Warning device

Abstract

To provide notification of a dangerous area with easy preparation and a high alerting effect.SOLUTION: A dangerous area notification device 50 includes a dangerous area determination unit 55 that moves a dangerous line 6 by a predetermined distance in a dangerous direction 7 and determines as a dangerous area 8 an area sandwiched between the dangerous line 6 which has not been moved and the dangerous line 6 which has been moved, a danger determination unit 56 that determines, based on a relationship between an imaging position of a camera 51, which takes a shot image of a danger marker 5, and a position of the dangerous area 8, whether the imaging position of the camera 51 is dangerous, and an alarm giving unit 57 that gives an alarm when the danger determination unit 56 determines that the imaging position is dangerous.SELECTED DRAWING: Figure 6

Description

Technical Field

[0001] The present invention relates to a warning device.

Background Art

[0002] In operations at factories, plants, construction sites, etc., depending on the work location, there are risks including the following types of accidents that can occur. · Falls from heights. This includes not only falls from heights of several meters but also falls (trips) from steps of several tens of centimeters. · Entanglement in moving machinery · Falling of objects from above To prevent these accidents, a system that alerts workers to risks detected from the images of cameras introduced at the work site can be considered.

[0003] Patent Document 1 describes a technology for preventing accidents where a general vehicle breaks through regulatory members (cones) scattered between a work area (safe area) and a driving lane of a general vehicle (dangerous area) and heads towards a road construction site where workers are working. In this technology, a camera fixedly installed near the road construction site monitors general vehicles that break through the cones. An alarm is issued when a change occurs in the cones or the like in the captured image of the camera.

[0004] Patent Document 2 describes an operation support system for a work machine such as an excavator operated by a worker. In this technology, markers installed at the work site are detected using a camera mounted on the excavator, and a no-entry area defined by an imaginary line connecting adjacent markers is set. When the work machine enters the no-entry area, the movement of the work machine is stopped.

Prior Art Documents

Patent Documents

[0005]

Patent Document 1

Patent Document 2

[0006] The fixed camera method described in Patent Document 1 is difficult to apply to walkways in factories, plants, construction sites, and other similar locations for the following reasons. • When the access road at the site is in the air at a high altitude, and there is no physically suitable place to install a camera. • When the condition of surrounding structures is frequently altered as a result of the work. • When there are many target locations and the number of surveillance cameras to be installed becomes enormous.

[0007] Therefore, we will consider using a mobile camera system like the one in Patent Document 2, which uses a camera attached to the worker themselves to identify dangerous areas. The markers in Patent Document 2 use two-dimensional codes, such as QR codes (registered trademark), from which information can be extracted by mechanical image processing. This makes it possible to assign mechanically readable semantic information (such as the marker's location information) to each individual marker. However, because mobile cameras have a limited field of view, dangerous areas in the camera's blind spots will not be captured in the images. Therefore, the range in which dangerous areas can be identified is limited to the mobile camera's field of view.

[0008] On the other hand, multiple workers are working at a construction site, and hazard signs visible to the human eye, such as cones, can help workers identify dangerous areas over a wide area. Therefore, to enhance safety, it is desirable to use both human-visible hazard signs and machine-readable QR codes. However, if both types of hazard signs are to be placed as separate objects at the site, they may block passageways and other areas, and installation and relocation may be time-consuming.

[0009] Therefore, the main objective of the present invention is to provide a method for notifying people of dangerous areas that is easy to prepare and has a high alerting effect. [Means for solving the problem]

[0010] The warning device of the present invention has the following features. The present invention includes a danger marker detection unit that detects danger markers, which have a danger line in the length direction and a mark indicating the direction of danger in the width direction, from images captured by a camera. Based on the danger line and danger direction indicated by the danger marker detected by the danger marker detection unit, the danger line is moved by a first distance in the danger direction, and the area enclosed by the danger line before and after the movement is determined as a danger area by the danger area determination unit. A hazard determination unit determines whether the camera's shooting position is hazardous or not based on the positional relationship between the camera's shooting position, which captured the image of the hazard marker, and the hazardous area. The system is characterized by having a warning issuing unit that issues a warning when the danger determination unit determines that there is danger. Other methods will be described later. [Effects of the Invention]

[0011] According to the present invention, it is possible to notify people of dangerous areas that are easy to prepare and have a high alerting effect. [Brief explanation of the drawing]

[0012] [Figure 1] This is an explanatory diagram showing the hazardous areas to which the present invention is applied in Example 1. [Figure 2] Figure 1 is a schematic diagram showing a plan view of Example 1. [Figure 3] This is an example of a hazard marker related to Example 1. [Figure 4] This is an explanatory diagram of the hazard marker for Example 1. [Figure 5] This is an explanatory diagram of the hazardous area defined by the hazard marker in Figure 4, relating to Example 1. [Figure 6] This is a diagram showing the configuration of the hazardous area notification device according to Example 1. [Figure 7]It is a flowchart showing the processing of the danger marker detection unit related to Example 1. [Figure 8] It is an explanatory diagram showing an example of a method for detecting a danger marker related to Example 1. [Figure 9] It is a flowchart showing the processing of the danger line detection unit related to Example 1. [Figure 10] It is a flowchart showing the processing of the danger direction detection unit related to Example 1. [Figure 11] It is an explanatory diagram of a danger marker showing the details of the processing of the danger direction detection unit related to Example 1. [Figure 12] It is a flowchart showing the processing of the danger area determination unit related to Example 1. [Figure 13] It is a flowchart showing the processing of the danger determination unit related to Example 1. [Figure 14] It is an explanatory diagram of a shooting screen for explaining the processing of FIG. 13 related to Example 1. [Figure 15] It is a plan view of the danger marker of FIG. 11 related to Example 1. [Figure 16] It shows a modified example of the operator position determination process related to Example 1. [Figure 17] It is a plan view showing a high-place work site where two passageways are close to each other related to Example 2. [Figure 18] It is a flowchart with the processing of the danger area determination unit added related to Example 2. [Figure 19] It is an explanatory diagram showing an example where a rectangular opening is open in the central part of the area related to Example 2. [Figure 20] It is an explanatory diagram showing the danger area detected from FIG. 19 related to Example 2. [Figure 21] It is a configuration diagram of the danger area notification device related to Example 3. [Figure 22] It is a plan view of the work site when using a safety marker related to Example 3. [Figure 23] It is an explanatory diagram showing an example of a safety marker related to Example 3. [Figure 24]This is a plan view showing the hazardous area formed by the combined use of hazard markers and safety markers in Example 3. [Figure 25] This is a flowchart showing the processing of the hazardous area determination unit when using a safety marker according to Example 3. [Figure 26] This is a camera image diagram showing the first positional relationship between the hazard marker and the safety marker in Example 3. [Figure 27] This is a camera image diagram showing the second positional relationship between the hazard marker and the safety marker in Example 3. [Figure 28] This is a camera image diagram showing the third positional relationship between the hazard marker and the safety marker in Example 3. [Figure 29] This is a perspective view showing the location where a movable safety fence is installed in the opening of Example 3. [Figure 30] This is a plan view showing the location of Figure 29 in Example 3. [Figure 31] This is a photographic image showing that one hazard marker in Example 3 was separated into two parts. [Figure 32] This is an image diagram showing that one hazard marker in Example 3 was separated into three parts during the photograph. [Figure 33] This is a hardware configuration diagram of the hazardous area notification device for each embodiment. [Modes for carrying out the invention]

[0013] Hereafter, Examples 1 to 3 for carrying out the present invention will be described with reference to the figures, etc. [Examples]

[0014] Figure 1 is an explanatory diagram showing a hazardous area to which the present invention is applied. In this diagram, the elevated walkway 1, running from the foreground to the background, is shown in the center as an image taken by a camera 51 (Figure 5) attached to the helmet or chest of a worker 22 (Figure 2). Safety fences 2 are installed on both sides of the elevated walkway 1, but the safety fences 2 are interrupted in the section of the opening 3. The elevated walkway 1 is suspended in the air like a bridge, and there is no floor surface outside the safety fences 2 on the left and right. Note that the location of safety fence 2 is indicated by a thick black line as safety fence location 4 to clearly distinguish it from hazard marker 5. The thick black line at safety fence location 4 is added for illustrative purposes only; in reality, it is the same color as the floor surface of the other elevated walkways 1.

