JP6034767B2 - Rail detection device, railroad front monitoring camera device, railcar, rail detection method and control method - Google Patents

Description

  The present invention relates to a rail detection device, a railway front monitoring camera device, a railway vehicle, a rail detection method, and a control method.

  A camera device is mounted on the forefront of a railway vehicle to identify railway signals and signs ahead of the train (see, for example, Patent Document 1), and to detect crossing obstacles and the like.

  There is also a technology that installs a wide-angle camera and a telephoto camera together, finds a place to be imaged from a wide field of view obtained by the wide-angle camera, and points the telephoto camera at that place to take an image ( For example, see Patent Document 2). Such technology can also be used for forward monitoring of railway vehicles.

JP 2008-215938 A Japanese Patent No. 5152758

  As described above, when a part to be imaged is searched from a wide field of view obtained by a wide-angle camera and the telephoto camera is pointed at the part, the part to be imaged may not be determined quickly. .

  For example, there may be a case where an erroneous detection of a location to be imaged occurs, such as when there is another target at the location to be imaged, or when it is difficult to visually recognize sunlight or the like reflected at the location to be imaged. When such erroneous detection occurs, the location to be imaged may not be determined quickly because an incorrect imaging direction is instructed or the location to be imaged changes frequently in a short time. This may interfere with the forward monitoring of the railway vehicle.

  The present invention has been performed under such a background, and can eliminate erroneous detection of a location to be imaged, a rail front monitoring camera device, a railroad vehicle, a rail detection method, and a rail detection method, and An object is to provide a control method.

  The first aspect of the present invention is a viewpoint as a rail detection device. The present invention selects a predetermined track pattern from a track pattern group having different rail curvatures in a rail detection device that detects a rail from image information obtained as a result of imaging by a camera that images the front of a railway vehicle. The selection means compares the curvature of the rail in the previously selected line pattern with the curvature of the rail in the line pattern scheduled to be selected this time, and based on the similarity of the two curvatures, The line patterns that permit selection are narrowed down, and a predetermined line pattern is selected from the narrowed line patterns.

  A second aspect of the present invention is a railway front monitoring camera device. The present invention relates to a railway front monitoring camera device that includes a wide-angle camera and a telephoto camera that capture an image in front of a railway vehicle, and sets the imaging direction of the telephoto camera according to image information of the wide-angle camera. In the image information of the wide-angle camera, the rail vanishing point where the two rails of the railroad appear to overlap or are interrupted is detected from the image information of the wide-angle camera, and the direction of the detected rail vanishing point is imaged by the telephoto camera The control means for moving the telephoto camera in the determined imaging direction is determined, and the control means includes a plurality of types of line patterns as rail image information as a line pattern group, and is actually captured by a wide-angle camera. Select a line pattern that approximates the image information of the rail from the line pattern group, replace the selected line pattern with the actual rail image information, and delete the rail. When adopting the point detection, the curvature of the rail in the previously selected line pattern is compared with the curvature of the rail in the line pattern scheduled to be selected this time. The control which narrows down the track | line pattern which permits selection is performed.

  A third aspect of the present invention is a viewpoint as a railway vehicle. The present invention is a railway vehicle comprising the railway front monitoring camera device of the present invention.

  The fourth aspect of the present invention is a viewpoint as a rail detection method. The present invention relates to a rail detection method executed by a rail detection device that detects a rail from image information obtained as a result of imaging by a camera that images the front of a railway vehicle. A selection step of selecting a track pattern of the rail, and the selection step compares the curvature of the rail in the previously selected track pattern with the curvature of the rail in the track pattern to be selected this time, and based on the similarity of the curvatures of the two Then, the line patterns that allow selection from the line pattern group are narrowed down, and a predetermined line pattern is selected from the narrowed line patterns.

  A fifth aspect of the present invention is a control method for a railway front monitoring camera device. The present invention relates to a railway front monitoring camera apparatus that has a wide-angle camera and a telephoto camera that capture an image in front of a railway vehicle, and sets the imaging direction of the telephoto camera according to image information of the wide-angle camera. In the control method, the control means detects the rail vanishing point at which the two rails of the rail appear to overlap or are interrupted on the image information of the wide-angle camera from the image information of the wide-angle camera, and the detected rail vanishing point It has a control step of determining the direction to the imaging direction of the telephoto camera and moving the telephoto camera in the determined imaging direction, and the control step comprises a plurality of types of track patterns as rail image information as a track pattern group, A line pattern that approximates the image information of the actual rail captured by the wide-angle camera is selected from the line pattern group, and the selected line pattern is Compared with the curvature of the rail in the previously selected track pattern and the curvature of the rail in the track pattern scheduled to be selected this time, when using it to detect the rail vanishing point instead of the image information of The step of narrowing down the line pattern which permits selection from the line pattern group based on the above.

