JP2006250775A - Detecting device of obstruction at pantagraph peripheries - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To detect obstruction at the pantagraph periphery safely at high speed by means of contactless image processing. <P>SOLUTION: The detecting device of obstruction at the pantagraph periphery shooting pantagraph periphery with two cameras 1 and 2 of right-and-left comprises an image setting means 10, a deleting means 20 of out of examination subject extent, a deleting means 30 of the pantagraph portion, a deleting means 40 of the trolley portion, a masked image forming means 50 for forming masked images with background of black and residuals of white, a mask-processing image forming means 60 for forming mask-processed imageries of right-and-left deleting the imagery data of black portions of the masked imagery, a block retrieving means 70 for retrieving characteristic blocks for either the left imagery or the right imagery, a stereo measurement means 80 for deriving the averaged three-dimensional position data of an envelope as a three-dimensional position of characteristic block, a proximity test means 90 for pantagraph selecting the high degree of proximity of the characteristic block as the obstruction through comparison of the three-dimensional position of characteristic blocks with the three-dimensional position of characteristic pantagraph, and an obstacle detecting data setting means 100 for setting the three-dimensional data of obstacle. <P>COPYRIGHT: (C)2006,JPO&NCIPI

Description

  The present invention relates to monitoring of a pantograph by image processing, and more particularly to an apparatus for detecting obstacles around the pantograph.

A pantograph obstacle refers to an obstacle that comes into contact with the pantograph of a traveling railway vehicle and interferes with traveling.
Conventionally, there are dedicated measurement vehicles called inspection vehicles and vehicle limit measurement vehicles as obstacle measurement means, which are operated at regular intervals (hereinafter referred to as “measurement vehicles”). . This measurement vehicle has a plurality of measurement items such as vehicle body tilt and rail displacement, and the obstacle is one of the measurement items.
This obstacle measurement method includes a contact sensor method, a laser sensor method, and a light cutting sensor method, and has the following characteristics.
1. Contact sensor system A stick-shaped elastic body is attached to the contact sensor, which is attached to a pantograph or a vehicle to detect contact with an obstacle.
2. Laser sensor method Depending on the laser irradiation shape, there are spot type laser, scan type laser, lens type laser (irradiation in fan shape), etc., and the absolute difference to the measurement object due to the phase difference of reflected wave and deformation of the irradiated laser shape This is a method for measuring distance.
3. Light cutting method This is a method of measuring the three-dimensional shape of the measurement target surface by projecting the stripe onto the measurement target, receiving the deformation of the stripe along the unevenness of the measurement target.
Japanese Patent Application No. 2002-151840 “Trolley Wire Position Measuring Device” "Modernization technology for maintenance of train track facilities" by The Institute of Electrical Engineers of Japan 1995 "Latest Electric Railway Engineering" by The Institute of Electrical Engineers, published by Corona Corporation 2000

Obstacle measurement methods include a contact sensor method, a laser sensor method, and a light cutting sensor method, and have the following problems.
(1) In the case of the contact sensor method, it is dangerous if the support and the contact sensor collide with each other at a high speed.
(2) The laser sensor system includes a spot type laser, a scan type laser, and a lens type laser. Since the spot type laser and the scan type laser are sensors that measure only one distance to a measurement point, they can travel at high speed. The vicinity of the pantograph cannot be measured over a wide range from the vehicle. Further, since a lens type laser usually uses a strong laser of CLASS-C or higher (JIS-C-6802 standard), it cannot be used in a place where a person enters.
(3) In the case of the light cutting method, since the light projected from the sensor cannot be seen by sunlight during the day, it is limited to use at night or in a tunnel.

