JPH026269A - Device for detecting train number - Google Patents

Abstract

PURPOSE:To clearly read a train number and detect the accurate train number by estimating the optimum throttle value from the image of a train number indicator and its periphery which is photographed by a camera and undergone image processing and adjusting the throttle value for each train by means of an automatic throttle adjusting mechanism. CONSTITUTION:The passage of the front face or rear face of a train traveling on rails 11 is detected by means of a train detecting sensor 30 and a sensor reflector 31. The periphery of a train number indicator 20 is photographed by means of a fixed focus camera 40 using a CCD element when a train exists at the time of detecting the passage of the front face of train whereas, when the train is passing at the time of detecting the rear face of train. Then, the photographing area 21 on the periphery including a train number indicator 20 installed on the front face or rear face of the train 10 is photographed by the camera 40 and the image signal thereof is inputted into an image processing device 60. The image processing device 60 judges whether the image is taken in at the optimum throttle value from the taken image to estimate the optimum throttle value, which is commanded to an automatic throttle adjusting mechanism 50 to operate the automatic throttle adjusting mechanism.

Description

[Detailed Description of the Invention] [Objective of the Invention] (Industrial Application Field) This invention captures images of the train number display installed on the front and rear sides of a train and its surroundings. The present invention relates to a train number detection device that captures video information from a camera with an appropriate aperture value depending on the image state of the train, and performs train number detection.

(Prior Art) Conventionally, a train number detection device that automatically detects a train number by capturing an image of a train number display attached to the front or rear of a train using a camera and processing the image signal has been used. Train number display The aperture value of the camera for imaging is fixed at a constant value according to the amount of light received by the optical sensor installed around the camera, and the train number is displayed using software technology by image processing of the image signal from the camera. I was trying to detect characters.

(Problems to be Solved by the Invention) However, with such conventional train number detection devices, when an image is taken of the area around the train number display attached to the front or rear of the train, Because each company uses a different type of display, and even within the same company, the train number display may have different designs or stains depending on whether the train is new or old, so the characters displayed on all train number displays have the same contrast. Therefore, there was a problem in that it was not possible to reliably read the train numbers of all trains using only image processing technology.

In order to overcome this problem, it may be possible to install multiple cameras with different aperture values in one place for fixed aperture cameras, but if you increase the number of cameras installed in this way, Differences in the resulting imaging screens require an increase in image processing software, an increase in the number of camera input interfaces for image processing devices, and an increase in the frequency of maintenance due to an increase in the number of cameras.
It is expected that there will be various effects such as increased wiring costs, spoiling the appearance of the platform, and increased equipment costs when the storage capacity of the interface board in the image processing device is exceeded. There was also the problem that the number of units could not be easily increased.

In view of these problems, it may be possible to use a camera equipped with an automatic aperture adjustment mechanism, but normally cameras with such an automatic aperture adjustment mechanism are sensitive to the brightness of external light around the camera. Even if the aperture value is adjusted, it is not possible to determine the appropriate aperture value according to the difference in brightness of the characters in the frame of the train number display installed on each train, and to take images with the determined aperture value. There was a problem.

This invention was made in view of these conventional problems, and the characters on the train number display can be read from a screen captured by a camera around the train number display attached to the front or rear of the train. We provide a train number detection device that can extract the frame using image processing software, activate the automatic aperture adjustment mechanism on the front of the camera based on the image inside the character frame, and detect the characters within the character frame with the clearest image. The purpose is to

[Structure of the Invention] (Means for Solving the Problems) The train number detection device of the present invention includes a train detection sensor that detects the passage of the front or rear side of a train running along a track, and a signal of the train detection sensor. A camera is installed on the front or rear of the train to take images of the area around the train number display, and the images taken by this camera are captured and the clearest image is selected from the images within the character frame of the train number display. This device includes an image processing device that predicts the aperture value that will be obtained, and an automatic aperture adjustment mechanism that operates according to the aperture command value output by the image processing device.

(Function) In the train number detection device of the present invention, a train detection sensor detects the passing of a train, and a camera captures an image around a train number display attached to the front or rear of the train.

Then, the image signal from this camera is processed by an image processing device to predict the aperture value that will give the clearest image from the image within the character frame of the train number display, and the automatic aperture adjustment mechanism is activated based on the aperture value obtained. Activate it to image the train number display at the optimal aperture value.

(Example) Hereinafter, embodiments of the present invention will be explained in detail based on the drawings.