  According to the present invention, it is possible to eliminate erroneous detection of a location to be imaged.

It is a block block diagram of the railroad front monitoring camera apparatus which concerns on embodiment of this invention. It is a figure for demonstrating a rail vanishing point, and a rail is a figure which shows a straight drive state. It is a figure for demonstrating a rail vanishing point, and is a figure which shows the state which the rail curves to the left. It is a figure for demonstrating a rail vanishing point, and is a figure which shows the state in which the rail is curving to the right. It is a figure for demonstrating a rail vanishing point, and is a figure which shows the state in which the rail is the downward slope ahead. It is a figure for demonstrating a rail vanishing point, and is a figure which shows the state in which the rail is an uphill gradient ahead. It is a figure for demonstrating the process of a rail vanishing point change, and a rail is a figure which shows a straight-ahead state. It is a figure for demonstrating the process of the change of a rail vanishing point, and a rail is a figure which shows the state of a left curve a little. It is a figure for demonstrating the process of the change of a rail vanishing point, and is a figure which shows the state in which the rail is carrying out the left curve much larger than FIG. It is a figure for demonstrating the process of the change of a rail vanishing point, and is a figure which shows the state in which the rail has left-curved much larger than FIG. It is a figure for demonstrating the process of the change of a rail vanishing point, and is a figure which shows the state in which the rail is carrying out the left curve much larger than FIG. It is a figure which extracts the rail vanishing point and shows the state of the change of the rail vanishing point explained in FIGS. It is a figure which shows the state of the change of the rail vanishing point shown in FIG. 12 on a time axis. It is a figure which compares the kind of track pattern and the state of transition of a track pattern by comparing a desirable transition and a round-robin transition. It is a figure which shows the data of a line pattern. It is a figure which shows the data regarding the centerline of the data of the line pattern shown in FIG. It is a figure which shows the relationship between the data regarding the centerline shown in FIG. 16, and numerical formula. It is a figure which shows the similarity matrix of a line pattern, and the selectable range. It is a figure which shows the similarity matrix of a line pattern, a similarity rank, and a selectable range. It is a figure which shows the kind of line pattern which concerns on this Embodiment, and the state of the transition of a line pattern. It is a flowchart which shows operation | movement of the imaging position determination part of FIG.

  A railway front monitoring camera device 1 (hereinafter abbreviated as camera device 1) according to an embodiment of the present invention will be described with reference to FIGS.

  As shown in FIG. 1, the camera device 1 is mounted at a front portion in the traveling direction of a railway vehicle 2 (hereinafter abbreviated as a vehicle 2), and images a situation in front of the vehicle 2 including a rail R. The image information in front of the vehicle 2 acquired by the camera device 1 is used, for example, for finding an obstacle on a railroad track or a railroad crossing in order to operate the vehicle 2 safely.

  The camera device 1 includes a wide-angle camera 3 and a telephoto camera 4 that capture an image in front of the vehicle 1, and sets the imaging direction of the telephoto camera 4 according to image information of the wide-angle camera 3. The telephoto camera 4 is placed on an electric pan head 5. The pan / tilt head 5 is controlled in driving by the imaging position determination unit 6 and moves in the direction instructed by the imaging position determination unit 6 to set the imaging direction of the telephoto camera 4. The image information of the telephoto camera 4 is taken into the forward monitoring image processing unit 7 and subjected to image analysis, and the image analysis result is used for monitoring the front of the vehicle 2. In the present embodiment, movement in the imaging direction means turning on a plane substantially parallel to the ground.

  The imaging position determination unit 6 detects, on the image information of the wide-angle camera 3, a rail vanishing point at which the two rails R of the railway appear to overlap or are interrupted from the image information of the wide-angle camera 3, and the detected rail disappearance The direction of the point is determined as the imaging direction of the telephoto camera 4, and the camera platform 5 is driven to move the telephoto camera 4 in the determined imaging direction.