  The pantograph peripheral obstacle detecting device according to claim 1 of the present invention that solves the above-described problem is to continuously capture the pantograph peripheral part with two left and right cameras installed in the vehicle, and perform image processing on the captured image. Then, in the apparatus for detecting obstacles around the pantograph, an image showing an object to be detected as an obstacle from the continuous input images that are the left image and the right image taken by the two cameras is displayed. Image setting means for setting each as a processed image, inspection-out-range-out deleting means for deleting outside the inspection target range from the left and right processed images, and pantograph part deleting means for deleting pantograph portions from the left and right processed images, respectively A trolley part deleting unit that deletes trolley line parts from the left and right processed images; A mask image creating means for creating a mask image in which the background portion of the left and right processed images from which the portion and the trolley portion have been deleted is black and the other is white, and the left and right mask images are superimposed on the input image, respectively. In addition, mask processing image generation means for generating left and right mask processing images from which the image data of the black portion of the mask image has been deleted, and a white portion of the mask processing image for either the left image or the right image. A lump extraction means for extracting a lump feature, and the left and right mask processed images created by the mask processed image creation means are set as stereo measurement images, and a lump feature for the left image or right image extracted by the lump extraction means Stereo measurement means to obtain the average of the 3D position data of the contour line as the 3D position of the mass feature by performing stereo measurement along the contour line The pantograph proximity inspection means for comparing the three-dimensional position of the lump feature obtained by the stereo measurement means with the three-dimensional position of the pantograph feature obtained in advance, and selecting one having a high degree of proximity as an obstacle, It comprises obstacle detection data setting means for setting the three-dimensional data of the obstacle selected by the pantograph proximity inspection means.

  The pantograph peripheral obstacle detecting device according to a second aspect of the present invention for solving the above-mentioned problems is the pantograph peripheral obstacle detecting device according to the first aspect, wherein the lump extracting means is configured to detect a white portion with respect to the mask processed image for both the left image and the right image. Triangulation calculation that extracts the chunk feature and calculates the three-dimensional position of the chunk feature by triangulation method from the center of gravity of the left and right chunk features extracted from the left image and the right image, respectively, instead of the stereo measurement means It has the part.

  The pantograph peripheral obstacle detecting device according to claim 3 of the present invention that solves the above-described problem is that, in claim 1 or 2, the boundary lines of the trolley line deleted portion of the mask image created by the mask image creating means are respectively provided. It is characterized by adding a trolley line deletion unit recovery processing means for recovering the trolley line deletion part by changing the image data of the part sandwiched between the white parts to white if there is an adjacent white part.

  The pantograph peripheral obstacle detecting device according to claim 4 of the present invention that solves the above-described problem is that in claim 1, 2, or 3, the existing structure is excluded from the mass feature selected by the pantograph proximity inspection means. Thus, an object inspection means for performing object object inspection is added.

(1) Effects of the basic concept (Claim 1)
Compared to the contact sensor method, non-contact measurement is possible, so that there is no collision between the obstacle and the sensor, and it is possible to cope with high speed traveling.
Compared to the laser sensor method, there is no mechanical operation part and it can cope with high-speed running, and it is composed of a passive sensor such as a CCD camera.
Since obstacles can be detected at a normal driving speed, obstacles in the dynamic behavior of the vehicle can be detected.
Since an image at the time of occurrence of an obstacle is also recorded at the same time, the state and cause at that time can be estimated by visual observation from the image at the time of occurrence of the obstacle.

(2) Claim 2
In addition to the effect of claim 1, the calculation amount can be reduced because the calculation by the triangulation method using the center of gravity positions of the left and right chunks is used instead of the stereo measurement which takes processing time for setting the three-dimensional position data of the chunk feature. The processing time for obstacle detection processing can be shortened.

(3) Claim 3
In addition to the effect of claim 1 or 2, even if an object reflected in the processed image is divided by deleting the trolley line portion, the division portion is restored and a mass feature is created from the actual object shape. An object detection error can be prevented.

(4) Claim 4
In addition to the effects of claim 1, 2 or 3, when existing structures are not to be detected as obstacles, they are not detected, and only when an object other than the existing structure is close to the pantograph. Can be detected as an obstacle.

  Examples 1 to 5 described below are the best modes for carrying out the present invention.