[0015] To prevent falls from opening 3, a danger marker 5 is affixed to the floor surface of opening 3. The danger marker 5 has the following two pieces of information written on it, with a mark (directional indicator pattern 31 in Figure 3) that is visible to the human eye. The "dangerous section" is a line segment along the length of danger marker 5 and indicates the width of the opening 3. In Figure 2, it corresponds to the section of line segment CD within danger marker 5. The "dangerous direction" is the width direction of the danger marker 5 and indicates which direction, perpendicular to the dangerous section, is outside the opening 3 (for example, a dangerous area without a floor). In Figures 1 and 2, the dangerous direction is the direction in which the cross section parallel to the dangerous section of the triangular mark gradually lengthens; that is, in Figure 1, the direction from left to right of the danger marker 5.

[0016] The hazardous area notification device 50 shown in Figure 6 (described later) detects hazard markers 5 from images captured by a camera 51 worn by worker 22, and identifies the "hazardous area," which is a place where there is a possibility of falling, based on the hazardous section and direction. The hazardous area notification device 50 then issues a warning to worker 22 approaching the hazardous area to prevent them from stepping into the hazardous area and falling. Furthermore, the hazard marker 5 is painted with marks that allow the human eye to visually identify the hazardous area and direction, and in conjunction with the machine recognition of the hazardous area notification device 50, it also helps workers to visually identify the hazardous area. In other words, the hazard marker 5 is composed of a single object that combines a human-visible hazard warning sign with a machine-readable code.

[0017] In Figure 1, the hazard marker 5 was explained using an example of a location with a risk of falling, but the present invention using the hazard marker 5 can be applied to any dangerous area, not just areas with a risk of falling. For example, by installing the hazard marker 5 to surround an area where there is a risk of a large component falling over, it becomes possible to prevent being trapped or crushed by the falling component. Also, by indicating the operating range of a movable machine with the hazard marker 5, it becomes possible to prevent contact with or entanglement with the machine.

[0018] Figure 2 is a schematic diagram showing Figure 1 in plan view. Worker 22 is facing from the bottom to the top of the figure and is moving upwards. Both sides of the elevated walkway 1 are lower areas 21, and the floor surface of the lower areas 21 is several tens of centimeters to several meters lower than that of the elevated walkway 1. In Figure 2, the safety fence area 4 is indicated by a thick black line. The hazard marker 5 is a quadrilateral ABCD with sides AB and CD in the length direction and sides AC and BD in the width direction.

[0019] Figure 3 shows an example of a danger marker 5. In Figures 1 and 2, the vertical direction of the drawing was defined as the length direction of danger marker 5 (dangerous section), but in Figure 3, the horizontal direction of the drawing is defined as the length direction of danger marker 5. The danger marker 5 has a strip-like shape, with its length being longer than its width. The danger marker 5 has an area where multiple directional indicator patterns 31 are drawn, and an area of ​​background 32 where no directional indicator patterns 31 are drawn. Because multiple directional indicator patterns 31 of the same type are arranged along the same danger line 6, the danger line detection unit 53 shown in Figure 6 (described later) can perform image recognition along the length of the danger marker 5 (danger line 6). The color of the background section 32 is such that it can be clearly distinguished from the floor surface of the elevated walkway 1. For example, in factories, the floor surface is often green or gray, so yellow or orange is preferable for the background section 32. The color of the direction indicator pattern 31 is such that it can be clearly distinguished from the background section 32. For example, if the background section 32 is yellow or orange, black is preferable for the direction indicator pattern 31.

[0020] The danger marker 5 should preferably be at least 5 cm in width and at least 30 cm in length so that, for example, a person standing can see the direction indicator pattern 31 of the danger marker 5 placed on the floor within 5 m of their standing position. Similarly, the size of the direction indicator pattern 31 should preferably be at least 4 cm square so that it can be seen by a person standing. Furthermore, it is desirable that the width of the danger marker 5 and the pitch of the direction indicator pattern 31 (the distance between corresponding vertices of adjacent patterns) are fixed values. By having these fixed values, the width of the danger marker 5 and the pitch of the direction indicator pattern 31 on the screen can be used as a reference for measuring distance and interval, allowing for easy and highly accurate measurement of the distance and interval between two points on the screen. Since the pitch of the direction indicator pattern 31 is a fixed value, the number of direction indicator patterns 31 is approximately proportional to the length of the danger marker 5.

[0021] The shape of the direction indicator pattern 31 is shown as a triangle in Figures 1 and 2, but any shape is acceptable as long as it can identify the dangerous direction (all upward in Figure 3). The danger markers 5a-5e in Figure 3 are shown below as examples. Hazard marker 5a is formed by making the direction indicator pattern 31 a right triangle and having the same pitch as the direction indicator pattern 31. By making the direction indicator pattern 31 a right triangle, the end of the hazard marker 5 and the vertex of the direction indicator pattern 31 that is not right angle coincide, so that regardless of the length of the hazard marker 5, the other end becomes a similar right triangle, and it is possible to prevent a decrease in the detection accuracy of the direction indicator pattern 31 due to the absence of a part of the direction indicator pattern 31.

[0022] In the case of danger marker 5a, two or more directional indicator patterns 31 are required to determine danger direction 7. On the other hand, danger markers 5b-5d can determine danger direction 7 with only one directional indicator pattern 31. Furthermore, each has the following effects. Because the hazard marker 5b has a more complex directional indication pattern 31, the probability of misidentifying it as a similar-shaped pattern other than the hazard marker is reduced. Because the danger marker 5c has a relatively simple shape, the probability of detecting the direction indicator pattern 31 is high even when the danger marker 5 is far away or when the downward angle of the camera 51 is small, causing the distortion of the danger marker 5 to increase.

[0023] The danger marker 5d makes it easy to identify the direction indicator pattern 31 because the round portion of the direction indicator pattern 31 looks the same regardless of the viewing direction. Furthermore, by using pattern matching to detect the round portion of the direction indicator pattern 31 first, the detection accuracy of the danger marker 5 can be further improved. In the width direction of the danger marker 5e, the size of the direction indicator pattern 31 is smaller than the width of the danger marker 5. As a result, in the danger marker detection unit 52, the contour of the extracted region 202 in Figure 8 and the extracted region contour portion 203 coincide, making the processing of the danger marker detection unit 52 (Figure 6) easier and more accurate.

[0024] The above describes various embodiments of the danger marker 5 based on differences in the shape of the direction indicator pattern 31. Furthermore, one or more of the physical configuration features of the danger marker 5, as described below as (Configuration 1)-(Configuration 4), may be used in combination. (Configuration 1) Regarding the method of installing the hazard marker 5 on the floor, it is attached to the floor surface using adhesive on the back, similar to adhesive tape. This makes it easy to install the hazard marker 5. Furthermore, since it can be stored in the form of a typical roll of adhesive tape, it is easy to handle and the manufacturing cost is also low.

