristics of the image data to be added (step S205). The identification information adding unit 16d converts the shape and size of the image data to be added according to the base image data (step S206), and updates the base image data by superimposing the image data to be added to the base image data (step S207). ).

  Thereafter, the identification information adding unit 16d determines whether or not other image data to be added remains, and if there is image data to be added remaining (step S208; Yes), the process proceeds to step S203. Then, when the processing of all the image data is completed (step S208; No), identification information is given to each solar panel of the base image data and output as composite image data (step S209), and the processing ends. .

  Next, the use of features will be further described. As described above, infrared image data is cut out from the infrared video data 35b to generate infrared image synthesis data 15b, and visible light image data is cut out from the visible video data 35c to form visible image synthesis data. 15c is generated. Here, the feature used when associating the infrared light image data is not limited to the feature of the infrared light image data itself. If the correspondence between the visible light image data can be specified, the correspondence between the infrared light image data captured at the same time can also be specified. Similarly, if the correspondence between infrared image data can be identified, the correspondence between visible light image data captured at the same time can also be identified.

  FIG. 7 is an explanatory diagram of mutual use of features in infrared light and visible light image data. In FIG. 7A, infrared image data F1a and visible light image data F1b are captured at time t1. In addition, infrared light image data F2a and visible light image data F2b are captured at time t2. At time t3, infrared light image data F3a and visible light image data F3b are captured.

  Between the infrared light image data F1a and the infrared light image data F2a, there is a luminance distribution pattern that can be used as a feature. On the other hand, there is no luminance distribution pattern that can be used as a feature between the infrared light image data F2a and the infrared light image data F3a.

  Further, there is no image of an article that can be used as a feature between the visible light image data F1b and the visible light image data F2b. On the other hand, there is an image of a tree that can be used as a feature between the visible light image data F2b and the visible light image data F3b.

  Therefore, the image data (F1a, F1b) imaged at time t1 and the image data (F2a, F2b) imaged at time t2 are associated with each other by the feature between the infrared light image data (F1a, F2a). Then, the image data (F2a, F2b) imaged at time t2 and the image data (F3a, F3b) imaged at time t3 are associated with each other by the characteristic between the visible light image data (F1b, F2b).

  The infrared image composite data LFa shown in FIG. 7B is composite image data generated from the infrared light image data F1a to F3a through mutual use of such features. Further, the visible image composite data LFb shown in FIG. 7B is composite image data generated from the visible image data F1b to F3b by performing mutual use of such features.

  Next, a specific example of giving identification information will be described. FIG. 8 is an explanatory diagram of a specific example of giving identification information. First, the common panel specifying unit 16c performs image processing on each image data to recognize the solar panel, and performs numbering that assigns a unique identification number in the image data.

  As a result, in the example shown in FIG. 8A, identification numbers “1” to “34” are assigned to the solar panels in the image data F1a, and “1” to “1” are assigned to the solar panels in the image data F2a. An identification number “36” is assigned.

  Further, the common panel specifying unit 16c determines the positional relationship of the solar panels in each image data. In the example shown in FIG. 8B, the positional relationship of the solar panels is shown by forming a link structure having a connection relationship between the solar panels adjacent vertically and horizontally for each image data.

  After that, the common panel specifying unit 16c determines the correspondence between the image data from the features that exist in common between the image data, and the common panel specifying unit 16c determines the common relationship from the correspondence between the image data and the positional relationship of the solar panels in the image data. Identify solar panels.

  In the example shown in FIG. 8A and FIG. 8B, the solar panel assigned the identification number “22” in the image data F1a, and the solar panel given the identification number “19” in the image data F2a Is a common solar panel.

  The identification information adding unit 16d connects the link structure of the image data F1a and the link structure of the image data F2a as shown in FIG. 8C with reference to the common solar panel specified by the common panel specifying unit 16c. To do. After that, renumbering is performed to reassign the identification number to each solar panel.

  In this way, by reassigning the identification number, it is possible to give unique identification information over a plurality of image data. In the example shown in FIG. 8D, identification numbers “1” to “54” are assigned.

  In the description so far, the distribution pattern of the brightness of the solar panel and the case of using an article around the solar panel as a feature have been described. However, a sign member that can be used as a feature may be installed in advance. .

  FIG. 9 is an explanatory diagram regarding the installation of the marker member. The sign members In1 to In3 shown in FIG. 9 are fixed to the frame of the solar panel 40 at appropriate intervals, for example, so that when the flying object 30 images the solar panel 40, any sign member is always reflected in the video data. To.

  In addition, it is preferable that the marking members In1 to In3 are set in color and shape so that they can be identified with visible light, and set in a heat generation state so as to form a predetermined luminance distribution pattern with infrared light. Further, the plurality of labeling members In1 to In3 individually have different colors, shapes, and heat generation states.

  As described above, when the labeling members In1 to In3 are appropriately arranged, it is possible to reliably extract the features from the video data and to join the image data with high accuracy.