FIG. 2 shows the hardware configuration of an embodiment of the present invention, in which a train number display 20 is attached to the front or rear of the train 10, and the train 10 running on the rails 11 has a train number display 20 attached to the front or rear of the train 10. The train detection sensor 30 detects the passing of the rear side.
The sensor reflector 31 detects the passing of the train number display 20 by the fixed focus camera 40 using a COD element when the train is present when the front of the train is detected, and when the train is passing when the rear of the train is detected. Capture images of the surrounding area.

Note that this camera 40 is equipped with an automatic aperture adjustment mechanism 50.

In order to actually capture images after the train 10 has stopped on the platform, a train detection sensor 30 and a sensor reflector 31 are installed several meters before the train 10 stops, and a signal to detect the presence of a train or the passing of a train is sent to the cable 32. An image signal is input to the image processing device 60 through the camera 40 after a certain period of time has elapsed after the input of this signal.

Then, this camera 40 images the imaging area 21 around the train number display 20 attached to the front or rear surface of the row JlalO, and inputs the image signal to the image processing device 60.

The image processing device 60 predicts the optimum aperture value by determining from the captured image whether the image is captured with an appropriate aperture value, and instructs this value to the automatic aperture adjustment mechanism 50 to automatically adjust the aperture value. The aperture adjustment mechanism 50 is activated.

The detailed hardware configuration of this image processing device 60 is described in the first section.
As shown in the figure, an arithmetic control unit 61 is a part that decodes and executes program instructions.

The image processing processor 62 is an image processing program memory 6
By extracting the various subroutines stored in 3 and setting the necessary parameters, we can obtain the density value when black is 0 and white is 255, and all intermediate colors are expressed with values between 0 and 255. This part executes image-related processing such as image conversion, smoothing, image binarization, and block labeling.

Video input module 64 provides video input to camera 40.
Synchronizes the video signal with the A/D converter that converts the video signal from O to 255 using a vertical synchronization signal,
This is the part that captures the video signal from the beginning of the screen.

The image memory 65 is used to store an image captured by the camera 40 as a numerical value of 0 to 255 for each pixel, and is used as a working area where it is temporarily stored when image processing is performed.

The digital human power module 66 is a module that interrupts the arithmetic control unit 61 when there is a contact signal from the outside, and transfers execution to the train detection sensor program starting from the address written in the interrupt destination. The output module 67 is a module that converts digital values from 0 to 56 into O to IV and outputs them to the outside.The program memory 68 is a memory that stores in advance a group of execution programs necessary for this process, and the data The memory 6 is an area for storing an optimum aperture value setting table, data obtained by various calculations, and constants.

Furthermore, the power-on module 70 is a module that causes an interrupt to occur when the power is turned on, and causes the arithmetic control unit 61 to execute the initial setting program starting from the address in the memory at the destination of this interrupt.

Furthermore, the clock module 71 is a one-second clock to prevent execution from starting until an external device operates before executing the next program, and a clock interrupt occurs in the arithmetic control unit 61 every second. Then, the timer program is activated.

The train detection sensor 30 and the sensor reflector 31 are connected to the train 1.
0, the signal transmitted from the transmitter of the train detection sensor 30 across the rails 11 is reflected by the reflector plate 31, and some of the signals return to the train detection sensor 30 again. It is installed in a position where it can be received by the receiver. If the train 10 does not exist between the train detection sensor 30 and the sensor reflection plate 31, a signal is received, and if the train 10 exists between the train detection sensor 30 and the sensor reflection plate 31, a signal is received. No more reception.

Therefore, if you want to detect the train 10 at the head of the train,
When the state changes from no train to train presence, the contact changes from open to closed, and a detection signal is input to the digital input module 66 via the cable 32.

The automatic aperture adjustment t! The treetop 50 includes a moving coil 51, an iron piece 52, magnetic pole pieces 53, 53 . It is composed of a spring 54, a gear 55, and aperture gears 56a and 56b.

Each of the aperture gears 56a and 56b has a slit 57.
a, 57blfi is provided. 58 is a lens.

Therefore, when the voltage signal output from the aperture value module 67 is applied to the movable coil 51 via the variable resistor 59, the movable coil 51 moves around the center against the spring 54 in proportion to the voltage level. 52.

The gear 55 rotates in conjunction with this rotation, causing the aperture gears 56a and 56b to rotate simultaneously. This aperture gear 5
6a and 56b have slits 57a.

57b, the light incident from the lens 58 passes through the overlapping openings of both slits 57a and 57b, enters the CCD image sensor 42, and is captured as a video signal.

