TWI597471B - Use of image processing hanger detection device - Google Patents

Description

Hanger detection device using image processing

The present invention relates to a hanger detecting device using image processing for detecting a hanger by image processing and detecting an abnormal position of a traveling position of a trolley vehicle or an angle of a hanger.

First, the necessity of detecting the hanger will be described.

The railway operator records the tram line while driving and determines the driving position of the tram from the video image. Previously, as a technique for determining the traveling position of a train, there is a GPS (Global Positioning System) or ATC (Automatic Train Control) technology, which has a certain degree of accuracy without a running obstacle of the electric train, but When determining the driving position of the train from the image, it is required to further correctly grasp the driving position.

Further, since the normal hanger is vertically erected, if the inclination is inclined with respect to the vertical direction due to temperature changes or the like, the current collecting characteristics of the electric train are deteriorated, and it is important to grasp the angle of the hanger. Conventionally, the following Patent Documents 1 and 2 are well known as techniques for detecting a hanger. Patent Document 1 is a patent for detecting a hanger using a template matching process or a vertical edge detection process from a captured image. Further, Patent Document 2 is a patent for detecting a hanger from a captured image by performing a Hough transform process or a luminance histogram process, or detecting an intersection of two straight lines.

[Previous Technical Literature] [Patent Literature]

[Patent Document 1] Japanese Patent No. 5402096

[Patent Document 2] Japanese Patent No. 5402096

[Patent Document 3] Japanese Laid-Open Patent Publication No. 2013-015336

The invention disclosed in the above-mentioned Patent Documents 1 and 2 can perform the hanger detection as long as it is a simple line configuration, but the intersection of the plurality of lines 4 and 5 as shown in FIG. 8 at the intersection line or the overlapping portion. It is difficult to detect the hanger 6.

Patent Document 1 detects a hanger by using vertical edge detection. In general, the edge detection system detects a portion having a large differential value between two pixels as an edge. The locality detection of 2 pixels has a problem of more false detection.

Patent Document 2 is a patent for detecting a hanger by Hough transform. Since the Hough transform finds the hanger using the relative feature of the image and the voting amount of the hanger feature, it is difficult to detect the hanger 6 which is small relative to the entire image as shown in FIG.

As a related patent, there is Patent Document 3 mentioned above. This invention is an invention for measuring the wear of overhead lines, and is a patent for GSTH processing (refer to Figs. 4 and 5) for background removal and automatic detection of the wear of overhead lines. The GSTH process of Patent Document 3 is characterized by the point at which the illumination light on the lower surface of the overhead line is reflected as white. Since the hanger which is the object of the present invention is not reflected as a white portion, it is difficult to detect the hanger using Patent Document 3.

The present invention has been made in view of such circumstances, and an object thereof is to provide a hanger detecting device that can detect a hanger using image processing with higher precision.

In order to solve the above-mentioned problem, the hanger detecting device using the image processing according to the first aspect of the invention is characterized in that the line sensor camera and the image processing unit are provided to detect the hanger in the image, and the line sensor camera is from the vehicle. a ceiling mounted on the roof at equal intervals and supporting the overhead line; and the image processing unit disposed inside the vehicle and analyzing the image obtained by using the line sensor camera; The image processing unit includes: The GSTH processing unit performs GSTH processing on the line sensor direction image input from the line sensor camera in sequence, and sequentially obtains the hang in the image in the photographic line direction and the time direction of the line sensor camera. GSTH imagery; and A hanger detecting unit that detects the hanger from the GSTH image.

Further, in the first aspect of the invention, in the first aspect of the invention, the GSTH processing unit performs the imaging of the line sensor camera as the GSTH processing. After the shrinking process is performed in the line direction, the first GSTH image is obtained, and the difference between the line sensor image and the first GSTH image is obtained to obtain a difference image, and the difference image is subjected to shrink processing in the time direction. Performing an expansion process to obtain a second GSTH image that emphasizes the hanger in the line sensor image, and The hanger detecting unit detects the hanger from the second GSTH image.

According to a third aspect of the invention, in the first aspect of the invention, the GSTH processing unit is configured to perform the GSTH processing, and the line sensor direction of the line sensor camera is the photographic line direction of the line sensor camera. Performing a shrinking process to obtain a gsth image, and performing a shrinking process on the gsth image in the time direction, and then performing an expansion process to obtain the GSTH image that emphasizes the hanger in the line sensor image, and The hanger detecting unit detects the hanger from the GSTH image.