  As described above, the inspection target specifying system according to the present embodiment extracts a plurality of image data from mega solar image data captured by the flying object 30, and extracts features that are commonly present in adjacent image data. By connecting the image data based on the extracted features, the positional relationship of the plurality of solar panels 40 included in the mega solar can be determined, and identification information for uniquely identifying each solar panel 40 can be given. For this reason, when detecting the presence or absence of abnormality on each solar panel forming the mega solar, the position where each solar panel is located can be efficiently identified.

  Note that each configuration illustrated in the present embodiment is functionally schematic and does not necessarily need to be physically configured as illustrated. That is, the form of distribution / integration of each device is not limited to the one shown in the figure, and all or a part thereof may be functionally or physically distributed / integrated in an arbitrary unit according to various loads or usage conditions. Can be configured.

  In the aerial image processing system and the aerial image processing method according to the present invention, when detecting the presence or absence of an abnormality on each solar panel forming a mega solar by aerial photography by a flying object, the position where each solar panel is located is determined. Suitable for efficient identification.

DESCRIPTION OF SYMBOLS 10 Aerial image processing apparatus 11 Display part 12 Input part 13, 33 Wireless communication part 15, 35 Storage part 15a Panel installation data 15b Infrared image synthetic | combination data 15c Visible image synthetic | combination data 16, 36 Control part 16a Cutout part 16b Feature extraction Unit 16c common panel specifying unit 16d identification information adding unit 16e abnormality detection unit 30 flying body 31 imaging unit 32 GPS unit 34 rotor blade 35a flight history data 35b infrared video data 35c visible video data 36a flight control unit 36b imaging control unit 40 solar panel

Claims (7)

A flying object equipped with an imaging device for aerial imaging of video data including a plurality of solar panels forming part of the mega solar while flying over the mega solar, and the aerial imaged by the imaging device of the flying object An aerial image processing system including an aerial image processing device that performs image processing on aerial image data included in video data,
The aerial image processing apparatus includes:
Cutting means for cutting out a plurality of aerial image data from the video data so that a common solar panel is included between adjacent aerial image data;
Feature point extracting means for extracting feature points existing in common in the aerial image data cut out by the cutting means;
Based on the feature points extracted by the feature point extraction means, a specification means for specifying a common solar panel located in each adjacent aerial image data cut out by the cut-out means,
Based on the common solar panel specified by the specifying means, identification information for uniquely identifying the solar panel is assigned to each of the solar panels included in the plurality of aerial image data cut by the cutting means An aerial image processing system, comprising: an identification information providing unit. The imaging device captures infrared image data and visible image data, respectively.
The cutting means cuts out a plurality of infrared light aerial image data from the infrared light video data, and cuts out a plurality of visible light aerial image data from the visible light video data,
The feature point extracting means extracts feature points from at least one of the infrared image data of the infrared light and the aerial image data of the visible light,
The specifying means specifies a common solar panel between the aerial image data of the infrared light or between the aerial image data of the visible light,
The identification information adding unit uniquely identifies each solar panel included in the aerial image data of the plurality of infrared light based on the common solar panel specified by the specifying unit. The identification information is uniquely given to each of each solar panel included in the aerial image data of the plurality of visible lights, and identification information for uniquely identifying the solar panel is given. Aerial image processing system.   The aerial image processing system according to claim 2, wherein the feature point extracting unit extracts the feature points from a luminance distribution in the aerial image data of the infrared light. It further comprises a marker member installed in the vicinity of the solar panel,
The aerial image processing system according to claim 1, wherein the feature point extraction unit extracts the feature points using an image of a marker member included in the aerial image data.   The cutout means calculates an interval for cutting out the aerial image data from the video data based on a flight speed of the flying object, a flight altitude of the flying object, and an angle of view of the imaging device. The aerial image processing system according to claim 1. The specifying means determines the positional relationship of a plurality of solar panels included in one aerial image data,
The identification information providing means determines the positional relationship of each solar panel included in the plurality of aerial image data based on the common solar panel and the positional relationship between the plurality of solar panels included in the aerial image data. It discriminate | determines and provides the said identification information to each solar panel. The aerial image processing system as described in any one of Claims 1-5 characterized by the above-mentioned. A flying object equipped with an imaging device for aerial imaging of video data including a plurality of solar panels forming a part of the mega solar while flying over the mega solar, and an aerial image taken by the imaging device of the flying object An aerial image processing method by an aerial image processing system including an aerial image processing device that performs image processing on aerial image data included in video data,
Cutting out a plurality of aerial image data from the video data so that a common solar panel is included between adjacent aerial image data;
A feature point extracting step for extracting a feature point that exists in common in the adjacent aerial image data cut out by the cutting step;
Based on the feature points extracted by the feature point extraction step, a specific step of specifying a common solar panel located in each adjacent aerial image data cut out by the cutout step;
Based on the common solar panel specified by the specifying step, identification information for uniquely identifying the solar panel is assigned to each of the solar panels included in the plurality of aerial image data extracted by the cutting step An aerial image processing method comprising: an identification information adding step.

Images