[0025] (Configuration 2) Hazard Marker 5 is a sheet-like mat, or a mat with magnetic or other adhesive properties that prevents it from shifting during installation. In the case of a sheet-like magnet, it can be installed in places where adhesives are difficult to use, such as gratings. Furthermore, it can respond immediately and easily to sites where hazardous areas change frequently, and the unsafe time until Hazard Marker 5 is installed can be minimized.

[0026] (Configuration 3) The danger marker 5 may be fitted with a fluorescent function or made of a reflective material. However, it is desirable to have different fluorescent or reflective properties for the direction indicator pattern 31 and the background 32. For example, by using a reflective material for the background 32 and a non-reflective material for the direction indicator pattern 31, the danger marker 5 can be easily detected even in dimly lit areas, and the direction indicator pattern 31 and the background 32 can also be easily distinguished.

[0027] (Configuration 4) The danger marker 5 may be displayed electronically. That is, a liquid crystal display, LED array, or electronic paper is installed and the danger marker 5 is displayed thereon. In particular, if the dangerous area changes depending on the situation at the time, such as when a fence is opened or closed, by displaying the danger marker 5 when it is dangerous and not displaying it when it is safe, it becomes possible to notify workers 22 of an accurate warning and reduce the frequency of false alarms (warning that it is dangerous when it is safe).

[0028] Furthermore, as explained in Figure 3, there may be multiple types of hazard markers 5, and multiple types of hazard markers 5 may be used and installed depending on the type or degree of danger. For example, by changing the type of hazard marker 5, such as the density or shape of the directional indicator pattern 31, according to the distance from the floor surface below the area where there is a risk of falling, it becomes possible to issue warnings appropriate to the degree of danger. Alternatively, by changing the distance at which a warning is issued (such as the distance 234 in Figure 15) according to the type of hazard marker 5, it becomes possible to issue a warning earlier in cases of high danger, thereby improving safety.

[0029] Figure 4 is an explanatory diagram of the danger marker 5. In Figure 4, the directional indicator pattern 31 within the danger marker 5 is triangular. Of the four sides of the danger marker 5, the side where multiple directional indicator patterns 31 sides align, i.e., side CD in Figure 4, corresponds to the danger zone, which is the danger line 6 in Figure 5.

[0030] Figure 5 is an explanatory diagram of the hazard area 8 defined by the hazard marker 5 in Figure 4. Note that in Figure 5, the hazard line 6 and the side CD of the hazard marker 5 are drawn slightly offset for illustrative purposes, but in reality they are the same line segment. The danger direction 7 is the direction opposite to the danger marker 5, vertically from danger line 6, that is, the direction opposite to side AB when viewed from danger line 6. When installing danger marker 5, at opening 3, side CD should be aligned with the edge of opening 3, and side AB should be placed on the passage side. When danger marker 5 is installed, the danger area 8 is the quadrilateral CDFE enclosed by danger line 6 (line segment CD) and line segment EF, which is moved a certain distance in danger direction 7.

[0031] The distance between line segment CD and line segment EF (i.e., the length of line segment CE) is a fixed width within the operating range of this system, and this width can be freely determined at the time of starting operation of this system. For example, since a person's step is generally slightly less than 1m, a margin is taken and it will be about 2m. When the danger marker 5 is installed, the direction indicated by the triangle of the direction indication pattern 31 from the danger line 6, that is, the direction from line segment CD to line segment AB, is the safe direction. As in the embodiment described later, the danger marker 5 is judged to be safe or dangerous from the captured image. Since the worker 22 can also visually confirm the dangerous direction, this system can also present the dangerous area 8 to the worker 22 who is not wearing this system.

[0032] Figure 6 is a diagram showing the configuration of the hazardous area notification device 50. The hazard area notification device (warning device) 50 detects hazard markers 5 from images captured by the camera 51 of the worker 22, and determines whether the worker 22's current location is safe or not based on the positional relationship between the hazard area 8 detected from the hazard markers 5 and the worker 22. Therefore, the hazard area notification device 50 includes a camera 51, a hazard marker detection unit 52, a hazard line detection unit 53, a hazard direction detection unit 54, a hazard area determination unit 55, a hazard judgment unit 56, and a warning issuing unit 57.

[0033] The following is an overview of the hazardous area notification device 50. The danger marker detection unit 52 detects danger markers 5, which have marks indicating danger lines 6 in the length direction and danger directions 7 in the width direction, from the image captured by the camera 51. The hazard area determination unit 55, based on the hazard line 6 and hazard direction 7 indicated by the hazard marker 5 detected by the hazard marker detection unit 52, moves the hazard line 6 in the hazard direction 7 by a first distance (the distance of side CE in Figure 5), and determines the area enclosed by the hazard line 6 before the move (side CD in Figure 5) and the hazard line 6 after the move (side EF in Figure 5) (a rectangle CDFE in Figure 5) as the hazard area 8. The hazard determination unit 56 determines whether the shooting position of camera 51, which captured the image of the hazard marker 5, is hazardous or not, based on the positional relationship between the shooting position of camera 51 and the hazardous area 8. The warning issuing unit 57 issues a warning when the danger determination unit 56 determines that there is danger.

[0034] The configuration of the danger marker 5, as illustrated in Figure 3, is shown below. Hazard marker 5 is composed of a rectangle that contains multiple marks placed on the same line along the direction of the hazard line 6 (left-right direction in Figure 3). Each mark contained within the danger marker 5 is composed of an asymmetrical shape that indicates the danger direction 7 with respect to the direction perpendicular to the danger line 6 (the up and down direction in the drawing in Figure 3).

[0035] The specific configuration of the hazardous area notification device 50 is shown below. The hazard area determination unit 55 determines the hazard line 6 at the detection position of the hazard marker 5 as the first side, and the second side by moving the first side a predetermined distance toward the hazard direction 7. The rectangle formed by connecting the first side and the second side as constituent elements is determined as the hazard area 8. The danger determination unit 56 determines that the shooting position of camera 51 is dangerous when the shooting point 241 in the captured image, which indicates the shooting position of camera 51, is included in the danger area 8 (or the warning start area 235 in Figure 15 which includes the danger area 8).

[0036] Alternatively, a camera 51 may be attached to a worker 22, and wireless communication with the camera 51 may be used to process the images captured by the camera 51 in a hazardous area notification device 50 located in a different place from the camera 51 (a configuration in which the camera 51 is physically separated from the hazardous area notification device 50). Furthermore, by having one worker 22 wear multiple cameras 51 and ensuring that the fields of view of each attached camera 51 do not overlap as much as possible, blind spots can be reduced. The video captured by camera 51 is input to the hazard marker detection unit 52 for each frame. The hazard marker detection unit 52 searches for hazard markers 5 in the frame. If no hazard markers 5 are detected in the frame, the location is considered safe, and a safety signal 65 is output. When a safety signal 65 is output, the hazard determination unit 56 unconditionally determines that the worker 22 is safe at the time the frame was captured, and no warnings are issued.

[0037] On the other hand, if the danger marker detection unit 52 detects a danger marker 5 on the screen, it outputs danger marker location information 61. The danger marker location information 61 is the screen coordinates corresponding to A, B, C, and D of the danger marker 5 in Figure 4. If there are multiple danger markers 5 on the screen, the danger marker location information 61 for each is output, and the processing described later is performed for each danger marker location information 61. The danger marker position information 61 is input to the danger line detection unit 53 and the danger direction detection unit 54, and danger line information 62 and danger direction information 63 are output, respectively. The danger line information 62 is the coordinates of the two endpoints corresponding to the danger line 6 on the screen, and the danger direction information 63 is vector information on the screen for the danger direction 7.