According to a fourth aspect of the invention, the GSTH processing unit performs the GSTH processing on the line sensor image by performing the following brightness inversion processing as described in any one of the first to third aspects of the invention. The brightness inversion processing is directed to the line direction of the line sensor camera, and when the brightness value of the main component in the direction of the line is light, the brightness value is inverted row by row, and the brightness of the main component is When the value is dark, the brightness value is not inverted.

According to the present invention, it is possible to extract only the hanger which is present in a specific direction in the image and which is thinner than the specific fineness. Therefore, it is possible to accurately detect a normal hanger in which only the angle is set within the management value.

1‧‧‧ Vehicles

2‧‧‧Line sensor camera

3‧‧‧Image Processing Department

4‧‧‧ overhead lines

5‧‧‧ hanging wire

6‧‧‧ hanger

7‧‧‧structures

11‧‧‧ original image

12‧‧‧Shrinking images

12-m‧‧‧Shrinking buffer

13‧‧‧Open image

13-m‧‧‧Expansion processing buffer

14‧‧‧GSTH imagery

31‧‧‧ arithmetic device

31a‧‧‧GSTH Processing Department

31b‧‧‧swing detection department

31c‧‧‧ memory

32‧‧‧recording device

I‧‧‧ line sensor image

I _A ‧‧‧Brightness reversal image

I _B ‧‧‧First GSTH image

I _C ‧‧‧Differential image, gsth image

I _D ‧‧‧Second GSTH image, GSTH image

S _D ‧‧‧Exploration

S _L ‧‧‧Exploration

S1~S6‧‧‧Steps

S11~S15‧‧‧Steps

Fig. 1 is an explanatory view showing an installation example of a hanger detecting device using image processing according to the first embodiment of the present invention.

Fig. 2 is a block diagram showing the configuration of a device for detecting a hanger using image processing according to the present invention.

Fig. 3 is an explanatory view showing an example of an original image for explaining GSTH processing.

Fig. 4 is an explanatory view showing an example of shrinkage processing by GSTH processing.

Fig. 5 is an explanatory view showing an example of expansion processing of GSTH processing.

Fig. 6 is an explanatory view showing a converted image obtained by performing GSTH processing on the original image shown in Fig. 3.

Fig. 7 is a flow chart showing the flow of the hanger detection in the first embodiment of the present invention.

Fig. 8 is a schematic view showing an example of a line sensor image.

Fig. 9 is a view showing an image in which the brightness of the line sensor shown in Fig. 8 is inverted.

Fig. 10 is a view showing an example in which the image shown in Fig. 9 is subjected to GSTH processing in a region where the image is long in the x direction.

Fig. 11 is a view showing an image in which the difference between the image shown in Fig. 9 and the image shown in Fig. 10 is obtained.

Fig. 12 is a view showing an example in which the image shown in Fig. 11 is subjected to GSTH processing in a region where the image is long in the y direction.

Figure 13 is a flow chart showing the flow of the hanger detection in the second embodiment of the present invention.

Hereinafter, the hanger detection using the image processing of the present invention will be described with reference to the drawings. The device will be described, but the present invention is not limited to the following embodiments.

In the following description, the processing referred to as GSTH processing shows that the expansion processing shown in FIG. 5 is performed after the shrinking processing shown in FIG. 4, and the processing described as the gsth processing is a process of simplifying the GSTH processing, which is a display. The expansion process shown in Fig. 5 is not performed, and only the shrinkage process shown in Fig. 4 is performed.

[Example 1] [Use of GSTH-treated hanger detection device: GSTH (A-GSTH (A))]

The details of the hanger detecting device using image processing according to the first embodiment of the present invention will be described with reference to Figs. 1 to 12 .

As shown in FIG. 1, the hanger detecting device using image processing of the present embodiment includes a line sensor camera 2 provided on the roof of the train car 1, and an image processing unit 3 provided inside the vehicle 1.

Further, the overhead line 4 is supported by a plurality of hangers 6 that are provided at substantially equal intervals on the suspension wire 5.

The line sensor camera 2 is disposed on the side of the roof of the vehicle 1 in such a manner that its optical axis is orthogonal to the traveling direction of the vehicle 1 and the angle and elevation angle of the hanger 6 are accommodated in the field of view. Make settings. Here, one of the dot chain lines in Fig. 1 indicates a photographing line. The image (hereinafter referred to as a line sensor image) I shown in FIG. 8 obtained by the line sensor camera 2 is input to the image processing unit 3.

Further, in the example shown in FIG. 8, the line sensor image I is displayed with the overhead line 4, the hanging line 5, and the four hangers 6.