  Here, the rail vanishing point will be described in detail with reference to FIGS. FIG. 2 is an example of the rail vanishing point D1 in the straight rails R1 and R2. In the case of the straight rails R1 and R2, the rail vanishing point D1 is a point where the two rails R1 and R2 appear to overlap on the image information beyond the resolution limit at a distance.

  FIG. 3 is an example of the rail vanishing point D2 in the rails R3 and R4 that are curving to the left. In the rails R3 and R4 that are curved to the left, the rail vanishing point D2 is a point where the rails R3 and R4 appear to overlap on the image information.

  FIG. 4 is an example of the rail vanishing point D3 in the rails R5 and R6 that are curving to the right. In the rails R5 and R6 that curve to the right, the rail vanishing point D3 is a point where the rails R5 and R6 appear to overlap on the image information.

  FIG. 5 is an example of the rail vanishing point D4 in the rails R7 and R8 that are inclined downward in the front. In the rails R7 and R8 having a downward slope in front, both the rails R7 and R8 are interrupted on the image information before the two rails R7 and R8 appear to overlap each other beyond the resolution limit on the image information. In such a case, the vicinity of the farthest intermediate point between the rails R7 and R8 when the rails R7 and R8 are interrupted from the image information is defined as a rail vanishing point D4.

  FIG. 6 is an example of the rail vanishing point D5 in the rails R9 and R10 that are inclined upward in the front. In the rails R9 and R10 having an upward slope in front, both the rails R9 and R10 reach the peak of the upward slope before the two rails R9 and R10 appear to overlap each other beyond the resolution limit on the image information. . In such a case, the vicinity of the intermediate point between the rail R9 and the rail R10 when the rails R9 and R10 reach the peak of the upward slope is defined as a rail vanishing point D5.

  Next, a change in the position of the rail vanishing point will be described with reference to FIGS. FIGS. 7 to 11 show, as an example, changes in the series of rail vanishing points D10 to D14 until the rail R of the railway curves from the straight traveling state to the left.

  In FIG. 7, the rail R is in a straight state, and the rail vanishing point D <b> 10 is positioned at the upper portion in the vertical direction at the central portion in the horizontal direction. In FIG. 8, the rail R has begun to turn slightly to the left, and the rail vanishing point D11 is located slightly above the horizontal center and at the top in the vertical direction compared to the rail vanishing point D10. In FIG. 9, the rail R is further bent to the left from the state of FIG. 8, and the rail vanishing point D12 is further to the left than the center portion in the horizontal direction and slightly above the vertical direction compared to the rail vanishing point D11. Located in. In FIG. 10, the rail R is further bent to the left from the state of FIG. 9, and the rail vanishing point D13 is located slightly to the left in the horizontal direction and slightly above the vertical direction compared to the rail vanishing point D12. In FIG. 11, the rail R is further bent to the left from the state of FIG. 10, and the rail vanishing point D <b> 14 is located slightly to the left in the horizontal direction and slightly lower in the vertical direction than the rail vanishing point D <b> 13. When such rail vanishing points D10 to D14 of FIGS. 7 to 11 are plotted on the same space axis, they are as shown in FIG.

  As shown in FIG. 7, when the rail R is in a straight state, the rail vanishing point D10 remains at a single point with almost no movement. Eventually, as shown in FIG. 8, when a left curve is reached, a new rail vanishing point D11 is detected, and a new imaging direction of the telephoto camera 4 is determined. Further, as shown in FIGS. 9 to 11, new rail vanishing points D12 to D14 are detected one after another in the left curve, and new imaging directions of the telephoto camera 4 are determined one after another.

  FIG. 13 shows such a change in the rail vanishing points D10 to D14 on the time axis. That is, when approaching the left curve, the curvature of the rail is loose, and the time t1 for switching from the rail vanishing point D10 to the rail vanishing point D11 is relatively long. When a left curve is entered, the time t2 at which the rail vanishing point D11 switches to the rail vanishing point D12 becomes shorter than the time t1. Subsequently, a time t3 at which the rail vanishing point D12 is switched to the rail vanishing point D13 is shorter than the time t2. Furthermore, the time t4 when the rail vanishing point D13 is switched to the rail vanishing point D14 is shorter than the time t3.

  Next, rail detection for rail vanishing point detection will be described with reference to FIG. FIG. 14 is a diagram showing a plurality of types of line patterns prepared in advance. In the example of FIG. 14, five types of line patterns S, L1, L2, R1, and R2 are illustrated.