(1) Basic concept (Example 1)
An object of the present invention is to provide an apparatus for detecting obstacles around a pantograph by image processing.
In this embodiment, as shown in FIG. 1, two CCD cameras 1 and 2 installed in a vehicle are used as image input means.
The pantograph 5 and the trolley line 6 are photographed by the CCD cameras 1 and 2, and images near the pantograph 5 during traveling are continuously acquired while synchronizing with the cameras 1 and 2. The acquired image is output to the image processing unit 3 and stored in the recording unit 4 as a continuous input image.
The image processing unit 3 also detects the panda graph 5 and the trolley line 6 before detecting the obstacle 7 around the panda graph from both the left and right images that are input images.

The pantograph detection method is not particularly limited, and may be processed by a known technique. For example, this is performed by “Image Processing Device” of Japanese Patent Application No. 2003-325411 and “Pantograph Detection Device” of Japanese Patent Application No. 2003-354636.
The method for detecting the trolley wire is not particularly limited, and may be processed by a known technique. For example, this is performed by “Trolley line position measuring device” in Japanese Patent Application No. 2002-151840 and “Image processing device” in Japanese Patent Application No. 2003-325411.
On the premise of the pantograph and trolley line detection data obtained in this way, the image processing unit 3 processes the input images of the left and right CCD cameras 1 and 2 and follows the flowchart shown in FIG. In order to detect the object 7, as shown in FIG. 10, a memory 9, an image setting unit 10, a non-inspection range deletion unit 20, a pantograph part deletion unit 30, a trolley line part deletion unit 40, a mask image creation unit 50, a mask A processed image creation unit 60, a lump extraction unit 70, a stereo measurement unit 80, a pantograph proximity inspection unit 90, and an obstacle detection data setting unit 100 are provided. The image processing unit 3 may be configured by hardware or may be realized by software for each function. In this way, the image processing unit 3 can be provided by a recording medium such as a CD-ROM, Downloading via a network is also possible.

The flowchart of FIG. 9 includes “processing for the left image” and “processing for the right image”.
“Processing for the left image” is as follows: processing image setting, deletion outside the inspection target range, deletion of the pantograph part, deletion of the trolley line part, creation of the mask image, creation of the mask processing image, and mass extraction in order It carries out by implementing (step S1-step S7).
As shown in FIG. 2, “processed image setting” refers to an image in which a target image detected as an obstacle 7 is detected from a continuous image that is an input image stored in the recording unit 4. 9 is a process of setting as a processed image by sequentially taking in the image 9 (step S1), and is executed by the processed image setting unit 10.
As shown in FIG. 3, the “deletion outside the inspection target range” is a process of clearing the image data outside the target range in the processed image, in other words, overwriting the non-target range with the black data (step S2). ), Performed by the out-of-inspection-range deleting unit 20. Specifically, the range overwritten with black data is the upper, lower, left, and right frame portions of the processed image.

As shown in FIG. 4, “deleting a pantograph part” is obtained by performing a pantograph detection process on a pantograph mask (described later) created in advance for an image from which image data not subject to inspection is cleared. This is a process of converting the coordinates to the pantograph position, superimposing the pantograph mask and the processed image, and clearing the image data of the pantograph part (step S3), and is executed by the pantograph part deleting unit 30.
As shown in FIG. 5, “deletion of the trolley line portion” means that a trolley with a set width is obtained based on the trolley line position obtained by the trolley line detection process for an image in which the image data of the pantograph portion is cleared. This is a process of deleting a line part (step S4), and is executed by the trolley line part deleting unit 40.
As shown in FIG. 6, “creation of a mask image” is a binary image (referred to as a “mask image” in which a background portion is black and a portion other than that is white based on a preset threshold value of the image. Is created by the mask image creating unit 50 (step S5).
As shown in FIG. 7, the “creation of a mask processed image” is an image obtained by superimposing a mask image and an input image and clearing image data of a black portion of the mask image from the input image (this is referred to as “mask processed image”). (Step S6), and is performed by the mask processing image creation unit 60.

“Bulk extraction” is a process of extracting a white part of a block from a mask image and detecting a block feature (step S7), as shown in FIG.
The chunk feature has the following data:
・ Outline data ・ Center of gravity position ・ Range surrounding the lump, and position, area, perimeter length, 3D position (no data at this point)

  As for the processing for the right image, as in the left image processing described above, image setting, deletion outside the inspection target range, deletion of the pantograph portion, deletion of the trolley line portion, creation of the mask image, creation of the mask processing image It carries out by implementing in order (step S1-step S6). In the present embodiment, no lump extraction is performed in the “processing for the right image”.