The image processing unit 3 analyzes the line sensor image I and detects the hanger 6 , and the arithmetic unit 31 and the recording device 32 are configured.

As shown in FIG. 2, the arithmetic unit 31 includes a GSTH processing unit 31a, a hanger detecting unit 31b, and a memory 31c.

The GSTH processing unit 31a inputs parameters based on the self-recording device 32 via the memory 31c. The line sensor image I input from the line sensor camera 2 via the memory 31c is subjected to image brightness inversion processing and GSTH processing with respect to the line sensor image I.

Here, the image brightness inversion processing is processing for inverting the luminance value for each line in the x direction with respect to the line sensor image I. However, when the luminance value of the main component in the x direction is a rich color, it is determined that it is nighttime or tunnel and is not subjected to the luminance inversion processing.

Further, the GSTH processing is a grayscale top hat processing (Gray Scale-Top Hat processing), that is, a top hat handler is applied to a grayscale image as its name suggests. The top hat processing is performed by subtracting the image from the original image (the image is shrunk by the same number of times → the image after the expansion process).

Hereinafter, the GSTH process for the GSTH processing unit 31a will be briefly described using FIGS. 3 to 6. The GSTH process is performed by the following process.

(1) As shown in FIG. 4, for the original image 11 shown in FIG. 3, a range of N pixels (three pixels in FIG. 3) (hereinafter referred to as a search range) S _D -m (m) is performed. =1~M) Offset, one side writes the brightness value of the darkest pixel to the other buffer (shrink processing buffer) 12-m shrink processing, and creates the shrink processing image 12 (12-M ).

(2) Next, as shown in FIG. 5, the range of the N pixels (three pixels in the present embodiment) (hereinafter referred to as the search range) S _L -m is shifted with respect to the contraction processed image 12 The brightness value of the brightest pixel is respectively written into the expansion processing of the other buffer (expansion processing buffer) 13-m, and the open image 13 (13-M) is created.

(3) Then, the open image 13 (13-M) created by (2) is subtracted from the original image.

By performing the above processing, a GSTH image 14 that emphasizes brighter pixels surrounded by darker portions of the image as shown in FIG. 6 can be obtained.

The GSTH image (converted image) I _D shown in FIG. 12 subjected to the GSTH processing after the image brightness inversion processing in the GSTH processing unit 31a is input to the hanger detecting unit 31b via the memory 31c.

The hanger detecting unit 31b inputs parameters based on the self-recording device 32 via the memory 31c. The converted image is input from the GSTH processing unit 31a via the memory 31c, and the hanger 6 is detected by a binarization process or the like.

Here, since the erected hangers 6 are disposed at a specific equidistant distance (distance), the position in the image can be grasped by counting the number of the detected hangers 6.

Further, as shown in Fig. 12, in the present embodiment, only the hanger 6 which is vertically disposed is detected, and the inclined hanger 6 is not detected. The number of hangers 6 between the poles is managed in advance as equipment information. Therefore, by interlocking with the pole detection signal, it can be detected as the hanger 6 of the abnormal angle when the number of the hangers 6 of each pole is small.

The information of the hanger 6 detected by the hanger detecting unit 31b is output to the memory 31c as a result of the hanger detection.

The memory 31c temporarily stores various materials. Various materials are stored in the recording device 32. In addition, the above parameters are used to set the number of pixels N of the search range, the number of pixels w, h or the binarization threshold value necessary for performing GSTH processing.

Hereinafter, the flow of processing of the hanger detecting device using image processing of the present embodiment will be described based on FIG.

As shown in FIG. 7, in the hanger detecting device using image processing of the present embodiment, first, an example of the line sensor image I shown in FIG. 8 is input from the line sensor camera 2 (step S1). In addition, the x direction shown in FIG. 8 corresponds to the photographing line of the line sensor camera 2 of FIG. Moreover, the y direction shown in FIG. 8 is the time when the line sensor camera 2 performs shooting line by line. The image captured by the vehicle 1 at constant speed is shown in Fig. 8.

Next, the GSTH processing unit 31a performs a process of inverting the luminance value with respect to the line sensor image I (step S2). Thereby, the luminance inversion image I _A as shown in FIG. 9 is obtained. Further, as described above, the luminance inversion is processed every line of the x-direction of the line sensor image I, and when the luminance value of the main component in the x direction is dark, it is determined to be nighttime or tunnel and not Since the brightness is reversed, the tunnel portion where the background shown in the upper portion of Fig. 8 is dark and the background of the sky or the like is assimilated.