[0038] The danger line information 62 and danger direction information 63 are input to the danger area determination unit 55, and the danger area 8 on the screen is determined. The danger area 8 is output as danger area information 64, which includes the coordinates of four points on the screen (corresponding to points C, D, E, and F in Figure 5). The hazardous area information 64 is input to the hazard determination unit 56, which determines whether the location of worker 22 is safe or not at the time the screen is captured. If it is determined that the location of worker 22 is near hazardous area 8 and is therefore dangerous, this fact is input to the warning unit 57. The warning unit 57 notifies worker 22 that it is dangerous. If it is determined that the location of worker 22 is safe, processing for that screen ends, and worker 22 is not notified of anything.

[0039] When the warning unit 57 determines that the worker 22 is in a dangerous location, it notifies the worker 22 of the dangerous location with an alarm sound, vibration, light, etc. If the worker 22 is wearing a wearable display such as a goggle-type display or a transparent or translucent display attached to glasses, a warning signal may be output on the screen. Detailed information such as the location of the dangerous area 8 and the direction in which the dangerous area 8 is located may be displayed on the screen of the wearable display. By displaying detailed information, the worker 22 can more easily intuitively understand the dangerous area, thereby increasing safety.

[0040] Notification to worker 22 may also be by voice. In the case of voice notification, the direction of the hazardous area 8 may be identified from the location information of the hazardous area 8 in the screen captured by camera 51, and a warning may be issued by indicating the direction. That is, if the hazardous area 8 is located towards the right of the screen, by telling worker 22 by voice, "There is a hazardous area 8 to the right," worker 22 can easily intuitively understand the dangerous area, thereby increasing safety. In the case of voice notification, using a bone conduction device will ensure that the content is accurately conveyed even in noisy environments, further increasing safety. The details of each component of the hazardous area notification device 50 will be explained below using Figures 7 to 13.

[0041] Figure 7 is a flowchart showing the processing of the danger marker detection unit 52. The following explanation will describe the process by which the danger marker detection unit 52 sequentially examines the entire area (all pixels) of the input screen to determine the presence or absence of a danger marker 5, referring to Figure 7. First, the danger marker detection unit 52 extracts one pixel from the screen in S101, and in S102 determines whether the color of the extracted pixel is similar to the color of the background 32 of the danger marker 5. If they are not similar, the process moves to S107, and the process from S101 is repeated until processing is completed for all pixels on the screen (S107, Yes) (S107, No). If the extracted pixels have similar colors (S102, Yes), the danger marker detection unit 52 investigates the pixels located around the pixels extracted in S101 and extracts multiple pixels (extraction region 202 in Figure 8) whose background 32 of the danger marker 5 has similar colors (S103).

[0042] Figure 8 is an explanatory diagram showing an example of a method for detecting the danger marker 201. First, the danger marker detection unit 52 extracts pixels whose background 32 of the danger marker 5 has a similar color (extraction region 202). The danger marker detection unit 52 detects corner points (circular locations indicated in the extraction region 202) in relation to the extraction region 202, and detects the extraction region contour 203, which is composed of the four sides of the danger marker 5, from the four corner points and the locations where the corner points are aligned in a straight line. In addition to the above explanation, a method of detecting straight lines from the contour can also be applied to extract the quadrilateral.

[0043] Returning to Figure 7, in S104, the danger marker detection unit 52 determines whether the outer contour of the region composed of the extracted pixels is in the shape of a quadrilateral. Danger markers 5 often appear as a parallelogram or trapezoid on the screen. If the extracted region contour is a quadrilateral (S104, Yes), the process moves to S105; otherwise, it moves to S107 (S104, No). The danger marker detection unit 52 detects the direction indicator pattern 31 inside the quadrilateral based on the danger marker position information 61 on the screen (coordinates of the four points of the quadrilateral detected in S104) (S105). The detection in S105 uses the difference between the brightness of the background area 32 and the brightness of the direction indicator pattern area inside the quadrilateral, for example, as shown in area 204 of Figure 8, and then extracts the corner points (for example, the circled areas in the extraction area 202) to detect the direction indicator pattern 31 and determine whether the quadrilateral in question is a danger marker 5. The danger marker detection unit 52 sets the quadrilateral as a danger marker 5 when it extracts one or more direction indicator patterns 31 from inside the quadrilateral (S106). By extracting multiple direction indicator patterns 31, it is possible to prevent false detection of danger markers due to patterns that coincidentally resemble the direction indicator patterns 31.

[0044] Figure 9 is a flowchart showing the processing of the danger line detection unit 53. The danger line detection unit 53 detects the direction indicator pattern 31 within the danger marker 5, similar to S105 (S111).

[0045] The danger line detection unit 53 examines the degree of overlap between the positional information of each side of the detected direction indicator pattern 31 and the four sides of the danger marker position information 61 (S112). The degree of overlap indicates the degree to which two sides overlap; the degree of overlap increases when both the difference in the slope and the difference in the intercept of the straight line approximating the two sides are small. The danger line detection unit 53 sets the side with the highest number of overlapping occurrences among the four sides of the danger marker position information 61 as danger line 6 (danger line information 62) (S113). For example, if three triangular regions are detected, the danger line detection unit 53 compares the degree of overlap between the sides of each triangle and the four sides of the danger marker 5, and sets the side of the danger marker 5 with the most overlap with the sides of the triangle (the rightmost side in the extracted region contour 203 of Figure 8) as danger line 6.

[0046] Figure 10 is a flowchart showing the processing of the danger direction detection unit 54. The danger direction detection unit 54 detects the direction indicator pattern 31 within the danger marker 5, similar to S105 or S111 (S121). In other words, the process in S105 may be common to S105 or S111. The danger direction detection unit 54 detects the direction of the danger direction 7 from the multiple direction indicator patterns 31 that it has detected (S122).

[0047] Figure 11 is an explanatory diagram of the danger marker 5, which shows the details of the processing in S122. The danger marker 5 on the screen, captured by camera 51, shows three directional indicator patterns 31, and the direction (vector) of each danger direction 7 can be determined as follows. The first directional pattern 31a is triangle pqr, where vector 221 passes from vertex p, which is tangent to line segment AB, through the midpoint s of side qr, which is tangent to line segment CD. Vector 221 may also be the normalized form of vector ps. The vector 222 is obtained from the second direction indicator pattern 31 in the same way as the first direction indicator pattern 31a. The vector 223 is obtained from the third direction indicator pattern 31 in the same way as the first direction indicator pattern 31a. • Vector 231 extending line segment AC Vector 232 extending line segment BD Note that these vectors 221-223, 231, and 232 are normalized vectors.

[0048] Returning to Figure 10, the danger direction detection unit 54 takes the average of the multiple vectors described in Figure 11 and determines the danger direction 7 (danger direction information 63) as a result (S123). Note that the danger direction detection unit 54 may use only vectors 221-223 extracted from the direction indicator pattern 31 for the multiple vectors calculated in S123, or it may add vectors 231 and 232 in addition to vectors 221-223 and average them. By adding vectors 231 and 232 to the S123 calculation, detection errors and noise are reduced, and a more accurate vector indicating the dangerous direction 7 can be calculated.

[0049] Figure 12 is a flowchart showing the processing of the hazardous area determination unit 55. The hazard area determination unit 55 calculates a straight line (straight line EF in Figure 5) by moving the hazard line 6 (segment CD in Figure 5) detected by the hazard line detection unit 53 a predetermined distance in the hazard direction 7 determined by the hazard direction detection unit 54 (S131). The danger area determination unit 55 determines that the quadrilateral CDFE generated by the line segment EF calculated in S131 and the danger line 6 (line segment CD) is the danger area 8 on the screen (danger area information 64) (S132).