As “processing for left and right images”, based on the results of “processing for left image” and “processing for right image”, setting of 3D position data of mass features, inspection of pantograph proximity, pantograph proximity as follows: This is performed by sequentially performing the degree inspection (steps S8 to S10).
“Set the 3D position data of the lump feature” is to set the left and right masked images as stereo measurement images, perform stereo measurement along the outline of the lump features extracted from the left image, This is a process for obtaining the average of the position data as the three-dimensional position of the mass feature (step S8), and is performed by the stereo measurement unit 80.
“Pantograph proximity degree inspection” means comparing the three-dimensional position of the pantograph obtained from the pantograph detection result obtained in advance with the three-dimensional position of the detected mass feature, and setting the proximity distance as an obstacle in advance. This is a process of selecting a block feature having a height closest to the pantograph as an obstacle (step S9), and is executed by the pantograph proximity inspection unit 90. That is, the object within the set distance and closest to the height (distance) of the pantograph is selected as an obstacle.
“Setting obstacle detection data” is a process of setting the three-dimensional position data as obstacle data if there is a block feature selected as an obstacle in the “pantograph proximity degree inspection” (step S7). ), The obstacle detection data setting unit 100 implements this.

In the pantograph part deletion process (step S3) described above, a pantograph mask is used to delete the pantograph part from the processed image.
The “pantograph mask” is composed of a binary image in which the pantograph part is black and the other part is white.
In the pantograph part deletion process, coordinate conversion is performed on the pantograph mask obtained in the pantograph detection process obtained in advance, and the image data of the black part of the pantograph mask (panda cuff part) is overlapped with the panda graph mask and the processed image. Is deleted from the processed image to delete the pantograph portion from the processed image.
The procedure for creating a pantograph mask is shown in the flowchart of FIG. 16, and the pantograph creation device is shown in FIG.
The pantograph mask creating apparatus includes a basic image setting unit 140, a binary image creating unit 150, an expansion processing unit 160, an isolated part deleting unit 170, an image data reversing unit 180, and a pantograph mask setting unit 190.

As shown in FIG. 16, pantograph mask creation is performed by sequentially executing reference image setting, binary image creation, dilation processing, isolated portion deletion, image data inversion, and pantograph mask setting processing (see FIG. 16). Step S21 to Step S26).
“Setting the reference image” is a process of setting the image data used in the pantograph model of the pantograph detection process as a reference image as shown in FIG. 11 (step S21), and is performed by the basic image setting unit 140. .
As shown in FIG. 12, “creating a binary image” means a binary image in which a background portion is black and a portion other than white is set based on a predetermined image density threshold with respect to a reference image. This is an image creation process (step S22), and is performed by the binary image creation unit 150.
As shown in FIG. 13, the “expansion process” is a process of expanding the white portion of the binary image and expanding the width of the pantograph portion that is currently white (step S23). Implemented by unit 160.
As shown in FIG. 14, “deletion of isolated part” is a process of changing the image data of the black part surrounded by the white part to white with respect to the binary image (step S24). 170.

“Reversal of image data” is a process of performing black and white reversal on a binary image so that the pantograph portion is black and the other portions are white as shown in FIG. 15 (step S25). 180.
“Pantograph mask setting” is a process of setting a binary image that has been binarized and inverted as a pantograph mask (step S26), and is executed by the pantograph mask setting unit 190.
As described above, as an effect of the present embodiment, compared to the contact sensor method, non-contact measurement can be performed, so that there is no collision between the obstacle and the sensor, and it is possible to cope with high speed traveling.
Further, compared with the laser sensor system, there is no mechanical operation part and it can cope with high-speed running, and since it is composed of a passive sensor such as a CCD camera, there is no safety problem even in places where people enter.
Since obstacles can be detected at normal driving speed in this way, obstacles due to the dynamic behavior of the vehicle can be detected, and images at the time of obstacles are recorded at the same time. The state and cause at that time can be estimated by visual inspection.