Next, in the GSTH processing unit 31a, the luminance inversion image I _A shown in FIG. 8 is subjected to GSTH processing in the x direction (step S3). That is, the shrinking process shown in FIG. 4 in which the darkest pixel is obtained in the region of w × 1 [pixel] which is long in the x direction is performed, and then the expansion process shown in FIG. 5 for obtaining the brightest pixel is performed. Thereby, the first GSTH image I _{B obtained by} extracting the white portion shown in FIG. 10 in the x direction by the width w or less is obtained. Further, w is corresponding to the width of the line of the overhead line 4, and is experimentally calculated. In the example shown in FIG. 10, the first GSTH image I _{B is} an image obtained by removing the structure of the hanger 6 and the pole or the building (hereinafter referred to as interference) 7 with respect to the brightness inversion image I _A . .

Next, similarly to the GSTH processing unit 31a, the first GSTH image I _{B is} subtracted from the luminance inverted image I _{A to} obtain the difference image I _C shown in FIG. 11 (step S4). In the example shown in FIG. 11, the difference image I _{C is} an image obtained by extracting interference from the hanger 6 and a pole or a building. Further, when the differential processing may result in a negative region, the carry below 0 is 0.

Next, similarly to the GSTH processing unit 31a, the difference image I _C is subjected to GSTH processing in the y direction (step S5). That is, the shrinking process shown in FIG. 4 for obtaining the darkest pixel is performed in the region of 1×h [pixel] which is longer in the y direction, and then the expansion process shown in FIG. 5 for obtaining the brightest pixel is performed. Thereby, as shown in FIG. 12, an image (hereinafter referred to as a second GSTH image) I _{D obtained by} extracting the white portion in the y direction and having a length of h or less is obtained. The other h corresponds to the diameter of the hanger 6 and is experimentally calculated. In the example shown in FIG. 12, the second GSTH image I _{D is} an image in which only the hanger 6 is extracted with respect to the difference image I _C .

The second GSTH image I _D is input as a converted image to the hanger detecting unit 31b via the memory 31c.

Next, in the hanger detecting unit 31b, the hanger 6 is detected using the binarization processing or the like from the second GSTH image I _D shown in FIG. 12 (step S6). Since the erected hangers 6 are arranged at specific intervals (distances), the number of the detected hangers 6 is counted and the position in the image is grasped.

Perform the above steps P1 to P6 until all the line sensor images I The input is completed.

In the hanger detecting apparatus using image processing of the present embodiment thus constituted, the GSTH processing is combined with the contraction processing shown in FIG. 4 and the expansion processing shown in FIG. 5, and from w×1 [pixel] and 1×. A straight line is obtained in the area of h [pixel], and only a straight line which is specific to the specific direction and is finer than the specific fineness can be extracted from the straight line appearing in the luminance inversion image I _A .

Therefore, the hanger detecting device using image processing in the present embodiment has the following advantages in utilizing the feature of GSTH.

. It is possible to detect only the hanger 6 photographed in a specific direction and a specific thickness.

. It is possible to eliminate the disturbance 7 of the tunnel tunnel opening or the pole 7 and the like having different characteristics from the hanger 6, and only detect the hanger 6.

. In the present embodiment, only the normal hanger 6 disposed in the vertical direction is detected, and the hanger 6 having an abnormal inclination other than the management value is not detected. Therefore, by comparing the number of the hangers 6 to be detected in the present embodiment with the number of the normal hangers 6 of each of the poles previously managed as the equipment data, the abnormal hanger can also be detected. 6.

[Embodiment 2] [A hanger detection device using simplified GSTH processing: GSTH (gsth(A))]

Hereinafter, a hanger detecting device using image processing according to a second embodiment of the present invention will be described with reference to FIGS. 1 to 6 and FIGS. 8 to 13. The hanger detection device using image processing in the present embodiment is different from the first embodiment in that the processing by the GSTH processing unit 31a is simplified. The device configuration is the same as that of the image processing device of the first embodiment, and the description of the image processing unit 3 will be mainly described below.

The flow of the processing of the hanger detecting device for image processing of the present embodiment will be described based on Fig. 13 .

As shown in FIG. 13, in the hanger detecting device using image processing of the present embodiment, first, an example of the line sensor image I is shown in FIG. 8 from the line sensor camera 2 (step S11), and then, In the GSTH processing unit 31a, a process of inverting the luminance value for the line sensor image I and the photographic line whose luminance value of the main component in the x direction is light is inverted (step S12). Thereby, the luminance inversion image I _A shown in FIG. 9 is obtained.