[0050] Figure 13 is a flowchart showing the processing of the hazard determination unit 56. First, the hazard determination unit 56 determines that the worker's position is safe (S146) when the safety signal 65 shown in Figure 6 is input (S141, Yes). On the other hand, when no safety signal 65 is input (S141, No), the danger determination unit 56 selects the point closest to the shooting point (shortest point) among the four points of the rectangle of the danger area 8 (S142).

[0051] Figure 14 is an explanatory diagram of the shooting screen used to explain the process of S142 in Figure 13. The captured screen 230 shows the quadrilateral ABCD of the danger marker 5 explained in Figure 11, and also shows the quadrilateral CDFE of the danger area 8 formed from the danger marker 5. The danger determination unit 56 selects, for example, the center of the bottom of the captured screen 230 as the shooting point 241, and selects the vertex of the danger area 8 (in this case, vertex C) that has the shortest distance 242 from the shooting point 241 as the shortest point (S142). Note that shooting point 241 is the point closest to the feet of the worker 22 who is wearing the camera 51 and taking the picture. When using a camera 51 with a narrow field of view, the distance between shooting point 241 and the worker's feet becomes larger, so adding a fixed distance to the distance between shooting point 241 and the shortest point reduces the error in the distance to the outermost point. In the following explanation, we will use an example where the distance between shooting point 241 on the screen and the actual feet of the worker 22 is sufficiently small, or where the error has been corrected.

[0052] Furthermore, if the camera 51 is installed on the worker 22's helmet, the danger marker 5 may temporarily move out of the camera 51's field of view due to the worker 22 turning their head. Therefore, the danger determination unit 56 may consider not only the current image from the camera 51 but also past images from the camera 51 to determine the worker 22's current position and the direction of the danger marker 5 visible from that current position. Therefore, for example, the hazard determination unit 56 tracks the direction of the worker's face from images taken by the camera 51 from the past to the present, and from an acceleration sensor attached to the worker's helmet. If there is a change in the direction of the worker's face (such as turning to the side) at the point when the hazard marker 5 is no longer visible in the image from the camera 51, the hazard determination unit 56 may continue the warning issued at the time the hazard marker 5 was visible in the image from the camera 51, assuming that the worker simply turned around without moving.

[0053] Thus, the danger determination unit 56 acquires the camera 51's turning information, and if the camera 51's turning information indicates that it turned during a period when the danger line 6 is not detected in the camera 51's captured images in the time-series change of the camera 51's captured images, it may determine that the camera 51's shooting position did not change during the period when it was not detected.

[0054] Returning to Figure 13, the hazard determination unit 56 estimates the distance between the shooting point 241 and the shortest point in S142 (S143). The distance estimation process in S143 is achieved by first fixing the height of the camera 51, measuring its characteristics, and setting the distance for each pixel position in the screen. The height of the camera 51 depends on the height of the worker 22 and where the camera 51 is mounted (for example, on a helmet or in a chest pocket). Therefore, by setting the height of the worker 22 and the mounting location of the camera 51 in advance, the estimated distance to the shortest point can be corrected.

[0055] When correcting the distance to the shortest point in S143, if the worker is tall, a wider area will be captured, so the distance corresponding to the pixel at the same position will be longer. Also, since the height of the camera 51 changes when the worker 22 is leaning forward or crouching, it is also possible to estimate the worker 22's posture using an acceleration sensor or the like and then correct the distance to the shortest point. However, since falls in dangerous positions are most likely to occur when the worker 22 moves, that is, when they are walking upright, it is thought that sufficient effect can be obtained from the standpoint of accident prevention even without correcting for the worker 22's posture. The above describes a method for pre-setting the distance to the shortest point based on the characteristics of camera 51.

[0056] Another method for determining the distance to the shortest point in S143 is to estimate the distance to the shortest point from the width of the detected danger marker 5 and the pitch of the direction indicator pattern 31 on the screen. That is, the number of pixels on the screen corresponding to the width of the detected danger marker 5 and the pitch of the direction indicator pattern 31 is obtained, and the location to be measured is estimated by the ratio of these pixel counts. This method has the effect of eliminating the need for prior distance setting or correction for height, etc., but it has the characteristic that the error increases as the distance from the reference danger marker 5 position increases. The hazard determination unit 56 determines whether the distance to the shortest point obtained in S143 is less than or equal to a predetermined value (i.e., whether the distance to the shortest point is close to the worker 22) (S144). If it is less than or equal to the predetermined value (S144, Yes), the hazard determination unit 56 determines that the position of the worker 22 is dangerous because it is close to the hazard area 8 (S145). If it is not less than or equal to the predetermined value (S144, No), the hazard determination unit 56 determines that the position of the worker 22 is safe (S146).

[0057] Figure 15 is a plan view of the danger marker 5 in Figure 11. The area shaded in the diagram is the warning start area 235 where the worker 22's position is determined to be dangerous in S145. The warning start area 235 is formed in the region less than or equal to the distance 234, which is the predetermined value in S144, from the quadrilateral CDFE of the dangerous area 8. In Figure 15, for illustrative purposes, the distance 234 is drawn to be about twice the width of the danger marker 5, but in reality, it is desirable to set the distance 234 to a distance (for example, about 2m) at which worker 22 can safely stop after being warned when they enter the dangerous area 8.

[0058] Figure 16 shows a modified example of the position determination process for worker 22 in S144. The warning start area 235 in Figure 15 extends to the area in front of the danger line 6 (side CD) of the danger marker 5 (towards the elevated walkway 1 in Figure 1), so it was possible to appropriately determine the danger (S145) and issue a warning before the worker 22 crossed the opening 3 (danger line 6) in Figure 1. However, when worker 22 is moving at high speed, as shown in Figure 16 (position 22a → position 22b → position 22c), issuing a warning only after entering the warning start area 235 may result in a delayed warning. Therefore, the danger determination unit 56 estimates the direction of movement 22d of worker 22 from the change in worker 22's position.

[0059] Furthermore, the hazard determination unit 56, taking into account the worker's latest position 22c and the worker's direction of movement 22d, may issue the warning S145 before the worker enters the warning area 235 if it anticipates that the worker will enter the warning area 235 in the future. This further enhances safety through earlier warnings. Similarly, the danger determination unit 56 may detect the direction of movement of the camera 51 from the image captured by the camera 51, and if the direction of movement of the camera 51 is toward the danger area 8, it may control the warning issuing unit 57 to issue a stronger warning than when the camera 51 is in a dangerous position. [Examples]

[0060] The following describes Example 2 of the present invention. Example 1 focused on a single hazard marker 5. Example 2 describes a method for forming a hazard area 8 from multiple hazard markers 5 that are close to each other. Figure 17 is a plan view showing a work site at a height where two passages 301 and 302 are in close proximity. Passage 301 is in a fixed position, and passage 302 is a ladder (boarding bridge) that is temporarily connected to passage 301 in a T-shape (actually close with some gap remaining) in order to enter and exit passage 301.

[0061] When the danger lines 6 of multiple danger markers 5 (first danger marker 303, second danger marker 304) are close together, the multiple danger areas 8 formed from these danger lines 6 may overlap. In this case, the danger area determination unit 55 may use the area of ​​the union of the multiple danger areas 8 for the overlapping parts between the multiple danger areas 8, or it may cancel out a part of the area of ​​that union to narrow the danger areas 8. The details of the method for narrowing the danger areas 8 will be explained below.

