Railroad wayside facility inspection system and railroad wayside facility inspection processing method

Abstract

To provide a railroad wayside facility inspection system which can perform inspection of a railroad wayside facility with efficiency and accuracy.SOLUTION: A railroad wayside facility inspection system comprises an image acquisition unit, an inspection adaptability acquisition unit, an image selection unit, and an anomaly detection unit. The image acquisition unit is provided in a railroad vehicle, and acquires captured images from a plurality of imaging devices which can capture a railroad wayside facility which is an inspection object included in wayside on which the railroad vehicle travels from multiple directions. The inspection adaptability acquisition unit acquires an inspection adaptability parameter for determining whether or not to consider the captured image as an analysis object image for inspection of the railroad wayside facility, based on at least a content of the captured image. The image selection unit selects the analysis object image from the plurality of captured images, based on the inspection adaptability parameter. The anomaly detection unit detects presence / absence of abnormality of the railroad wayside facility by using the selected analysis object image.SELECTED DRAWING: Figure 3

Description

[Technical field]

[0001] An embodiment of the present invention relates to a railway wayside facility inspection system and a railway wayside facility inspection processing method. [Background technology]

[0002] Conventionally, in order to operate railway vehicles safely and smoothly, it is necessary for the railway line facilities installed along the railway line to function normally. Examples of railway line facilities include electrical equipment (e.g., electric wires, insulators, curved pull fittings, movable brackets, hanger ears, etc.), communication equipment, signal equipment, and civil engineering equipment (e.g., tracks, tunnels, bridges, etc.). Inspection of these railway line facilities has been performed by analyzing images including the railway line facilities taken while the train is traveling using an imaging device mounted on the train. [Prior art documents] [Patent documents]

[0003] [Patent Document 1] Patent Publication No. 2022-71407 [Patent Document 2] Patent Publication No. 2021-181893 Summary of the Invention [Problem to be solved by the invention]

[0004] In the conventional inspection method, a huge amount of images are captured and then image-processed to perform the inspection, which results in a high processing load. In addition, images captured while the train is in motion are not necessarily suitable for detecting anomalies, and there is a problem that the accuracy of anomaly detection varies and erroneous judgments are likely to occur due to the influence of noise, etc.

[0005] Therefore, if a railway wayside equipment inspection system could be provided that could inspect railway wayside equipment (detect anomalies) efficiently and with high accuracy, it would be useful because it would reduce the burden of inspection work and enable higher accuracy and frequency. [Means for solving the problem]

[0006] A railway wayside equipment inspection system according to an embodiment includes an image acquisition unit, an inspection adaptability acquisition unit, an image selection unit, and an anomaly detection unit. The image acquisition unit is provided on a railway vehicle and acquires captured images from a plurality of imaging devices capable of capturing images of railway wayside equipment, which is an inspection target included in a railway line along which the railway vehicle travels, from multiple directions. The inspection adaptability acquisition unit acquires an inspection adaptability parameter for determining whether or not the captured image is to be an analysis target image for inspection of the railway wayside equipment, based on at least the content of the captured image. The image selection unit selects the analysis target image from the plurality of captured images, based on the inspection adaptability parameter. The anomaly detection unit detects the presence or absence of an abnormality in the railway wayside equipment, using the selected analysis target image. [Brief description of the drawings]

[0007] [Figure 1] FIG. 1 is an exemplary schematic perspective view showing an arrangement of a railway vehicle equipped with a railway wayside facility inspection system according to an embodiment and railway wayside facilities. [Diagram 2] FIG. 2 is an exemplary schematic explanatory diagram showing an imaging direction of an imaging device that captures an image in the railway wayside facility inspection system shown in FIG. [Diagram 3] FIG. 3 is an exemplary schematic block diagram showing a system configuration capable of executing real-time processing in a railway wayside facility inspection system according to an embodiment. [Figure 4] FIG. 4 is an exemplary schematic explanatory diagram showing an inspection management table (inspection management information) for each image acquired from each imaging device in the railway wayside facility inspection system of the embodiment. [Diagram 5]FIG. 5 is an exemplary schematic explanatory diagram showing record contents of an inspection management table (inspection management information) of the railway wayside facility inspection system according to the embodiment. [Figure 6] FIG. 6 is an exemplary schematic block diagram showing a system configuration capable of executing offline processing in a railway wayside facility inspection system according to an embodiment. [Figure 7] FIG. 7 is an exemplary schematic explanatory diagram showing an integrated inspection management table (inspection management information) obtained by offline processing of images acquired from each imaging device in the railway wayside facility inspection system of the embodiment. [Figure 8] FIG. 8 is an exemplary schematic explanatory diagram showing a user display screen (operation screen) of the railway wayside facility inspection system according to the embodiment. [Figure 9] FIG. 9 is an exemplary schematic explanatory diagram showing a user display screen (result screen) of the railway wayside facility inspection system according to the embodiment. DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS

[0008] Hereinafter, an embodiment of the present invention will be described with reference to the drawings. The configuration of the embodiment described below, and the actions and results (effects) brought about by the configuration are merely examples, and are not limited to the contents described below.