Then, similarly to the GSTH processing unit 31a, the luminance inversion image IA shown in FIG. 8 is subjected to gsth processing in the x direction (step S13). That is, the shrinking process shown in FIG. 4 for obtaining the darkest pixel in the region of w×1 [pixel] which is long in the x direction is performed, and the expansion shown in FIG. 5 for obtaining the brightest pixel is not performed. deal with. Thereby, the first gsth image I _C extracted from the white portion in the x direction by the width w or less as shown in FIG. 11 is obtained. The other w is corresponding to the width of the line of the overhead line 4 and the like, and is experimentally calculated.

Next, similarly to the GSTH processing unit 31a, the gsth image I _C is subjected to GSTH processing in the y direction (step S14). That is, the shrinking process shown in FIG. 4 in which the darkest pixel is obtained in the region of 1×h [pixel] which is long in the y direction is performed, and then the expansion process shown in FIG. 5 for obtaining the brightest pixel is performed. Thereby, the GSTH image I _{D obtained by} extracting the white portion in the y direction and having a length of h or less as shown in FIG. 12 is obtained. Further, the h system corresponds to the diameter of the hanger 6, and is experimentally calculated.

The GSTH image I _D is input as a converted image to the hanger detecting unit 31b via the memory 31c.

Next, in the hanger detecting unit 31b, the hanger 6 is detected from the GSTH image I _D shown in FIG. 12 using a binarization process or the like (step S15). Since the hangers 6 are arranged at a specific interval (distance), the current position of the vehicle 1 can be grasped by counting the detected hangers 6, and since the installation angle of the hangers 6 has a management value, inspection is performed. Whether to set the angle for a specific one.

The above steps P11 to P15 are performed until the input of all the line sensor images I is completed.

In the hanger detecting apparatus using image processing according to the present embodiment configured as described above, the processing is simplified by performing the gsth processing (only performing the contraction processing shown in FIG. 4) in the x direction in the luminance inversion image I _A , and Compared with the first embodiment, there is an advantage that the hanger detection processing can be speeded up. Further, by performing the gsth processing, although the hanger 6 detected on the image is shortened, the present embodiment is intended to determine the presence or absence of the hanger 6, so that the hanger 6 is detected to have a short effect.

[Industrial availability]

The present invention is preferably applied to a hanger detecting device using image processing.

2‧‧‧Line sensor camera

31a‧‧‧GSTH Processing Department

31b‧‧‧swing detection department

31c‧‧‧ memory

Claims (4)

A hanger detecting device using image processing, comprising: a line sensor camera and an image processing unit for detecting a hanger in an image, wherein the line sensor camera is set at an equal interval from a roof of a vehicle and supports a hangar for an overhead line; the image processing unit is disposed inside the vehicle and analyzes the image obtained by using the line sensor camera; and the image processing unit includes a GSTH processing unit for sexy from the line The line sensor sensor image input by the camera is sequentially subjected to GSTH processing in the direction and time direction of the line of the line sensor camera, and obtains the GSTH image of the hanger in the image; and the hanger detecting unit It detects the hanger from the above GSTH image. The hanger detecting device using image processing according to claim 1, wherein the GSTH processing unit performs the expansion processing on the line sensor image in the direction of the photographing line of the line sensor camera as the GSTH processing. And obtaining a first GSTH image, obtaining a difference image between the line sensor image and the first GSTH image, and obtaining a difference image in the time direction, performing a shrinking process on the difference image, and performing an expansion process to obtain the line sexy a second GSTH image of the hanger in the image of the detector, and the hanger detecting unit detects the hanger from the second GSTH image. The hanger detecting device using image processing according to claim 1, wherein the GSTH processing unit performs the GSTH processing, and the line sensor image is subjected to contraction processing only in the line direction of the line sensor camera to obtain gsth And for the above-mentioned gsth image, the shrinking process is performed in the time direction, and then the expansion process is performed, and the GSTH image of the hanger in the image of the line sensor is obtained. For example, the hanger detecting unit detects the hanger from the GSTH image. The hanger detecting device using image processing according to any one of claims 1 to 3, wherein the GSTH processing unit performs the GSTH processing on the line sensor image by performing the following brightness inversion processing; The rotation processing is directed to the photographic line direction of the above-mentioned line sensor camera. When the luminance value of the main component in the direction of the photographic line is light, the luminance value is inverted line by line, and when the luminance value of the main component is dark In the case of the case, the brightness value is not inverted.