[0062] In Figure 17, the hazard area 8 corresponding to hazard marker 303 is a quadrilateral CDFE, but another hazard marker 304 of passage 302 exists within that area. In this case, the hazard area determination unit 55 excludes the portion of the hazard marker 304 below the hazard line PQ in the figure (the portion beyond the opposing hazard marker 303) from the hazard area 8. Therefore, the hazard area 8 formed by hazard marker 303 becomes the region enclosed by the vertex CERPQD shown by the shaded area. Similarly, for the hazard area 8 formed by hazard marker 304, the portion beyond the opposing hazard marker 303 is excluded from the hazard area 8.

[0063] In this way, the danger marker detection unit 52 detects the first danger marker 5 and the second danger marker 5 from the same captured image. Then, if the second danger marker 5 is included within the first danger area 8 determined from the first danger marker 5, the danger line 6 of the second danger marker 5 is moved by a third distance (the distance of side PR) in the danger direction 7 of the first danger marker 5, and the area enclosed by the danger line 6 before the move (side PQ) and the danger line 6 after the move (side RF) is excluded from the first danger area 8. This prevents inappropriate warnings from being issued regarding the movement of worker 22 across the two passages 301 and 302.

[0064] Figure 18 is a flowchart that corresponds to Example 2 of the flowchart in Figure 12, with the processing of the hazardous area determination unit 55 added. In this flowchart, the following two hazard markers 5 are illustrated. From the first danger marker 5 (for example, danger marker 303 in Figure 17), the first danger line, the first danger direction, and the first danger area (quadrilateral CDFE in Figure 17) are detected. From the second danger marker 5 (for example, danger marker 304 in Figure 17), the second danger line, the second danger direction, and the second danger area are detected.

[0065] The hazard area determination unit 55 then performs the processes S151 to S153 for each hazard marker 5. The hazard area determination unit 55 sets a first hazard area corresponding to the first hazard line, similar to the first embodiment (S151). The hazard area determination unit 55 searches whether a second hazard line exists within the first hazard area of ​​S151 (S152). If a second hazard line does not exist (S152, No), the process ends, and the original first hazard area is set as hazard area 8.

[0066] On the other hand, if a second danger line exists (S152, Yes), the danger area determination unit 55 checks whether the first danger direction and the second danger direction are in opposite directions (S153). The determination of whether they are in opposite directions can be made by checking whether the result of the dot product of the vectors of the two danger directions is negative. If the two hazard directions are opposite (S153, Yes), the hazard area determination unit 55 excludes the area beyond the second hazard line (on the side of the first hazard direction and opposite to the first hazard line) from the first hazard area. As a result, the area enclosed by point CERPQD in Figure 17 becomes the hazard area 8.

[0067] If the first danger marker 5 and the second danger marker 5 are swapped in the above explanation, the first danger marker 5 becomes danger marker 304 in Figure 17, and the second danger marker 5 becomes danger marker 303 in Figure 17. In that case, the danger area determination unit 55 excludes the area beyond the second danger line (line segment CD) (the upper part in Figure 17) from the first danger area formed from the first danger line (line segment PQ). As a result, the area enclosed by quadrilateral SDQP in Figure 17 (more precisely, an area narrower than quadrilateral SDQP by the width of the safety fence 306) becomes the danger area 8. Then, the hazard area determination unit 55 merges (unions) the first hazard area after the exclusion process in S153 (the area enclosed by point CERPQD) and the second hazard area after the exclusion process in S153 (quadrilateral SDQP), thereby forming one hazard area 8 from the two hazard markers 5.

[0068] Furthermore, if the distance between the connection (proximity) point between the passage 302 and the passage 301 is narrow, that is, if the dangerous area 8 has an elongated shape and the maximum value of its width is less than a predetermined value, the dangerous area determination unit 55 may remove the connection (proximity) point from the dangerous area 8, assuming that there is no risk of falling from that point. For example, if the lengths of line segment SP and line segment DQ in Figure 17 are both smaller than a predetermined value (e.g., a few centimeters), it is determined that there is no risk of falling from the area of ​​quadrilateral SDQP, and the portion of quadrilateral SDQP is removed from danger area 8, leaving only quadrilateral CERS as danger area 8. This means that even if there is a small gap in the boarding bridge, the likelihood of a person falling through that gap is low, and excessive warnings will not be issued to passengers on board.

[0069] Figure 19 is an explanatory diagram showing an example where a rectangular opening 3 (hole) 311 is located in the center of area 310. Four hazard markers 311, 313, 314, and 315 are installed around the opening 3. However, in the area 312 where marker installation is not possible, located to the lower right of the opening 3, there is a lid that can be opened and closed, and an operating tool for it, so there is a place where a hazard marker 5 cannot be installed. For this reason, hazard marker 315 is shorter in length than the hazard marker 313 facing it.

[0070] Figure 20 is an explanatory diagram showing the hazardous area 8 detected from Figure 19. The hazardous area 8 corresponding to each hazard marker 5 is as follows: • The first hazard area, "Square GKLT," is detected from the first hazard marker, 313. • Second danger area "Rectangle JKNP" detected from the second danger marker 314. • Third danger area "Square SZYR" detected from third danger marker 315 • Fourth hazard area "Square GHQR" detected from the fourth hazard marker 316.

[0071] The hazard area determination unit 55 executes the process shown in Figure 18 to define the area remaining after removing the inner central quadrilateral UVWX from the outer quadrilateral GKNR as the hazard area 8 extracted from the four hazard markers 5. Even though hazard markers 5 cannot be placed in the marker-unplaceable area 312, the edges of the opening 311 (the parts touching line segments GK, KN, NR, and RG, respectively) all become hazard area 8, thus ensuring safety. Therefore, the hazard area determination unit 55 sets the width of the hazard area 8 (the distance between the hazard line 6 and the side parallel to it) to be wider than the width of the area 312 where markers cannot be installed. If the width of the area 312 where markers cannot be installed is wider than the width of the hazard area 8, the hazard area determination unit 55 sets the area adjacent to the opening 3 as the hazard area 8 by installing a hazard marker 5 so as to surround both the area where markers cannot be installed and the opening 3. [Examples]

[0072] The following describes Embodiment 3 of the present invention. In Embodiments 1 and 2, the hazard area 8 was formed based on the positional relationship of the hazard markers 5, but in Embodiment 3, a more detailed hazard area 8 is formed by using the safety marker 9 shown in Figure 22 in addition to the hazard marker 5.

[0073] Figure 21 is a diagram showing the configuration of the hazardous area notification device 50 in Example 3. The hazardous area notification device 50 in Figure 21 is further modified from the hazardous area notification device 50 in Figure 6 by adding a safety marker detection unit 71. The safety marker detection unit 71 detects a safety marker 9 that is different from the danger marker 5, and detects a safety line (for example, the identification line 92 in Figure 23) from the detected safety marker 9. The method for detecting the safety marker 9 is the same as that of the hazard marker detection unit 52, using image processing based on the color or mark painted on the marker (see Figure 23 for details). The safety marker detection unit 71 then outputs the safety marker information 72, which is the detection result of the safety marker 9 (coordinate information of the safety marker 9 in the image), to the hazard area determination unit 55.

[0074] Figure 22 is a plan view of the work site when using safety marker 9. In this work site, opening 3 is located in the upper right of the area, and a hazard marker 321 is installed at that location. However, there is also another safety fence 322 between it and worker 22. Worker 22 approaching safety fence 322 is physically restricted from moving due to the safety fence 322, even though the hazard marker 321 is nearby, and there is no risk of falling outside the passageway. In this case, it is preferable not to issue a warning even if worker 22 enters the hazard area 8 near hazard marker 321.