  As shown in FIG. 14, the imaging position determination unit 6 stores a line pattern group in a memory (not shown). The imaging position determination unit 6 selects a line pattern that approximates the actual rail image information captured by the wide-angle camera 3 from the line pattern group. Furthermore, the imaging position determination unit 6 adopts the selected track pattern for detection of the rail vanishing point in place of the actual rail image information. Thereby, when noise is included in the image information of the actual rail, this image information can be replaced with image information of a previously prepared line pattern that does not include noise. Thereby, since the detection of the rail vanishing point can be performed based on the image information of the rail that does not include noise, the rail vanishing point can be detected with high accuracy.

  The line pattern S in FIG. 14 is a line pattern in which the rail is in a straight traveling state. The line pattern L1 in FIG. 14 is a line pattern with a relatively loose left curve. The line pattern L2 in FIG. 14 is a line pattern having a relatively steep left curve. The line pattern R1 in FIG. 14 is a relatively gentle right curve line pattern. The line pattern R2 in FIG. 14 is a line pattern with a relatively steep right curve.

  In general, railroad rails are laid so that there are as many straight sections as possible and curved portions are as loose as possible. In view of such a rail installation situation, in each of the line patterns S, L1, L2, R1, and R2 in FIG. 14, the transition of the line pattern is usually performed between adjacent line patterns. That is, for example, a situation in which the transition of the line pattern changes from the line pattern S of FIG. 14 to the line pattern L2 or the line pattern R2 is not conceivable. If such a line pattern transition is performed, it can be determined that this is a result of erroneous detection of the rail.

  Dotted arrows between the line patterns S, L1, L2, R1, and R2 in FIG. 14 simply indicate combinations of the line patterns S, L1, L2, R1, and R2. Here, if the transition between the line patterns is permitted for all combinations of the line patterns S, L1, L2, R1, and R2, the imaging position determination unit 6 detects a rail that is erroneously detected. But there is a transition between line patterns. As a result, useless processing time occurs in the imaging position determination unit 6. Furthermore, the detection result of the rail vanishing point of the imaging position determining unit 6 is deteriorated because the result of the erroneous detection is directly used for the detection of the rail vanishing point. In this case, the processing time of the imaging position determination unit 6 becomes longer as the number of line patterns constituting the line pattern group increases.

  On the other hand, the broken-line arrows between the line patterns S, L1, L2, R1, and R2 in FIG. 14 indicate the desired line patterns S, L1, L2, R1, and R2 when detecting railroad rails. Indicates the combination of transitions between. That is, the broken-line arrows between the line patterns S, L1, L2, R1, and R2 in FIG. 14 permit only transitions between adjacent line patterns. According to this, it is possible to eliminate erroneous rail detection in the imaging position determination unit 6 and it is possible to eliminate useless processing time of the imaging position determination unit 6. Furthermore, according to this, even when the number of line patterns constituting the line pattern group increases, the processing time of the imaging position determination unit 6 does not increase. For this reason, the imaging position determination unit 6 is configured as described below in order to realize the transition indicated by the dashed arrows between the line patterns S, L1, L2, R1, and R2 in FIG. Have

As shown in FIG. 15, the imaging position determination unit 6 determines the positions of the upper and lower ends of the left and right rails and the center portion of six numerical data x _Ltop , x _Rtop and x _Lbottom , x _Rbottom, and x _Lmid , x _Rmid. Expressed by The similarities are calculated by reading the curvature of each line pattern from these data, and the constraint condition of the line pattern that can be selected next from the current line pattern is set. Thereby, a desirable combination of transitions between the respective line patterns S, L1, L2, R1, and R2 as indicated by broken lines in FIG. 14 can be realized. According to this, erroneous detection of the rail of the imaging position determination unit 6 can be eliminated, and the processing speed is improved. Further, according to this, even if the number of line patterns constituting the line pattern group prepared in advance increases, the processing speed of the imaging position determination unit 6 does not increase.

Next, a method for defining the similarity between line patterns using the curvature of the line patterns will be described. The upper end of the rail center line, the lower end, and an intermediate position can be determined line pattern data x _Ltop, the average of x _Rtop, x _Lmid, the average of x _Rmid, and x _Lbottom, the average of x _Rbottom. That is,
x _Ctop = (x _Ltop + x _Rtop ) / 2
x _Cmid = (x _Lmid + x _Rmid ) / 2
x _Cbottom = (x _Lbottom + x _Rbottom ) / 2
As required. FIG. 16 shows rail centerline data.