[0009] 1 is an exemplary schematic perspective view showing an arrangement of a railway vehicle 12 equipped with a railway wayside equipment inspection system 10 according to an embodiment, and railway wayside equipment 14. The railway wayside equipment inspection system 10 has a function of selecting an image suitable for inspection from a plurality of captured images taken while the railway vehicle 12 is traveling, and detecting the presence or absence of an abnormality in the railway wayside equipment 14 included in the selected captured image.

[0010] Railway line facilities include various facilities such as electrical facilities (e.g., electric wires, insulators, curved pull fittings, movable brackets, hanger ears, etc.), communication facilities, signal facilities, and civil engineering facilities (e.g., tracks, tunnels, bridges, etc.). In Fig. 1, "insulators" interposed between utility poles and electric wires are shown as an example of railway line facilities 14A and 14B as electrical facilities.

[0011] A plurality of imaging devices 16 are installed on the exterior wall of the body of the railway vehicle 12, for example, on the top surface (roof), to capture images used for inspecting railway wayside facilities 14. In the case of Fig. 1, for the sake of simplicity, the railway vehicle 12 is shown as a three-car formation (leading car 12a, middle car 12b, and last car 12c), and an example is shown in which, for example, four imaging devices 16 (16a, 16b, 16c, 16d) are arranged.

[0012] Each imaging device 16 is fixed to the roof of the railcar 12 so that the imaging directions are different so that the railroad wayside facilities 14 can be imaged from multiple directions. The imaging devices 16 are, for example, digital cameras with built-in imaging elements such as CCD (charge coupled device) and CIS (CMOS image sensor). Types of imaging elements include, for example, area scan sensors and line scan sensors. Area scan sensors are sensors used in normal digital cameras, and can capture images at the time of shutter release. Line scan sensors capture linear (line-shaped) images continuously, and can capture continuous, long objects without interruption by moving the camera or the object. For example, line scan sensors are suitable for overhead lines of the railroad wayside facilities 14, and can generate continuous images by dividing them into fixed time intervals. Two types of cameras with imaging elements may be selected or mixed according to the type of railroad wayside facilities 14. In addition, high dynamic range (HDR) cameras may also be used for the imaging devices 16. A high dynamic range camera takes multiple images in succession in a short period of time with different shooting conditions such as shutter time and sensitivity, and then combines these images to generate an image with little blown out highlights or crushed shadows, even in scenes with a large difference in brightness.In addition, instead of using a high dynamic range camera, the same effect can be obtained by using multiple cameras with the same orientation, but with different shooting conditions such as shutter time, sensitivity, aperture, and focus.

[0013] The imaging device 16 can output captured image data at a predetermined frame rate. Each imaging device 16 may have a wide-angle lens or a fisheye lens, and in the case of FIG. 1, for example, the imaging device 16 is oriented diagonally upward so as to include in its imaging range the railway line facilities 14A, 14B (insulators) arranged along the electric railroad tracks. The railway line facilities 14A, 14B (the "insulators" that support the electric railroad tracks) are arranged along the tracks. In addition, since the direction of the railroad vehicle 12 changes along the tracks, the direction (imaging direction) of the imaging device 16 fixed to the roof also changes, and the railway line facilities 14A, 14B arranged at each position can be continuously imaged.

[0014] For example, imaging device 16a is installed on the roof of the leading car 12a such that imaging direction d1 faces obliquely upward with respect to the traveling direction F (forward) of the railway vehicle 12. Furthermore, imaging devices 16b and 16c are installed on the roof of intermediate car 12b as shown in Fig. 2 such that imaging directions d2 and d3 face obliquely upward with respect to the center in the left-right direction of the railway vehicle 12. Imaging device 16d is installed on the roof of the rearmost car 12c such that imaging direction d4 faces obliquely upward with respect to the opposite direction (rear) of the traveling direction F of the railway vehicle 12. Therefore, when imaging devices 16 image the railway wayside facility 14A while the railway vehicle 12 travels in the traveling direction F, it is possible to image the front, rear, left, right, and bottom surfaces of the railway wayside facility 14A.

[0015] As described above, each imaging device 16 is equipped with a wide-angle lens, so that other types of railway wayside facilities 14 existing around the tracks can also be captured in the same frame. Therefore, the railway wayside facilities 14 to be inspected can be selected from the captured image to detect the presence or absence of abnormalities. In another embodiment, the imaging direction of each imaging device 16 may be variable according to the position of the railway wayside facilities 14 so that the railway wayside facilities 14 to be inspected is located, for example, near the center of the imaging area.

[0016] The images captured by each imaging device 16 are sequentially supplied to a processing unit of the railway wayside equipment inspection system 10 via an image transmission path 18 (signal line). As will be described later, the railway wayside equipment inspection system 10 supports both real-time processing and offline processing. In the case of real-time processing, the railway wayside equipment inspection system 10 is mounted on a traveling railway vehicle 12 and sequentially processes the captured images to detect the presence or absence of abnormalities. On the other hand, in the case of offline processing, the railway vehicle 12 accumulates the images captured by the imaging devices 16, and after the operation is completed, etc., the accumulated images are provided to the railway wayside equipment inspection system 10A arranged in a processing center or the like separate from the railway vehicle 12, and the accumulated images are processed to detect the presence or absence of abnormalities.