[0075] Therefore, in Embodiment 3, by installing a safety marker 9 along the worker 22 side (pathway side) of the safety fence 322, the hazard area determination unit 55 excludes the area near the detected safety marker 9 from the hazard area 8. As a result, the hazard determination unit 56 determines that the worker 22 approaching the safety fence 322 is safe, preventing false alarms (where a notification of danger is issued even though it is safe). Since false alarms are a fail-safe mechanism, they do not hinder accident prevention. However, if false alarms become frequent, workers 22 tend to downplay or ignore warnings of danger, indirectly reducing the effectiveness of accident prevention.

[0076] Figure 23 is an explanatory diagram showing an example of a safety marker 9. The safety marker 9 has a strip-like shape, similar to the hazard marker 5 in Figure 3. The pattern inside the safety marker 9 must be clearly different from that of the hazard marker 5. For example, the safety marker 9 in Figure 23 has a white background 91 with two green identification lines 92 placed within it. Unlike the hazard marker 5, the safety marker 9 does not have a concept of direction (hazard direction 7), but it does have the concept of a safety line corresponding to the hazard line 6. The safety line is a line along the length of the safety marker 9, for example, the line of the identification line 92.

[0077] Figure 24 is a plan view showing the hazardous area 8 formed by the combined use of hazard marker 5 and safety marker 9. Areas 331-333 represent the same area over time, with time progressing in the order of Area 331 → Area 332 → Area 333. In each area 331-333, one safety marker 9 is positioned vertically on either side of the hazard marker 5 in the center of the drawing. In other words, the two safety markers 9 and the one hazard marker 5 are parallel to each other. The total of three markers divide the area into four sections as follows. • Area 1 335 is formed outside the upper safety marker 9 and is outside the hazardous area 8, so no warning is issued even if worker 22 is present. • The second area 336 is formed between the upper safety marker 9 and the danger marker 5, and is outside the danger area 8, so basically no warning is issued even if workers 22 are present. However, as shown in Figure 15, a warning start area 235 (Figure 15, not shown in Figure 24) is formed around the danger area 8, so a warning is issued to workers 22 within the warning start area 235.

[0078] • The third area 337 is formed between the lower safety marker 9 and the hazard marker 5, and is essentially within the hazard area 8, so a warning is issued if worker 22 is present. Furthermore, a warning is also issued to worker 22 within the warning start area 235 surrounding the hazard area 8. • The fourth area 338 is formed outside the upper safety marker 9. Both the fourth area 338 and the third area 337 are located on the side of the danger direction 7 from the danger marker 5, but the fourth area 338 shields the danger area 8 and serves to exclude the lower direction of the diagram from the danger area 8 and its warning start area 235. Therefore, even if a worker 22 is present in the fourth area 338, no warning will be issued.

[0079] Figure 25 is a flowchart showing the processing of the hazardous area determination unit 55 when using the safety marker 9. In contrast to the flowchart in Figure 12 which does not use safety marker 9, in Figure 25, the hazardous area 8 set in S132 based only on the hazardous area 8 is designated as the provisional hazardous area 8 (S132b). Then, in Figure 25, by adding the following process, the provisional hazardous area 8 is modified using safety marker 9 to determine the final hazardous area 8. If a safety marker 9 exists in the area in front of the hazardous area 8 (second area 336 in Figure 24) (S133, Yes), the portion of the hazardous area 8 hidden by the safety marker 9 is erased (S135, see Figure 26 for details). (S133, Yes) is the case, for example, when a safety marker 9 exists between the hazard line CD in Figure 14 and the shooting point 241. If a safety marker 9 exists within hazardous area 8 (area 337 in Figure 24) (S134, Yes), the portion of hazardous area 8 beyond safety marker 9 is erased (S136, see Figure 27 for details).

[0080] Figure 26 is an image captured by camera 51 showing the first positional relationship between danger marker 5 and safety marker 9. The hazard area determination unit 55 checks whether a safety marker 9 exists on a straight line connecting the shooting point 401 in the captured image diagram to each point on the hazard line CD. The first straight line 402 is an extension of the line from the shooting point 401, passing through point C, and intersects with the safety marker 9 at point U. When this intersection occurs, it is determined that point C is hidden by point U of the safety marker 9. The second straight line 403 is an extension from the shooting point 401 that passes through point X on the danger line CD, and intersects with the safety marker 9 at point V. When they intersect in this way, it is determined that point X is hidden by point V of the safety marker 9. In other words, the section of line segment CX of the danger marker 5 is hidden by the section of line segment UV of the safety marker 9. The third straight line 404 is an extension of the line from the shooting point 401, passing through point D, and does not intersect with safety marker 9. In other words, the effective (unobstructed) section of danger marker 5 is the section of line segment XD.

[0081] The hazard area determination unit 55 removes the portion of quadrilateral EYXC (the section of line segment CX) that is hidden by the safety marker 9 from the provisional hazard area 8 composed of quadrilateral EFDC (S135), and makes the remaining quadrilateral YFDX (the section of line segment XD) the corrected (final setting) hazard area 8. Thus, if the hazard area determination unit 55 determines that the safety marker 9 is not included within the hazard area 8 determined from the hazard marker 5, and the safety marker 9 is located between the image position corresponding to the camera 51's shooting position and the image position of the hazard line 6, the hazard line 6 that is shielded by the safety line of the safety marker 9 from the image position corresponding to the camera 51's shooting position is excluded from the determination of the hazard area 8. In other words, as shown in Figure 26, if the danger area determination unit 55 sees a safety marker 9 closer than the danger marker 5, it deletes the danger area 8, thereby preventing the warning issuing unit 57 from issuing a warning.

[0082] Figure 27 is an image captured by camera 51 showing the second positional relationship between danger marker 5 and safety marker 9. If a safety marker 9 is present within the hazardous area 8 (quadrilateral EFDC), the hazardous area determination unit 55 erases the portion of the hazardous area 8 beyond the safety marker 9 (on the hazardous direction 7 side) (quadrilateral EZYX) (S136). As a result, the hazardous area determination unit 55 designates the area enclosed by the remaining points XYZFDC as the modified hazardous area 8. In this way, if the hazard area determination unit 55 determines that a safety marker 9 is included within the hazard area 8 determined from the hazard marker 5, it moves the safety line of the safety marker 9 in the hazard direction 7 by a second distance (the distance of side XE), and removes the area enclosed by the safety line before the move (side XY) and the safety line after the move (side EZ) from the hazard area 8. In other words, as shown in Figure 27, if the safety marker 9 is visible at a distance greater than the danger marker 5, the danger area determination unit 55 deletes the danger area 8 that is further away from the safety marker 9.

[0083] Figure 28 is an image captured by camera 51 showing the third positional relationship between danger marker 5 and safety marker 9. Figures 27 and 28 share the common feature of having a safety marker 9 within the hazardous area 8. In Figure 28, the distance between the hazardous marker 5 and the safety marker 9 is even closer than in Figure 27. In this case, the portion of the hazardous area 8 beyond the safety marker 9 (rectangle EFYX) is completely eliminated, so the modified hazardous area 8 becomes a quadrilateral XYDC. Furthermore, because the lengths of line segment XC and line segment YD are short enough for a person to step over (e.g., 10 cm or less), the danger area determination unit 55 determines that quadrilateral XYDC is no longer a danger area 8 if there is no risk of falling, and that there is no corrected danger area 8.