(Method to extract chunk features from left and right images and calculate chunk 3D position)
This embodiment is an improved example of the first embodiment.
That is, as shown in FIG. 19, a triangulation calculation unit 110 is provided instead of the stereo measurement unit 80, and as shown in FIG. 18, mass extraction is performed also in “processing for the right image” (step S <b> 7). The left lump feature and the right lump feature are respectively extracted from. The other configuration is the same as that of the first embodiment and provides the same function and effect, so the description thereof is omitted.
In the present embodiment, the “three-dimensional position data setting of the lump feature” is performed by the stereo measurement unit 80 according to the first embodiment performing stereo measurement along the outline of the left lump feature to set the three-dimensional position of the lump feature. Instead, it is processed by the triangulation calculation unit 110 calculating and setting the three-dimensional position of the lump feature by the triangulation method from the centroid position of the left and right lump feature (step S8).
As an effect of the present embodiment, in addition to the effect of the first embodiment, the calculation by the triangulation method using the center of gravity positions of the left and right blocks instead of the stereo measurement which takes a processing time for the setting of the three-dimensional position data of the block feature is performed. This reduces the amount of calculation and reduces the processing time for obstacle detection processing.

(Detection of obstacles with recovery of trolley wire deleted part)
This embodiment is an improved example of the first embodiment.
That is, as shown in FIG. 23, a trolley line deletion unit restoration processing unit 120 is added to the image processing unit 3 and, as shown in FIG. 22, after creating a mask image in the left image processing (step S5), a trolley line The deleted portion recovery process (step S11) is added. The other configuration is the same as that of the first embodiment and provides the same function and effect, so the description thereof is omitted.
In the first embodiment, when the trolley line portion is deleted from the mask image for the left image, as shown in FIG. 20, an object that exists across the trolley line is divided on the image and appears in the input image. In some cases, a mass feature having a shape different from that of the actual object is extracted.
Therefore, in this embodiment, in order to suppress the division by the trolley line part deletion process, the restoration process of the trolley line deletion part is performed on the mask image.
This “trolley line deleted portion restoration process” refers to checking the boundary line of the trolley line deleted portion as shown in FIG. 21, and if there is an adjacent white portion, the image data of the black portion sandwiched between the white portions is obtained. This is done by changing to white (step S11).

Specifically, as shown in FIG. 28, an object that exists across the trolley line in the mask image (shown in white) is replaced with a fixed width part (shown in black) by removing the trolley line. Divided into two. The divided portion becomes two black and white boundary lines A and B. When a pixel is searched horizontally by the set width from the upper end P ₁ of the boundary line A and the boundary line B is found, the upper end P _{1 of the} boundary line A is detected. The parallelogram surrounded by the four points of the lower end P ₂ and the upper end P ₄ and the lower end P ₃ of the boundary line B is converted from black to white. Here, the set width is set to be somewhat thicker than the width of the trolley line.
As an effect of the present embodiment, in addition to the effect of the first embodiment, even if an object shown in the processed image is divided by deleting the trolley line portion, the divided portion is restored and a mass feature is created from the actual object shape. Therefore, it is possible to prevent an obstacle detection error due to division.

(Obstacle detection by left and right block with recovery of trolley wire deleted part)
This embodiment is an improved example of the second embodiment.
That is, in the present embodiment, as shown in FIG. 25, a trolley line deletion unit restoration processing unit 120 is added to the image processing unit 3, and as shown in FIG. 24, the mask image is created (step S5). The recovery process (step S11) of the trolley line deletion part similar to Example 3 is added. The other configuration is the same as that of the second embodiment and provides the same function and effect, so the description thereof is omitted.
As an effect of the present embodiment, in addition to the effect of the second embodiment, even if an object shown in the processed image is divided by deleting the trolley line portion, the divided portion is restored and a mass feature is created from the actual object shape. Therefore, it is possible to prevent an obstacle detection error due to division.