[0017] The railcar 12 includes a train traffic management device 20. The train traffic management device 20 can acquire position information indicating the running position of the railcar 12. The train traffic management device 20 acquires the current running position (e.g., latitude and longitude information) of the railcar 12 that can be measured based on kilometer post information (distance from a point that is the starting point of the line) and GPS (Global Positioning System) information, and provides it to the railroad lineside facility inspection system 10 and the like. The railroad lineside facility inspection system 10 associates the captured images captured by each imaging device 16 with position information corresponding to the captured position, and processes them as integrated information. Note that the position information acquired by the train traffic management device 20 is generally information on the reference position (e.g., the leading position) of the railcar 12. Therefore, the position information is shifted from the actual position of the imaging device 16 arranged in each car of the railcar 12. For example, an error of three cars occurs between the imaging device 16a of the leading car 12a and the imaging device 16d of the last car 12c. Therefore, when linking image information with position information, the position information is corrected according to the position of the imaging device 16.

[0018] Incidentally, when capturing an image of the surroundings with the imaging device 16 fixed to the roof or the like of the railway vehicle 12, the quality of the captured image differs depending on the time of capturing the image, the weather, the scenery around the railway line, and the like. For example, during the day, the way in which the sun (light) hits the railway lineside facility 14 to be inspected changes depending on the relative angle with the sun. For example, in the case of front lighting, the pattern on the surface of the railway lineside facility 14 can be clearly captured. Conversely, in the case of back lighting, the surface becomes shadowed and pitch black, making it difficult to distinguish the pattern. In other words, the difficulty of image analysis of the captured image changes depending on the sunlight. Also, when the imaging device 16 faces the sun directly (back lighting), the screen becomes overexposed and nothing is visible, making it unusable for inspection. Also, when it goes into the shade, it becomes pitch black and nothing can be distinguished. In addition, in the case of back lighting, the outline portion has high contrast, and the image can be suitable for the purpose of distinguishing the external shape.

[0019] Furthermore, as the orientation of the imaging device 16 changes, the background of the railway line facility 14 being inspected also changes. For example, in the case of an imaging device 16 whose imaging direction has been adjusted for inspecting train lines, the imaging device 16 is oriented slightly upward, so the sky is often the background. However, if a bridge or building is included in the imaging range, it may overlap with the railway line facility 14, making it difficult to distinguish the railway line facility 14. In urban areas, for example, a high-rise apartment building may be in the background. Also, in mountainous areas, mountains may be in the background. In other words, if there is a pattern in the background, it may be more difficult to analyze the captured image.

[0020] However, one imaging device 16 captures only one side of the railway line facility 14 to be inspected. Therefore, abnormalities on hidden surfaces such as the side and back may be overlooked. Therefore, in the case of the railway line facility inspection system 10 of this embodiment, as shown in FIG. 1, multiple captured images obtained from multiple imaging devices 16 with different imaging directions are used. As a result, inspection can be performed from multiple sides. That is, in the case of the configuration shown in FIG. 1, for example, by photographing the railway line facility 14A with the imaging devices 16a to 6d, it is possible to capture and inspect almost all surfaces except the top surface. That is, even if the imaging device 16a is backlit, the imaging device 16d is frontlit, and an image suitable for analysis can be obtained. In addition, since the imaging directions of the imaging devices 16b and 16c are opposite, for example, the image on the side without the reflection of the building can be used. That is, by using images captured by multiple imaging devices 16 with different imaging directions, it is possible to improve the possibility of acquiring an image usable for analysis.

[0021] FIG. 3 is an exemplary schematic block diagram showing a system configuration capable of executing real-time processing in the railway wayside facility inspection system 10 according to the embodiment.

[0022] The railway wayside facility inspection system 10 is an image analysis device that can be realized with computer resources similar to those of a general information processing device such as a PC (Personal Computer), etc. The railway wayside facility inspection system 10 is composed of a CPU (Central Processing Unit), a ROM (Read Only Memory), a RAM (Random Access Memory), a storage unit such as an HDD (Hard Disk Drive) or an SSD (Solid State Drive), a communication interface, an input / output interface, etc.

[0023] The CPU implements a module that reads out a program installed and stored in a non-volatile storage device such as a ROM or SSD, captures images of the railway wayside in real time in accordance with the program, and detects in real time whether or not there is an abnormality in the railway wayside equipment 14 contained in the captured images. The railway wayside equipment inspection system 10 implements a real-time inspection management unit 22, a real-time image acquisition unit 24, an inspection adaptability acquisition unit 26, a real-time image selection unit 28, an abnormality detection unit 30, etc.

[0024] The real-time inspection management unit 22 controls the entire inspection process. In addition, for example, the real-time inspection management unit 22 acquires position information D1 from the train traffic control device 20. As described above, the imaging devices 16 installed at various locations from the front to the rear of the railcar 12 are different from the position information D1 indicating the representative position of the railcar 12. Therefore, the real-time inspection management unit 22 acquires position correction information for indicating the exact installation position of each imaging device 16 based on the position information D1 and the configuration of the railcar 12. The position correction information is provided to the real-time image acquisition unit 24, the inspection adaptability acquisition unit 26, the real-time image selection unit 28, and the like.