As shown in FIG. 16, there are three rail centerline data, and this point is approximated by a quadratic equation. When the vertical direction is the y-axis, the rail at the lower end y = 0 is almost vertical, so x (0) ′ = 0. Considering this condition, the rail center line is
x (y) = a + cy ² (1)
Can be expressed. FIG. 17 is a diagram illustrating the relationship between Equation (1) and the rail center line in the line pattern. At this time, the value of a represents the position on the x-axis on the image information, and the value of c is an index representing the curvature of the rail on the image information. Therefore, the line pattern similarity is defined using the value of c. When the value of c is 0, it is a straight line. When the value of c is positive, it is a right curve, and when it is negative, it is a left curve. The curve is steeper as the value of c is larger.

The value of c, which is an index of curvature, is calculated using the least square method.
c = (Σx _i Σy _i ² −NΣx _i y _i ² ) / ((Σy _i ² ) ² −NΣy _i ⁴ )
Is required. The line pattern data can be expressed as (x _Cbottom , 0), (x _Cmid , 0.5), (x _Ctop , 1) when the length in the y direction is normalized to 1. Substituting this into the above formula,
c = (1.75 _× Ctop− _0.5 × _Cmid− 1.25 _{× Cbottom} ) /1.625
Is required.

The difference between the similarities c _i and c _j obtained for the line patterns i and j is defined as the similarity d _{i, j} .
d _{i, j} = | c _j -c _i |
And create a similarity matrix with the values as components. As the line pattern that can be ordered up to the top M rank other than the target line pattern itself by ordering the similarities in order of closeness
Line pattern that can be selected from i = {j |. The upper M rank where the values of i and d _{i, j} are small}
A method of providing a constraint condition is conceivable. Alternatively, using the threshold value d _th for the value of the similarity d _{i, j} ,
Line pattern selectable from i = {j | d _{i, j} <d _th }
The method of thinking is also conceivable. When the next line pattern is selected for each line pattern, it is possible to eliminate erroneous detection when selecting the line pattern by selecting only from the selectable line patterns determined here.

Regarding the number M of selectable line patterns and the threshold value d _th , if there are few options, there is a possibility that the response to a change in the situation such as a curve becomes too slow. Further, according to this, it may take time to recover from erroneous detection. On the other hand, if there are too many options, the number of selectable line patterns will increase even if a transition restriction is provided between the line patterns, making it difficult to eliminate erroneous detection. Considering these points, adjustment to an appropriate value is necessary. Hereinafter, a method of obtaining a line pattern that can be selected from the similarity will be described with reference to FIGS. 18 and 19.

  FIG. 18 shows the similarity matrix in the left figure, and the selectable range of the line pattern in the right figure. As shown in the left diagram of FIG. 18, the similarity matrix shows that the similarity between the line patterns L2, L1, S, R1, R2 and the line patterns L2, L1, S, R1, R2 is naturally 0. 0. In other cases, the numerical value increases as the similarity decreases. In the example of FIG. 18, the similarity between the line pattern L2 and the line pattern L1 is 0.1. The similarity between the line pattern L2 and the line pattern S is 0.2. The similarity between the line pattern L2 and the line pattern R1 is 0.4. The similarity between the line pattern L2 and the line pattern R2 is 0.6.

  As shown in the right diagram of FIG. 18, when the transition between the line patterns having a similarity of less than 0.2 is selectable (circle mark in the figure), the line pattern between the line pattern L2 and the line pattern L1 Transitions can be selected between S and the line pattern L1 or the line pattern L2, between the line pattern R2 and the line pattern R1, and between the line pattern S and the line pattern R1 or the line pattern R2.

  Alternatively, the similarity ranking is determined based on the left diagram of FIG. 19 (same as the left diagram of FIG. 18) as shown in the right diagram of FIG. That is, when the similarity is 0, the similarity ranking is 0, when the similarity is 0.1, the similarity ranking is 1, and when the similarity is 0.2, the similarity ranking is 2. When the similarity is 0.4, the similarity ranking is 3, and when the similarity is 0.6, the similarity ranking is 4.