The amount of light incident on the CCD image sensor 42 is adjusted by the slits 57a of the aperture gears 56a, 56b,
This is determined by the size of the overlapping openings of the aperture gears 57b and 57b.

The rotation angle of the aperture gears 56a and 56b is controlled by adjusting the rotation angle of the aperture gears 56b, and the voltage applied to the movable coil 51 is adjusted by adjusting the resistance values of the five variable resistors, and the rotation angle changes. This will be done by

The operation of the train number detection device having the above configuration will be described next.

Based on the flowchart shown in FIG. 3, when the power is turned on, the digital value k previously set at the address Fo is written to the address F1, and the first aperture adjustment is executed using this aperture value k (step 210).

Next, when the signal from the train detection sensor 30 changes from absent to present, when that signal is input to the digital input module 66, the arithmetic control unit 66 informs that an interrupt has occurred.
1 is detected, the interrupt destination is analyzed, and the train detection sensor signal input processing program, which is a program starting from the first address stored in the interrupt destination, is retrieved from the program memory 68 and executed from the first address (step 220). .

Since this train detection sensor input processing program captures train 10 that is about to stop, train 1
In order to suspend the execution of the program for a period of time until 0 completely stops, the content t1 of the address TI stored in advance in the data memory 69 is retrieved and multiplied by a time delay of this t1 time.

This timer 10g terminates the train detection sensor signal input processing program when the arithmetic and control unit 6e counts up the clock every second of the time resistance module 71 and reaches t1 seconds, and starts the camera aperture value command output processing program. Transfer program execution to the address.

The arithmetic control unit 61 takes out the program starting from the first address of this aperture value command output processing program from the pre-stored program memory 68 and executes it from the first address (step 230).

In this aperture value command output processing program, the calculation control unit 6
1 stores the contents f1 of the aperture value F, which will be described later, in the data memory 69.
The camera 40 takes out the data from the aperture value output M and outputs the data to the aperture value output M, operates the automatic aperture adjustment V & mechanism 50 to perform aperture adjustment, and the camera 40 captures an image only for the time until the automatic aperture adjustment N mechanism 50 completes its operation. In order to prevent video from being captured, the content t2 of the address T2 previously stored in six data memories is taken out and multiplied by a time delay of this t2 time.

This timer program terminates the camera aperture value command output processing program when the arithmetic control unit 61 counts up the clock every second of the timer module 71 and does not reach t2 seconds, and starts the next image capture program. Transfer program execution to the address.

In the image capture program of step 240, the arithmetic control unit 61 connects the video input module 64 to the video input module 64 via multipath.
A command is given to the camera 40 to capture an image signal from the camera 40. The video input module 64 then connects the camera 40
The image is captured from the first signal in synchronization with the image bus, and the screen as shown in FIG.
is written as information for one screen, and stored in the memory 65 as an image name AI. After completing this program processing, execution of the program is moved to the start address of the next character frame detection program.

Therefore, the arithmetic control unit 61 stores the program starting from this start address in the program memory 68 that is stored in advance.
, and execution starts from the first address (step 250).

In this character frame detection processing program, the image 1 image memory 65
The character frame is detected from the image stored in the image by image processing, and a detailed flowchart is shown in FIG. The image processing processor 62 takes out the image AI and stores it in the image processing program memory 63.
A histogram calculation program is executed to divide the density of 0 to 255 stored in 32 into 32 parts and obtain the density distribution.

A diagram in which the median values of the histogram obtained by this calculation are connected by a curved line is shown by curve I in FIG. 8(a).

Subsequently, the image processing processor 62 retrieves and executes the density conversion program stored in the image processing program memory 63. This density conversion program focuses on the fact that the density close to 0 in the curve I rarely occurs, and the density is below a (this density a is assumed to have been written in advance at address A of the data memory 69). cut and
Concentration a is newly changed to concentration O, and concentration 255 is also changed to 255.
Perform linear transformation using the straight line ■, and make the curve obtained as a result of the calculation a curve ■ as shown in Figure 8 (b),
It is stored in the image memory 65 as the image name A2.

Since this image A2 has been subjected to density expansion processing, it is an image with improved contrast (step 251>
.

After performing the density enlargement process in this manner, a smoothing process program is executed (step 252>).

In the smoothing processing program, if a few black pixels in the white area or a few zero pixels in the ring area are mixed in, it will interfere with the next processing, so it is necessary to avoid high frequency noise etc. riding on these video signals. Smoothing is performed to remove minute parts. That is, the image gAAz is stored in the image memory 6
3x centered on the pixel (t, J) extracted from 5
The image processing program memory 63 stores a smoothing program that calculates the average value of the density of each point in the square area of 3, smooths it, and sets a new density f (f, j>).
, and executed by the image processing processor 62, and the result is stored in the image memory 65 as an image name A3.