[0025] Furthermore, when the railcar 12 reaches the inspection start position, the real-time inspection management unit 22 issues an instruction to start image acquisition to the real-time image acquisition unit 24, and starts real-time abnormality detection processing based on the acquired captured images. Furthermore, when the railcar 12 reaches the inspection end position, the real-time inspection management unit 22 issues an instruction to end image acquisition to the real-time image acquisition unit 24, receives the inspection result from the abnormality detection unit 30, and records it in the real-time inspection management table 32. In other words, the inspection result information is created.

[0026] The real-time image acquisition unit 24 sequentially acquires captured images from each imaging device 16 and provides them to the inspection adaptability acquisition unit 26 and the real-time image selection unit 28 .

[0027] The inspection adaptability acquisition unit 26 calculates (acquires) an adaptability parameter that is a threshold for determining whether or not the acquired captured image is an image suitable for anomaly detection. The calculation criteria can be calculated based on, for example, sunlight, background, orientation of the image capture device 16, and a combination thereof. The adaptability parameter is composed of multiple parameters, and is provided to the real-time image selection unit 28.

[0028] The adaptability parameter based on sunlight can be calculated, for example, from the brightness difference between the background and the object by image processing. Also, the solar altitude and azimuth angle can be calculated from the calendar, time, and latitude and longitude, and the relative relationship with the orientation (backlit, frontlit) and angle of view of the imaging device 16. Also, weather information, for example, can be acquired as external input data D2, and adaptability parameters can be calculated based on the formation of shadows, etc. Also, a comprehensive adaptability parameter can be calculated by combining these.

[0029] The adaptability parameter based on the background can be calculated, for example, from the texture by image processing to determine whether the background is suitable for image analysis. Also, since it is possible to obtain location information (map information, etc.) from the location information database DB1 and recognize whether a bridge, mountain, etc. is included in the background, the adaptability parameter can be calculated based on that data. Also, since it is possible to obtain past data (history information) related to the background for each location, the adaptability parameter can be calculated. Moreover, a comprehensive adaptability parameter may be calculated by combining these. The location information database DB1 can store information on the background for each location and the direction of the imaging device 16. Information can be entered manually, or past results can be left as history.

[0030] The adaptability parameter based on the orientation of the imaging device 16 can be calculated, for example, by detecting the orientation of the object to be inspected (railway facility 14) through image processing. Also, the adaptability parameter can be calculated using the orientation (history) of the object to be inspected (railway facility 14) in the past for each location. Also, a comprehensive adaptability parameter can be calculated by combining these.

[0031] The real-time image selection unit 28 selects a captured image to be sent to the anomaly detection unit 30 from among a plurality of captured images sequentially acquired by the real-time image acquisition unit 24 according to the adaptability parameter. For example, the captured image having an adaptability parameter equal to or greater than a threshold value is selected. When a plurality of adaptability parameters are calculated, all captured images having adaptability parameters equal to or greater than the threshold value may be selected, or captured images having all adaptability parameters equal to or greater than the threshold value may be selected. Note that by selecting fewer captured images, the processing load on the anomaly detection unit 30 can be reduced.

[0032] The anomaly detection unit 30 can use an anomaly detection algorithm using artificial intelligence (AI). For example, an object (railway facility 14) can be extracted by an object detection network, and dirt can be determined from an abnormality in the texture of the object's surface, or damage can be determined from an abnormality in its shape. The adaptability parameters can also be used to adjust the detection accuracy of the anomaly detection process. For example, the sunshine adaptability parameters can be used to adjust the contrast of the input image, or the line detection and binarization thresholds. The background adaptability parameters can also be used to adjust the operation of the object detection network.

[0033] The anomaly detection unit 30 may simultaneously recognize multiple types of railway wayside facilities 14 included in the captured image and perform anomaly detection processing, or may perform anomaly detection processing individually for each type of railway wayside facilities 14, for example, an anomaly detection unit 30a for insulators, an anomaly detection unit 30b for fasteners, an anomaly detection unit 30c for contact lines, etc., as shown in Fig. 3. In this case, the processing load of each anomaly detection unit 30 can be reduced, and a dedicated anomaly detection algorithm for each railway wayside facility 14 can be used, improving the detection accuracy.

[0034] In another embodiment, when there are multiple types of railway wayside facilities 14, a dedicated system configuration (inspection adaptability calculation unit, real-time image selection unit, anomaly detection unit) may be configured to match the type of railway wayside facilities 14. The real-time inspection management unit 22 may also be configured to appropriately manage limited hardware resources according to priority. For example, the priority of the railway wayside facilities 14 to be inspected may be determined, and anomaly detection processing may be performed according to the priority order.

[0035] The anomaly detection unit 30 sequentially stores the captured images of the inspection target in the inspection image database DB2. The stored captured images can be used by an inspector to visually check the presence or absence of anomalies in the railway wayside facility 14, to verify the content of the anomaly, and to consider how to respond to repairs. The inspection image database DB2 stores all captured images selected by the real-time image selection unit 28. In another embodiment, all images acquired by the real-time image acquisition unit 24 may be stored.