  If the transitions between the line patterns having similarity rank 0 or 1 in the similarity ranks in the right diagram of FIG. 19 are selectable (circles in the diagram), the selectable range as shown in the lower diagram of FIG. Become. The lower diagram in FIG. 19 is the same as the right diagram in FIG.

  By determining the selectable range of transition between line patterns as described above, selectable line patterns are limited as shown by solid arrows in FIG. 20, and each desired line pattern S shown by broken arrows is shown. , L1, L2, R1, and R2 are the same as the combination of transitions. In addition, the arrow of the dashed-two dotted line of FIG. 20 shows the combination of the limited selectable line pattern.

  Next, the operation of the imaging position determination unit 6 will be described with reference to the flowchart of FIG. The condition of “START” in FIG. 21 is a condition that the camera apparatus 1 is in operation and any line pattern has already been selected by the imaging position determination unit 6. Note that the processing from “START” to “END” in FIG. 21 is processing for one cycle, and when the condition of “START” is satisfied when the flow in FIG. 21 ends (END), The flow starts again. If the “START” condition is satisfied, the process proceeds to step S1.

  In step S <b> 1, the imaging position determination unit 6 determines whether there is a change in the line pattern based on the image information of the wide-angle camera 3. If it is determined in step S1 that the line pattern has changed, the process proceeds to step S2. On the other hand, if it is determined in step S1 that there is no change in the line pattern, the process repeats step S1.

  In step S2, the imaging position determination unit 6 determines whether the new line pattern to be selected is within a predetermined selectable range based on the change in the line pattern in step S1. If it is determined in step S2 that the new line pattern to be selected is within a predetermined selectable range, the process proceeds to step S3. On the other hand, if it is determined in step S2 that the new line pattern to be selected is outside the predetermined selectable range, the process returns to step S1 because there is an error in the determination in step S1.

  In step S3, the imaging position determination unit 6 changes the line pattern based on the determination in step S1 and ends the process for one cycle (END).

  As described above, the imaging position determination unit 6 detects the rail from the image information obtained as a result of imaging by the wide-angle camera 3 that images the front of the railway vehicle 1. At this time, the imaging position determination unit 6 selects a predetermined line pattern from the line pattern groups having different curvatures of the rail R, and the curvature of the rail in the previously selected line pattern and the rail in the currently selected line pattern. Are compared, the line patterns that are allowed to be selected from the line pattern group are narrowed down based on the similarity between the two curvatures, and a predetermined line pattern is selected from the narrowed line patterns. Thereby, according to the imaging position determination part 6, the erroneous detection of a rail can be eliminated. This also eliminates erroneous detection of rail vanishing points. In addition, according to the imaging position determination unit 6, the number of selectable line patterns does not change even when the number of line patterns constituting the line pattern group is increased, and the processing time of the imaging position determination unit 6 is not increased. be able to. As a result, the forward monitoring image processing unit 7 can perform forward monitoring based on image information without erroneous detection, and therefore can perform forward monitoring with high reliability. Furthermore, since the processing time of the imaging position determination unit 6 does not increase even if the number of line patterns constituting the line pattern group is increased in order to increase the detection accuracy of the rail, the forward monitoring image processing unit 7 changes the monitoring cycle. The reliability of forward monitoring can be improved without any problems.

  Further, the imaging position determination unit 6 can be realized by executing a predetermined program in which the information processing apparatus is installed in advance. Such an information processing apparatus has, for example, a memory, a CPU (Central Processing Unit), an input / output port, and the like. The CPU of the information processing apparatus reads and executes a control program as a predetermined program from a memory or the like. Thereby, the function of the imaging position determination unit 6 is realized in the information processing apparatus. An ASIC (Application Specific Integrated Circuit), a microprocessor (microcomputer), a DSP (Digital Signal Processor), or the like may be used instead of the CPU. The same applies to the function of the forward monitoring image processing unit 7.

  In addition, even if the predetermined program described above is stored in the memory of the information processing apparatus before shipment of the camera apparatus 1, it is stored in the memory of the information processing apparatus after shipment of the camera apparatus 1. It may be a thing. Further, a part of the program may be stored in a memory or the like of the information processing apparatus after the camera apparatus 1 is shipped. The program stored in the memory or the like of the information processing apparatus after the camera device 1 is shipped may be, for example, an installed program stored in a computer-readable recording medium such as a CD-ROM, or the like. The one downloaded via the transmission medium may be installed.