Next, perform image binarization processing (step 253)
.

In this binarization processing step 253, the image A3 is taken out from the image memory 65, and an all-pixel average density program for adding up the densities of all the pixels (256 x 256) to obtain an average value is taken out from the image processing program memory 63, and image processing is performed. The processor 62 executes the execution, and the resulting numerical value sI is stored at the address SI in the data memory 69.
The numerical value m1 written in 1 is extracted and added by the arithmetic control unit 61, and the higher density is white (255>) and the lower one is black ( 0)
A binarization conversion program that performs 2ffi conversion is taken out from the image processing program memory 63 and executed by the image processing processor 62, resulting in the result shown in FIG. 7(b).
A screen like this is stored in the image memory 65 as an image name A4.

Subsequently, a labeling processing program is executed (step 254).

In this labeling process, the white area 5 in FIG.
In order to label 01 to 512 with numbers 1 to 12 and save the numbers, a labeling program is extracted from the image processing program 63 and executed by the image processor 62, and the white color existing for labels 1 to 12 is stored in the data memory 69. Areas d1 to d12 are stored in table Ll.

In the next step 255, character frame selection processing, a character frame selection program that extracts only character frames 505 to 507 from these labels 1 to 12 and removes the others creates table Ll from six data memories. In order to obtain the circumscribed rectangles of the corresponding areas d1 to d1□ of labels 1 to 12, a circumscribed rectangle calculation program is taken out from the image processing program memory 63 and executed by the image processing processor 6.
2, and 7th for white areas 505 to 507.
As shown in Figure (c), in area 505, the minimum address (
XO51, YO51) and maximum address (XO52, YO
52> and find the minimum white area for labels 1 to 12,
Write each maximum address to table L2.

Then, table L2 is taken out and the minimum (XOII, YOll) and maximum addresses (XO1
2, YO12) as the base, the same address of labels 2 to 12 (XO21, YO21) (XO22, YO22),
Compare ...... and find that the difference between positions YOII and YO21 is within the tolerance ± pI, and the width (XO12 - XO
11) and (X022-XO21) is within the tolerance value ±22, and the height (YO12-YO11) and (YO
22-YO21> is within the tolerance value ±23.

It is assumed that these allowable values ρI, P2, I)3 are written in advance in six data memory addresses PI, P2, P3, respectively.

This is based on the minimum and maximum addresses of label 1, and labels 1, 2, . . . . , 12t, and if the other values are out of tolerance for label 1, then label 3 is compared with label 2 as a reference. ......, compare 12.

In this way, as is clear from FIG. 7(b), when label 5 is used as a reference, label 67 is within the allowable value, so only the minimum and maximum addresses of labels 5, 6.7 are stored in table L2. , and delete the others to end the character frame detection processing program and return to the character frame density deviation value σ calculation processing step 260 of the main program in FIG.

In the character frame one difference value σ calculation processing program, a standard deviation calculation program is taken out from the image processing program memory 63 and executed by the image processing processor 62 to obtain a standard one difference value. This standard deviation calculation program is a processing program of steps 261 to 264 as shown in the flowchart of FIG. The data is written to the address Xt within the address (step 261).

Subsequently, in order to obtain the average value, the content (Xt) of the obtained address Xt is divided by the square of N, that is, the calculation is performed according to the following formula, and the result is written to the address Xa (step 262).

Xt/N2-+Xa Next, the residual sum of squares is calculated based on the following equation (step 263).

i=1 j=1 The result of this operation is written to address Q of data memory 69.

Next, the standard deviation σ is calculated based on the following formula (step 264).

The result of this calculation is written to address V of the data memory 69, and the character frame density deviation value σ calculation program is thus completed, and the program proceeds to the next optimum aperture tL F 2 extraction program.

In the optimal aperture value F2 retrieval processing program in step 270, as shown in FIG. 10, the digital value k is written in advance in table L3 for the divided area of standard deviation σ, and the value is retrieved and written in address F2. . Through this process, the voltage applied to the automatic diaphragm adjustment mechanism 50 for aperture values 2 to 22 can be taken out from the table. In addition, in order to determine the optimum aperture value, the camera 40
It is necessary to experimentally measure the optimum aperture value for the standard deviation σ at the installation site and create a table based on that data.