[0036] The real-time inspection management unit 22 records the captured images used during the inspection selected by the real-time image selection unit 28 and the detection results of the anomaly detection unit 30 in chronological order in the real-time inspection management table database DB3. Fig. 4 is an exemplary and schematic explanatory diagram showing a real-time inspection management table 32 (inspection management information) for each captured image acquired from each imaging device 16 in the railway wayside facility inspection system 10. The real-time inspection management table 32 stores the inspection time and inspection results for each imaging device 16. Fig. 4 shows real-time inspection management tables 32a to 32d corresponding to the imaging devices 16a to 16d. The horizontal axis of the real-time inspection management table 32 indicates points such as kilometer posts.

[0037] In FIG. 4, an area m1 indicated by dots indicates an area that has been "inspected" and has "no abnormality". An area m2 indicated by diagonal hatching indicates an area that has "not been inspected" and has not yet been inspected. An area m3 indicated by black dots indicates an area that has "been inspected" and has "abnormality". The inspection status can be confirmed on a display unit 34 (see FIG. 1) controlled by the real-time inspection management unit 22. The display unit 34 is, for example, a liquid crystal display (LCD) or an organic electroluminescent display (OLED). The display unit 8 is covered with a transparent operation unit 36 ​​such as a touch panel. An inspection person can visually confirm an image displayed on the display screen of the display unit 34 through the operation unit 36. The inspection person can also perform an operation input by touching, pressing, or moving the operation unit 36 ​​with a finger or the like at a position corresponding to the image displayed on the display screen of the display unit 34. The operation unit 36 ​​may be an input device such as a keyboard or a mouse. The real-time inspection management table 32 may display the inspection results for each imaging device 16 in one real-time inspection management table 32. In this case, when the real-time inspection management tables 32a to 32d show different results at the same timing, it is desirable to display the worst inspection result among them to encourage detailed inspection of the target railway lineside facility 14. In other words, when there are two types of inspection results at the same point, that is, inspected (no abnormality) and abnormality detected, the integrated result is abnormality detected from the viewpoint of preventing overlooking of an abnormality. It is also possible to display the expiration of the inspection by referring to the start timestamp of the inspection result. When the system starts operating, the real-time inspection management table 32 has a single uninspected record for the entire inspection section, and is divided and updated in parts each time a valid inspection is performed. If the reliability of the result output by the anomaly detection algorithm together with the inspection result is low, updating may not be performed.

[0038] In the real-time inspection management tables 32a to 32d shown in FIG. 4, different inspection results are shown at the same timing because there are some parts that are not judged as abnormal or normal, but are not judged as favorable images based on the applicability parameters. For example, the applicability may be judged high by the imaging device 16a, but not judged low by the imaging device 16d. In the real-time inspection management table 32, a timestamp indicating expiration may be attached to each wayside facility 14, and the applicability parameters may be loosened to make the facility a special inspection target. If the wayside facility inspection system 10 is installed in a commercial vehicle and the captured images are acquired, it is possible to acquire information on changes in brightness and weather depending on the time of day over a long period of time (for example, from the first train to the last train). As a result, if the same section is run multiple times during business hours, inspection processing can be performed on the same section multiple times a day, which contributes to improving the inspection accuracy.

[0039] FIG. 5 is an exemplary schematic explanatory diagram showing the record contents of the real-time inspection management table 32 (inspection management information) of the railway wayside facility inspection system 10. As shown in FIG.

[0040] The head portion (record header R1) of the real-time inspection management table 32 records the inspection start kilometer post, end kilometer post, start timestamp, number of image inspection results, etc. In addition, an overall judgment result, etc. may be recorded. The overall judgment result is abnormal if an abnormality is found in one or more of the railway wayside facilities 14 in this record. It is normal only if all of the railway wayside facilities 14 are normal. In addition, in order to make it easier to search for the location of the abnormality, a new record may be created when an abnormality is found. For example, when inspection processing (abnormality detection) is performed at a frequency of 30 frames per second, the same location is captured in multiple captured images, so abnormalities are detected continuously. In addition, since inspection processing of the same location is continuously performed while the railway vehicle 12 (inspection vehicle) is stopped, in order to prevent an increase in the size of the real-time inspection management table database DB3, etc., recording of the image inspection results may be stopped until the train departs or until an abnormality is detected.

[0041] Following the record header R1 in the real-time inspection management table 32, image inspection results RE (RE1, RE2, ...) for each captured image taken by the imaging device 16 are recorded. The first part of the image inspection result RE (image header R11) records the image ID, the kilometer post where the image was taken, a timestamp, the image judgment result, the number of parts inspection results, etc. The image judgment result is abnormal if an abnormality is found in one or more railway wayside facilities 14 in the captured image. It is normal only if all railway wayside facilities 14 are normal. Image data RV is recorded after the image header R11. Note that the image data RV may display a link indicating the image storage location instead of the actual image data.

[0042] Following the image data RV, the inspection results PR (PR1, PR2,...) for each part are recorded. The inspection results PR store the ID, type, position (coordinates (X1, Y2)), part size (width W1, height H1), part judgment result, judgment reliability, etc. of the detected railway wayside facility 14.