  The predetermined program described above includes not only a program that can be directly executed by the information processing apparatus but also a program that can be executed by being installed on a hard disk or the like. Also included are those that are compressed or encrypted.

  As described above, by realizing the camera device 1 with the information processing device and the program, it is possible to flexibly cope with mass production and specification change (or design change).

  The program executed by the information processing apparatus may be a program that is processed in time series in the order described in this specification, or may be necessary in parallel or when a call is made. It may be a program that performs processing at timing.

  The above-described embodiment can be variously modified without departing from the gist thereof. For example, in the above-described embodiment, the pan head 5 is an operation of only turning in the left-right direction, but may also perform an operation in the up-down direction. According to this, the telephoto camera 4 can move the imaging direction vertically and horizontally. Further, the telephoto camera 4 may have a zoom mechanism.

DESCRIPTION OF SYMBOLS 1 ... Railway front monitoring camera apparatus, 2 ... Rail vehicle, 3 ... Wide angle type camera, 4 ... Telephoto type camera, 5 ... Pan head, 6 ... Imaging position determination part (control means), D1-D5, D10-D14 ... Rail vanishing point, R, R1-10 ... Rail

Claims (5)

In a rail detection device that detects a rail from image information obtained as a result of imaging by a camera that images the front of a railway vehicle,
A selection means for selecting a predetermined line pattern from a group of line patterns with different rail curvatures,
The selection means compares the curvature of the rail in the previously selected line pattern with the curvature of the rail in the line pattern scheduled to be selected this time, and based on the similarity between the two curvatures, Narrow down the line pattern to allow selection from, select a predetermined line pattern from the narrowed line pattern,
A rail detection device characterized by that. In the railway front monitoring camera device, which has a wide-angle camera and a telephoto camera that capture an image of the front of the railway vehicle, and sets an imaging direction of the telephoto camera according to image information of the wide-angle camera.
On the image information of the wide-angle camera, a rail vanishing point at which two rails of the railroad appear to overlap or are interrupted is detected from the image information of the wide-angle camera, and the direction of the detected rail vanishing point is the telephoto type Control means for determining the imaging direction of the camera and moving the telephoto camera in the determined imaging direction;
The control means includes a plurality of types of line patterns that are image information of the rail as a line pattern group, and the line pattern that approximates the actual image information of the rail imaged by the wide-angle camera is included in the line pattern group. When the selected track pattern is used for detection of the rail vanishing point instead of the actual rail image information, the curvature of the rail in the previously selected track pattern and the currently selected rail pattern are selected. Comparing the curvature of the rail in the line pattern, and performing control to narrow down the line pattern that allows selection from the line pattern group based on the similarity of both curvatures,
A railway front monitoring camera device characterized by that.   A railway vehicle comprising the railway front monitoring camera device according to claim 2. In a rail detection method executed by a rail detection device that detects a rail from image information obtained as a result of imaging by a camera that images the front of a railway vehicle,
A selection step of selecting a predetermined line pattern from a group of line patterns with different rail curvatures;
The selection step compares the curvature of the rail in the previously selected line pattern with the curvature of the rail in the line pattern scheduled to be selected this time, and based on the similarity between the two curvatures, Narrow down the line pattern to allow selection from, select a predetermined line pattern from the narrowed line pattern,
The rail detection method characterized by the above-mentioned. A control method for a railroad front monitoring camera device having a wide-angle camera and a telephoto camera that capture an image ahead of a railway vehicle, and setting an imaging direction of the telephoto camera according to image information of the wide-angle camera In
The control means detects, on the image information of the wide-angle camera, a rail vanishing point at which two rails of the railroad appear to overlap or are interrupted from the image information of the wide-angle camera, and the direction of the detected rail vanishing point A control step for determining the imaging direction of the telephoto camera and moving the telephoto camera in the determined imaging direction,
The control step includes a plurality of types of line patterns that are image information of the rail as a line pattern group, and the line pattern that approximates the actual image information of the rail imaged by the wide-angle camera is included in the line pattern group. When the selected track pattern is used for detection of the rail vanishing point instead of the actual rail image information, the curvature of the rail in the previously selected track pattern and the currently selected rail pattern are selected. Comparing the curvature of the rail in a line pattern, and having the step of narrowing down the line pattern that permits selection from the line pattern group based on the similarity of the curvatures of both.
A control method characterized by that.

Images