(Detection of obstacles with inspection of target objects)
The present embodiment is an improved example of the fourth embodiment.
That is, as shown in FIG. 27, the target object inspection processing unit 130 is added to the image processing unit 3, and as shown in FIG. 26, after the inspection of the pantograph proximity degree (step S9), the target object inspection (step S12). Is added. The other configuration is the same as that of the fourth embodiment and provides the same function and effect, so the description thereof is omitted.
In the fourth embodiment, when a nearby object is detected from the periphery of the pantograph and used as an obstacle, the “curved metal fitting” that pulls the trolley wire and the existing structure that supports it are also detected together. .
This is effective for detecting and checking everything close to the pantograph.

However, in the present embodiment, the following items are inspected so that existing structures are not detected as obstacles so that they are not detected. All applicable items are excluded as existing structures.
-Contact with trolley wire (mainly measures for bent metal fittings)
The degree of coincidence between the stereo corresponding point and the left and right block center of gravity position. The shape similarity of the left and right block feature. The pattern similarity of the left and right block feature. Specifically, in this embodiment, the target object inspection processing unit 130 processes the target object inspection. (Step S12).
Here, the “target object inspection” specifically refers to a trolley line contact inspection, a stereo matching point and a left / right lump center-of-gravity position check, a right / left lump feature shape similarity check, and a left / right lump feature pattern similarity check. Are performed in order.
The “trolley line contact inspection” is a process for inspecting whether the boundary of the trolley line deleted portion and the outline of the lump feature are in contact with each other.
“Stereo correspondence point and left and right lump center of gravity coincidence inspection” is a stereo measurement using the center of gravity of the left lump feature as the measurement point, and the distance between the corresponding point position on the right image and the center of gravity of the right lump feature is This is a process for judging that they match if the distance is smaller than a preset distance.

“Shape similarity inspection of left and right block features” is a method of setting a mask image (binary image) as an inspection target image and converting the image of the left block feature portion into a right image, The absolute value difference is calculated, and when the difference value is smaller than a preset value, the right and left shapes are judged to be similar.
“Pattern similarity inspection of left and right block features” is to set a mask processing image (grayscale image) as the inspection target image, convert the left feature portion image to the right image, and convert the image data and the right chunk feature portion image. The normalization correlation with data is performed, and when the correlation value is larger than a preset value, the left and right pictures are judged to be similar.
As an effect of the present embodiment, in addition to the effect of the apparatus of the fourth embodiment, when it is not desired to detect existing structures as obstacles, they are excluded so as not to be detected. Only when it is close to the pantograph, it can be detected as an obstacle.

  The present invention can be used for monitoring a pantograph by image processing, and in particular, can be used as an apparatus or method for detecting obstacles around the pantograph.

It is a perspective view which shows installation of the CCD camera for pantograph imaging. It is explanatory drawing which shows the setting of a process image. It is explanatory drawing which shows the deletion outside the inspection object range. It is explanatory drawing which shows a pantograph part deletion. It is explanatory drawing which shows trolley line part deletion. It is explanatory drawing which shows mask image creation. It is explanatory drawing which shows mask process image creation. It is explanatory drawing which shows lump extraction. It is a flowchart which shows the procedure of a pantograph peripheral obstacle detection which concerns on Example 1 of this invention. 1 is a configuration diagram of a pantograph peripheral obstacle detection device according to Embodiment 1 of the present invention. FIG. It is explanatory drawing which shows the setting of the reference image in pantograph mask preparation. It is explanatory drawing which shows the setting of the binary image in pantograph mask preparation. It is explanatory drawing which shows the expansion process in pantograph mask preparation. It is explanatory drawing which shows the deletion of the isolated part in pantograph mask creation. It is explanatory drawing which shows inversion of the image data in pantograph mask creation. It is a flowchart which shows the procedure of pantograph creation. It is a block diagram of a pantograph creation apparatus. It is a flowchart which shows the procedure of a pantograph surrounding obstacle detection which concerns on Example 2 of this invention. It is a block diagram of the pantograph surrounding obstacle detection apparatus which concerns on Example 2 of this invention. It is explanatory drawing of the lump characteristic by which lump extraction by the trolley line deletion of the object which exists so that it straddles a trolley line, and was divided | segmented. It is explanatory drawing which shows the trolley line deletion part and its recovery. It is a flowchart which shows the procedure of a pantograph surrounding obstacle detection which concerns on Example 3 of this invention. It is a block diagram of the pantograph periphery obstacle detection apparatus which concerns on Example 3 of this invention. It is a flowchart which shows the procedure of a pantograph periphery obstacle detection which concerns on Example 4 of this invention. It is a block diagram of the pantograph surrounding obstacle detection apparatus which concerns on Example 4 of this invention. It is a flowchart which shows the procedure of a pantograph periphery obstacle detection which concerns on Example 5 of this invention. It is a block diagram of the pantograph surrounding obstacle detection apparatus which concerns on Example 5 of this invention. It is an enlarged explanatory view showing a trolley line deleted portion of an object existing so as to straddle the trolley line and its recovery.