[0043] In this way, the railway line facility inspection system 10 of the present embodiment, which operates in real time, acquires captured images of the railway line facility 14 from multiple directions, and acquires an adaptability parameter for selecting an image suitable for the inspection process from among the captured images. Then, a detection process is performed to detect the presence or absence of an abnormality in the railway line facility 14 based on the selected captured images. As a result, the inspection (abnormality detection) of the railway line facility 14 can be performed efficiently and with high accuracy using captured images that are less affected by the sunlight and background. In other words, it is possible to solve problems such as the need for workers to perform visual inspection late at night after commercial operation has ended due to the difficulty of inspection (image acquisition) during the day, the shortage of workers, and the increased load due to the increase in inspection points due to the progression of aging. In addition, the railway line facility inspection system 10 may be applied to a commercial vehicle used for commercial operation, or a dedicated vehicle for inspecting the railway line facility 14. By installing the railway wayside facility inspection system 10 of this embodiment in a commercial vehicle, it is possible for the vehicle to travel on the same line multiple times in a day from the first train (early morning) to the last train (late night), enabling inspection processing (anomaly detection processing) of railway wayside facility 14 regardless of day or night. As a result, it is possible to increase the frequency of inspections, which contributes to improving the accuracy of anomaly detection.

[0044] Next, a system that performs inspection processing (anomaly detection processing) for railway wayside facility 14 offline will be described with reference to Fig. 6. Fig. 6 is an exemplary schematic block diagram showing a system configuration capable of performing offline processing in railway wayside facility inspection system 10A.

[0045] The railway wayside equipment inspection system 10A is similar to the railway wayside equipment inspection system 10, which is a real-time system, except that the images captured sequentially by the imaging device 16 are stored and anomaly detection processing for the railway wayside equipment 14 is performed in a management center or the like separate from the railway vehicles 12. Therefore, in Fig. 6, the same reference numerals are used for configurations that are substantially the same as those in Fig. 3, and detailed descriptions thereof will be omitted.

[0046] The railway vehicle 12 is equipped with a plurality of imaging devices 16 (e.g., imaging devices 16a to 16d) capable of capturing images of railway wayside facilities 14, which are inspection targets along the way, from multiple directions while traveling, and sequentially stores the captured images in an image storage unit 38. When storing the captured images, the image storage unit 38 associates them with location information.

[0047] The railway wayside facility inspection system 10A, which is placed in a management center or the like separate from the railway cars 12, is an image analysis device that can be realized with computer resources similar to those of general information processing devices such as a PC (Personal Computer). The railway wayside facility inspection system 10A is composed of a CPU (Central Processing Unit), a ROM (Read Only Memory), a RAM (Random Access Memory), a storage unit such as an HDD (Hard Disk Drive) or an SSD (Solid State Drive), a communication interface, an input / output interface, and the like.

[0048] The CPU, for example, reads out a program installed and stored in a non-volatile storage device such as a ROM or SSD, and implements a module for detecting offline whether or not an abnormality exists in railway wayside equipment 14 included in accumulated images of the railway wayside in accordance with the program. The railway wayside equipment inspection system 10A implements an offline inspection management unit 22A, an offline image acquisition unit 24A, an inspection adaptability acquisition unit 26, an offline image selection unit 28A, an abnormality detection unit 30, etc.

[0049] The offline inspection management unit 22A controls the entire inspection process. In addition, for example, the unit acquires position information D1 from the train traffic control device 20. As described above, the imaging devices 16 installed at various locations from the front to the rear of the railcar 12 are different from the position information D1 indicating the representative position of the railcar 12. Therefore, the offline inspection management unit 22A acquires position correction information for indicating the exact installation position of each imaging device 16 based on the position information D1. The position correction information is provided to the offline image acquisition unit 24A, the inspection adaptability acquisition unit 26, the offline image selection unit 28A, and the like.

[0050] Furthermore, when the railway vehicle 12 reaches a position where the inspection starts, the offline inspection management unit 22A instructs the image storage unit 38 to start storing images. Furthermore, when the railway vehicle 12 reaches a position where the inspection ends, the offline inspection management unit 22A instructs the image storage unit 38 to end storing images. Furthermore, when a request to start inspection is received from the inspector via the operation unit 36, the offline inspection management unit 22A instructs the offline image selection unit 28A to start selection so as to select captured images suitable for inspection from the metadata database DB4 in which the captured images are stored. Then, the inspection results are received from the anomaly detection unit 30 and recorded in the offline inspection management table.

[0051] Prior to performing offline anomaly detection processing, the offline image acquisition unit 24A acquires captured images of a batch or a specified inspection range from the image storage unit 38, stores them in the metadata database DB4, and provides them to the inspection adaptability acquisition unit 26, offline image selection unit 28A, etc. as appropriate.

[0052] The operation of the inspection adaptability acquisition unit 26 and the contents of the location information database DB1 are the same as those in the real-time processing (FIG. 3), and therefore a detailed description will be omitted. Note that, in FIG. 6, the captured images are processed after being stored in the metadata database DB4, but they may be provided to the inspection adaptability acquisition unit 26 before being stored and processed offline.