Here, Xij on the screen is from the address (1, 1> in the upper left of the screen to the address (1, 2>) horizontally, as shown in Figure 9.
,...,(1,N), the second line also has addresses (2,1), (2,2>,...(2,N>) from the left, and the left for each line sequentially. Add the coordinates from the left, and the last line is (N, 1), (N2), ..., (N, N
). Then, an arbitrary address (i, j) is composed of one row and j columns.

Next, the process moves to a comparison processing program, and the value f2 of the obtained address F2 is compared with the value f1 of the previous address F1 (step 280>).

In this comparison process, the previous aperture value fl stored in address F1 and the currently determined aperture value (m f 2) stored in address F2 are compared, and if they do not match, the content of address F2 is automatically set as the aperture value. Execute the data writing program to operate the aperture value adjustment mechanism 50 (
Step 290).

This data writing program is a process of writing the contents of address F2 to address F, and then returns to step 230, camera aperture value command output processing, to start execution, and in this way, address F,... This process is repeated until the F2 aperture values match.

When the aperture values of addresses F, . . . F2 match, the process moves to a character detection processing program (step 300).

A detailed flowchart of this character detection processing program is shown in FIG. 6. First, in the density enlargement processing of step 301, areas d5, da, and d7 of labels 5, 6.7 in the retable L+ are extracted from the data memory 69. , a mask program that extracts only the area surrounded by this area and changes the other areas to black is stored in the image processing program memory 6.
3 and stored in the image memory 65 (
jA A + is masked and the same program processing as the density enlargement processing program in step 251 is performed for each of the white areas 505, 506, and 507 in FIG. 7(C), and the results are stored as image base A5.

In this way, the contrast is enhanced by emphasizing the shading, and the next image binarization processing step 302 is executed.

This image binarization processing program performs the same program processing as the image binarization processing in the character frame detection processing program, converts the image (gE A 5) into a white and black binary image, and converts the image f Store it as A6.

Subsequently, in step 303, a minute black area erasing program is retrieved from the image processing program memory 63 and executed by the image processing processor 62, and although the contrast l~ has been improved due to the emphasis on shading, the noise on the screen has become noticeable. Eliminates noise on A6 images.

In this way, the aperture value that allows the characters in the character frame of the train number display 20 attached to the front or rear of each train to be clearly recognized is determined, and the aperture value of the camera 40 is automatically adjusted to the determined aperture value. However, by taking an image of the train number display 20, it is possible to accurately detect the train number.

In the above embodiment, the character frame of the train number display 20 is 3.
However, if the inside of the character frame is illuminated with fluorescent light, even if the center character frame is clear, the left and right character frames are not necessarily clear. By setting an appropriate aperture value for each character frame and capturing the image with the new aperture value, clear images of each character frame can be obtained.

[Effects of the Invention] As described above, according to the present invention, in order to obtain a clear image within the character frame area by capturing an image of the train number display mounted on the train and its surroundings with a camera and processing the image, The automatic aperture adjustment mechanism predicts the optimal aperture value for each train and adjusts the camera aperture value for each train, making it possible to clearly read the train number characters on the train number display.
Accurate train number characters can be written.

[Brief explanation of the drawing]

Fig. 1 is a block diagram of an embodiment of the present invention, Fig. 2 is a block diagram showing the overall configuration of the above embodiment, Fig. 3 is a flowchart explaining the operation of the above embodiment, and Fig. 4 is a block diagram of the above embodiment. FIG. 5 is a flowchart of the character frame density difference value σ calculation routine in the above embodiment, FIG. 6 is a flowchart of the character frame detection routine in the above embodiment, and FIG. 7 is a flowchart of the character frame detection routine in the above embodiment. An explanatory diagram showing the image processing process in the example, FIG. 8 is an explanatory diagram of the density enlargement process in the above embodiment, FIG. 9 is an explanatory diagram of coordinate display in the above embodiment, and FIG. 10 is a deviation value σ in the above embodiment. This is a table for converting from k to digital value k. 10... Train 20... Train number display 30... Train detection sensor 40... Camera 50
... Automatic aperture adjustment mechanism 60 ... Image processing device

Claims (1)

[Claims] A train detection sensor that detects the passage of the front or rear side of a train traveling along the tracks, and a camera that receives signals from this train detection sensor and images the area around the train number display installed on the front or rear side of the train. Then, an image processing device takes in the image taken by this camera and predicts the aperture value that will give the clearest image from the image within the character frame of the train number display, and the aperture command value output by this image processing device A train number detection device comprising an automatic diaphragm adjustment mechanism that operates accordingly.