Explanation of symbols

DESCRIPTION OF SYMBOLS 1, 2 CCD camera 3 Image processing part 4 Recording part 5 Pantograph 6 Trolley line 7 Obstacle 9 Memory 10 Image setting part 20 Out-of-inspection range deletion part 30 Pantograph part deletion part 40 Trolle line part deletion part 50 Mask image creation part 60 Mask processing image creation unit 70 lump extraction unit 80 stereo measurement unit 90 pantograph proximity inspection unit 100 obstacle detection data setting unit 110 triangulation calculation unit 120 trolley line deletion unit restoration processing unit 130 target object inspection processing unit 140 basic image setting unit 150 Binary image creation unit 160 Expansion processing unit 170 Isolated part deletion unit 180 Image data inversion unit 190 Pantograph mask setting unit

Claims (4)

In the device for detecting the obstacles in the pantograph peripheral part by continuously imaging the pantograph peripheral part with the two left and right cameras installed in the vehicle, and processing the picked-up image, the two cameras are photographed. Image setting means for setting each of the captured images detected as obstacles from the continuous input images as the left image and the right image as processed images, and out of the inspection target range from the left and right processed images The inspection object out-of-range deletion means for deleting each, the pantograph part deletion means for deleting the pantograph parts from the left and right processed images, the trolley part deletion means for deleting the trolley line parts from the left and right processed images, respectively, Out of the inspection target range, the background portion of the left and right processed images from which the pantograph portion and the trolley portion have been deleted are black, Mask image creation means for creating a mask image for white other than this, and the left and right mask processed images obtained by superimposing the left and right mask images on the input image, and deleting the image data of the black portion of the mask image Created by a mask processing image creation means to be created, a chunk extraction means for extracting a chunk feature of a white portion from the mask processing image for either the left image or the right image, and the mask processing image creation means The left and right mask processed images are set as stereo measurement images, and stereo measurement is performed along the contour lines of the mass features with respect to the left image or the right image extracted by the mass extraction means. Stereo measurement means for obtaining the average of the three-dimensional position of the mass feature, the three-dimensional position of the mass feature obtained by the stereo measurement means, and the pan obtained in advance Pantograph proximity inspection means for comparing the three-dimensional positions of the graph features and selecting one having a high degree of proximity as an obstacle, and obstacle detection for setting the three-dimensional data of the obstacle selected by the pantograph proximity inspection means A pantograph peripheral obstacle detection device comprising a data setting means. In Claim 1, the said lump extraction means extracts the lump feature of the white part with respect to the said mask process image about both the left image and the right image, respectively, and it replaces with the said stereo measurement means, and the left image and the right image An apparatus for detecting obstacles around a pantograph, comprising: a triangulation calculation unit that calculates a three-dimensional position of a lump feature from a center of gravity position of left and right lump features extracted from each by triangulation. 3. The boundary line of the trolley line deleted portion of the mask image created by the mask image creating means is respectively examined in claim 1 or 2, and if there is an adjacent white portion, the image data of the portion sandwiched between the white portions is whitened. An apparatus for detecting obstacles around a pantograph, characterized in that a trolley line deletion part recovery processing means for recovering a trolley line deletion part is added. The pantograph periphery according to claim 1, 2 or 3, wherein an object inspection means for inspecting a target object is added by excluding an existing structure from the mass feature selected by the pantograph proximity inspection means Obstacle detection device.

Images