[0053] The offline image selection unit 28A selects captured images to be sent to the anomaly detection unit 30 according to the adaptability parameters acquired (calculated) by the inspection adaptability acquisition unit 26, as in the online processing. The difference from real-time processing is that the metadata database DB4 stores captured images of the same location taken multiple times by the same imaging device 16. Regarding sunlight, since the altitude and azimuth of the sun change depending on the time, it is possible to select captured images that are more suitable for image analysis than real-time processing. In addition, it is also possible to select multiple captured images of the same location and send them to the anomaly detection unit 30. The selection procedure based on the adaptability parameter threshold is performed in the same way as in real-time processing.

[0054] The anomaly detection unit 30 basically operates in the same way as in real-time processing (FIG. 3). A measure is added to handle input of captured images of the same spot captured multiple times by the same imaging device 16. For example, as a preliminary step to anomaly detection, an inspection target captured in multiple images captured under different lighting conditions can be superimposed to synthesize a high dynamic range (HDR) image and used. In addition, by using images captured at night with little background for the railway line facilities 14, using images captured under backlighting with which the outline is easily distinguished to determine anomalies in the external shape, and using images captured under front lighting with which the texture is clearly visible to determine anomalies in the surface condition, it is possible to perform processing that makes the most of the characteristics of each captured image and improve the detection accuracy.

[0055] 6 shows an example in which the processes by the anomaly detection unit 30 are performed collectively, but if there are many types of railway wayside facilities 14 to be inspected, multiple components (inspection adaptability acquisition unit 26, offline image selection unit 28A, anomaly detection unit 30) may be provided according to the types. Also, the offline inspection management unit 22A is configured so that it can appropriately manage finite hardware resources according to priority.

[0056] The anomaly detection unit 30 sequentially stores the captured images of the inspection target in the inspection image database DB2. The stored captured images can be used by an inspector to visually check the presence or absence of anomalies in the railway wayside facility 14, to verify the content of the anomaly, and to consider how to respond to repairs. The inspection image database DB2 stores all captured images selected by the offline image selection unit 28A. In another embodiment, all images acquired by the offline image acquisition unit 24A may be stored.

[0057] The offline inspection management unit 22A records the captured images used during the inspection selected by the offline image selection unit 28A and the detection results of the anomaly detection unit 30 in the offline inspection management table database DB5 in chronological order. FIG. 7 is an exemplary and schematic explanatory diagram showing an offline inspection management table (inspection management information) in which the results of multiple inspections based on captured images captured by multiple imaging devices 16 are integrated after the inspections are performed. The horizontal axis of the offline inspection management table 40 indicates a location such as a kilometer post. The offline inspection management table 40 may store the results of the inspection for each imaging device 16, as in the real-time inspection management table 32. The offline inspection management table 40 may also store past history. In that case, the optimal one can be selected from several runs. The results may be updated by executing the process again after a new run. Also, the inspection staff may be able to grasp the uninspected parts via the display unit 34 and the operation unit 36. Also, the expiration of the inspection may be displayed by referring to the start timestamp of the inspection result. The previous results may be cleared by inputting information on accidents or construction. Also, the operation plan may be adjusted based on the inspection situation. The contents of the areas m1, m2, and m3 in the offline inspection management table 40 and the recorded contents of the offline inspection management table 40 are the same as those in FIG. 4 and FIG. 5, and therefore a description thereof will be omitted.

[0058] In this way, the railway line facility inspection system 10A, which operates offline in this embodiment, acquires images of the railway line facility 14 captured from multiple directions in advance, and acquires an adaptability parameter for selecting an image suitable for inspection processing from the stored captured images at a timing different from the timing of capturing the image. Then, a detection process is performed to detect the presence or absence of an abnormality in the railway line facility 14 based on the selected captured image. As a result, the inspection (abnormality detection) of the railway line facility 14 can be performed efficiently and with high accuracy using captured images that are less affected by the sunlight and background. In other words, it is possible to solve the problems of the conventional method, such as the need for workers to perform visual inspection late at night after commercial operations have ended due to the difficulty of inspection (image acquisition) during the day, the shortage of workers, and the increased load due to the increase in inspection points due to the progression of aging. In addition, the captured images used in the railway line facility inspection system 10A may be captured by a commercial vehicle used for commercial operations, or may be captured by a dedicated vehicle for inspecting the railway line facility 14. By using the captured images from commercial vehicles in the railway line equipment inspection system 10A, it becomes possible to use captured images taken multiple times within the same line throughout the day, from the first train to the last train, thereby contributing to improving the accuracy of detecting abnormalities.

[0059] Fig. 8 is an exemplary schematic explanatory diagram showing an operation screen 42a of the user display screen 42 displayed on the display unit 34 in the railway wayside facility inspection system 10, 10A. Also, Fig. 9 is an exemplary schematic explanatory diagram showing a result screen 42b of the user display screen 42.

[0060] The operation screen 42a has an inspection target selection switch 44a, an inspection time selection switch 44b, and an inspection result filter switch 44c, and by operating each of these, the contents to be displayed during the inspection process can be selected. The selection candidates are displayed in a search result list area 46, and by selecting one from the list and operating the image confirmation button 44d, the contents are displayed as in the result screen 42b shown in FIG. 9.

[0061] 9, images 48 (48a to 48d: cameras 1 to 4) corresponding to the images captured by each imaging device 16 (16a to 16d), an integrated inspection management table 50, and an inspection target list 52 are displayed.

[0062] The inspection management table 50 corresponds to the real-time inspection management table 32 or the offline inspection management table 40. When a desired position is clicked, the captured image corresponding to the position is displayed in the image 48 (48a to 48d) for each imaging device 16, and can be confirmed. The inspection target list 52 displays a list of the railway lineside facilities 14 that are the objects to be inspected and are shown in the image 48. When a railway lineside facility 14 is selected from the list, the inspection result is displayed on the image. In other words, by selecting a position on the inspection management table 50, the inspection result information of the position and the analysis target image are associated and displayed on the display unit 34, so that the inspector can quickly confirm the inspection result. Note that a plurality of real-time inspection management tables 32 shown in FIG. 4 may be displayed as the inspection management table 50. In this case, images of different positions can be displayed in the image 48, and comparison between positions can be performed.

[0063] In this way, according to the railway wayside facility inspection systems 10, 10A of the present embodiment, the inspection (abnormality detection) of the railway wayside facility 14 can be performed efficiently and with high accuracy.

[0064] In the above-described embodiment, an example of inspecting electrical equipment (e.g., insulators) has been shown as an example of railway wayside equipment 14, but the railway wayside equipment inspection systems 10, 10A can be similarly applied to other railway wayside equipment 14. For example, for communication equipment, signal equipment, civil engineering equipment (e.g., tracks, tunnels, bridges, etc.), it is also possible to execute a detection process for detecting the presence or absence of abnormalities based on captured images selected using adaptability parameters, and similar effects can be obtained.

[0065] Although the embodiment of the present invention has been described above, the above embodiment is merely an example and is not intended to limit the scope of the invention. The above embodiment can be implemented in various forms, and various omissions, substitutions, and modifications can be made without departing from the gist of the invention. The above embodiment is included in the scope and gist of the invention, and is included in the scope of the invention and its equivalents described in the claims. [Explanation of symbols]

[0066] 10,10A Railway lineside equipment inspection system 12. Railroad cars 14, 14A, 14B Railway facilities 16 Imaging device 20 Train operation control device 22 Real-time Inspection Management Department 22A Offline Inspection Management Department 24 Real-time image acquisition unit 24A Offline image acquisition unit 26 Inspection Adaptability Acquisition Unit 28 Real-time image selection section 28A Offline Image Selection 30 Anomaly detection unit 32 Real-time inspection management table 34 Display section 36 Control section 38 Image storage unit 40 Offline inspection management table 42 User display screen

Claims

1. an image acquisition unit that is provided in the railway vehicle and acquires images from a plurality of imaging devices that are capable of capturing images of railway wayside facilities that are inspection targets included in a railway line along which the railway vehicle travels from multiple directions; an inspection adaptability acquisition unit that acquires an inspection adaptability parameter for determining whether or not the captured image is to be an analysis target image for inspection of the railway wayside facility, based on at least a content of the captured image; an image selection unit that selects the analysis target image from the plurality of captured images based on the inspection suitability parameter; an anomaly detection unit that detects the presence or absence of an anomaly in the railway wayside facility by using the selected analysis target image; A railway line equipment inspection system equipped with

2. 2. The railway line equipment inspection system according to claim 1, wherein the inspection adaptability acquisition unit acquires the inspection adaptability parameter based on at least one of the following in addition to the content of the captured image: imaging condition information of the captured image, imaging position information of the captured image, and installation information of the railway line equipment.

3. 2. The railway wayside facility inspection system according to claim 1, wherein the anomaly detection unit sequentially selects the analysis target images while the railway vehicle is traveling, and detects the presence or absence of an anomaly in the railway wayside facility.

4. 2. The railway wayside equipment inspection system according to claim 1, wherein the anomaly detection unit selects the image to be analyzed from the captured images accumulated during the traveling after the railway vehicle has completed traveling, and detects the presence or absence of an abnormality in the railway wayside equipment.

5. The railway wayside facility inspection system according to claim 1 , wherein the anomaly detection unit outputs inspection result information indicating at least the presence or absence of an anomaly in the railway wayside facility in chronological order.

6. 6. The railway wayside facility inspection system according to claim 5, wherein the anomaly detection unit associates the inspection result information with the analysis target image, and displays, on a display device, the analysis target image that corresponds to the selected inspection result information.

7. an image acquisition process for acquiring captured images from a plurality of imaging devices provided on the railway vehicle and capable of capturing images of railway wayside facilities that are inspection targets included in a railway line along which the railway vehicle travels from multiple directions; an inspection adaptability acquisition process for acquiring an inspection adaptability parameter for determining whether or not the captured image is to be an analysis target image for inspection of the railway wayside facility, based on at least a content of the captured image; an image selection process for selecting the analysis target image from the plurality of captured images based on the inspection suitability parameter; an anomaly detection process for detecting the presence or absence of an anomaly in the railway wayside facility using the selected analysis target image; A railway line facility inspection processing method, comprising:

Images