[0084] Figure 29 is a perspective view showing the location where the movable safety fence 411 is installed in the opening 3. Figure 30 is a plan view showing the location shown in Figure 29. The opening 3 switches between a closed state, where it is shielded by overlapping with the movable safety fence 411 as shown in Figure 29, and an open state, where the movable safety fence 411 does not overlap with the space 412 of the opening 3, as shown in Figure 30. The safety marker 9 is installed at the bottom of the movable safety fence 411. Alternatively, instead of the movable safety fence 411, a door that opens and closes or a shutter that opens and closes vertically may be used, with the safety marker 9 installed at the bottom of the door or shutter.

[0085] In the open state, the safety marker 9 does not shield the hazard area 8 formed by the hazard marker 5, so a warning is issued to workers 22 near the opening 3. On the other hand, in the closed state, the safety marker 9 overlaps directly above the hazard area 8, thus eliminating the hazard area 8. As a result, the movable safety fence 411 makes it physically impossible to fall from the opening 3 to the lower area 21, thus eliminating the need to issue extra warnings to workers 22 near the opening 3. In other words, it is possible to issue appropriate warnings in accordance with changes in the environment.

[0086] Figure 31 is an image taken when a single hazard marker was separated into two. Since the camera 51 is attached to the worker, parts of the worker's hands or body may be captured on camera. For example, if worker 22's hand 502 is captured on screen and obscures part of the danger marker 5, it may be separated and detected as two separate danger markers 503 and 504. Therefore, the danger marker detection unit 52 may detect that the two danger markers 503 and 504 are located on their respective extensions and determine that the two danger markers 503 and 504 constitute a single danger marker 5, that is, in the figure, quadrilateral ABDC is the danger marker 5 and line segment CD is the danger line 6.

[0087] Figure 32 is an image taken when a single hazard marker was separated into three parts. When worker 22's hands 512 and 513 are visible and obscure the danger line 6, the danger line 6 is divided into three danger markers 514, 515, and 516. Danger marker 516 has a small area due to its position, and the direction indicator pattern 31 may not be detectable. Therefore, the danger marker detection unit 52 detects a danger marker 516 that is located on the extensions of two danger markers 514 and 515, and that is similar in color and partial shape to danger marker 5. In this case, the danger marker detection unit 52 may determine that the three danger markers 514, 515, and 516 are a single danger marker 5. Thus, if the danger marker detection unit 52 detects multiple danger markers 5 whose respective danger lines 6 lie on the same extension, it may consider these multiple danger markers 5 as a single danger marker 5.

[0088] Figure 33 is a hardware configuration diagram of the hazardous area notification device 50. The hazardous area notification device 50 is configured as a computer 900 having a CPU 901, RAM 902, ROM 903, storage 904, communication I / F 905, input / output I / F 906, and media I / F 907. Storage 904 is a large-capacity flash memory or HDD, connected internally or externally to the computer 900. Communication I / F 905 is connected to an external communication device 915. Input / Output I / F 906 is connected to an input / output device 916. Media I / F 907 reads and writes data to the recording medium 917. Furthermore, CPU 901 improves and controls each processing unit by executing a program (also called an application or app) loaded onto ROM 903 or RAM 902. This program can also be distributed via a communication line or by recording it on a recording medium 917 such as a recordable CD or DVD and distributing it.

[0089] The hazardous area notification device 50 of Examples 1 to 3 described above has a hazard determination unit 56 that detects hazard markers 5 attached to the floor or other surface to indicate a hazardous area 8 using a camera 51 worn by a worker 22, and determines whether the location where the worker 22 is located is safe or hazardous. By including information that allows for the identification of a hazard line 6 and a hazardous direction 7 by image recognition in the hazard marker 5, hazardous areas can be detected with high accuracy with a small amount of processing power. Furthermore, by including information that allows for visual identification of the danger line 6 and danger direction 7 in the danger marker 5, workers 22 can visually identify the danger area 8 even if they are not wearing the camera 51 or if the danger area notification device 50 is not functioning, such as when the camera 51's battery is dead.

[0090] Furthermore, the technology described in Patent Document 2 designates areas where markers scattered in various locations are connected by virtual lines as no-entry zones. However, various interpretations can arise regarding how to connect densely scattered coordinates, which can sometimes lead to misinterpretation of the intended virtual lines. On the other hand, in this embodiment, the danger area notification device 50 has danger markers 5 that contain line information capable of identifying danger lines 6, thus avoiding the problem of ambiguous interpretation of virtual lines as seen in Patent Document 2. Furthermore, the technology described in Patent Document 2 includes absolute position information (latitude, longitude, altitude) obtained by GPS (Global Positioning System) in the marker. On the other hand, since the danger marker 5 in this embodiment indicates the surrounding danger area 8, the absolute position of the danger marker 5 does not need to be included in the danger marker 5. This makes it possible to form a danger area 8 even in places where GPS signals are difficult to obtain, such as indoor factories and plants.

[0091] As an alternative, it is conceivable to use high-performance cameras such as TOF (Time of Flight) cameras or stereo cameras to detect depth information such as steps and openings. However, the price and detection accuracy of such high-performance cameras are challenges. Furthermore, while the cameras worn by workers need to be battery-powered, the power consumption of the camera itself and the processing required to acquire depth information increases, making it difficult to ensure a practical operating time. On the other hand, even if the camera 51 cannot acquire depth information, the hazardous area notification device 50 of this embodiment can mechanically identify the hazardous area 8 using the information from the hazard marker 5 as a clue.

[0092] Furthermore, the present invention is not limited to the embodiments described above, and it goes without saying that various other applications and modifications can be made as long as they do not depart from the gist of the invention as described in the claims. For example, the embodiments described above describe the configuration of the hazardous area notification device 50 in detail and specifically in order to explain the present invention in an easy-to-understand manner, and are not necessarily limited to those comprising all the components described. Also, it is possible to replace a part of the configuration of one embodiment with a component of another embodiment. It is also possible to add a component of another embodiment to the configuration of one embodiment. Furthermore, it is possible to add, replace, or delete other components for a part of the configuration of each embodiment.

[0093] Furthermore, some or all of the above configurations, functions, and processing units may be implemented in hardware, for example, by designing them as integrated circuits. Broadly defined processor devices such as FPGAs (Field Programmable Gate Arrays) and ASICs (Application Specific Integrated Circuits) may be used as hardware. Furthermore, each component of the hazardous area notification device 50 according to the above-described embodiment may be implemented on any hardware, as long as the respective hardware can send and receive information from each other via a network. Also, the processing performed by a certain processing unit may be implemented by a single piece of hardware, or by distributed processing by multiple pieces of hardware. [Explanation of Symbols]

[0094] 1. Elevated walkway 2 Safety fence 3 Opening 4. Safety fence locations 5. Danger Marker 6. Danger Line 7. Dangerous direction 8. Dangerous Areas 9. Safety Marker 22 workers 31 Directional Indicator Patterns 32 Background section 50. Dangerous Area Notification Device (Warning Device) 51 Camera 52 Hazard Marker Detection Unit 53 Danger line detection unit 54 Danger Direction Detection Unit 55 Dangerous Area Determination Section 56 Hazard Assessment Department 57 Warning Issuance Unit 71 Safety Marker Detection Unit 72 Safety Marker Information 235 Warning Start Area

Claims

1. A danger marker detection unit detects danger markers, which have a danger line in the length direction and a mark indicating the direction of danger in the width direction, from the image captured by the camera. Based on the danger line and danger direction indicated by the danger marker detected by the danger marker detection unit, the danger line is moved by a first distance in the danger direction, and the area enclosed by the danger line before and after the movement is determined as a danger area by the danger area determination unit. A hazard determination unit determines whether the camera's shooting position is hazardous or not based on the positional relationship between the camera's shooting position, which captured the image of the hazard marker, and the hazardous area. The system is characterized by having a warning issuing unit that issues a warning when the danger determination unit determines that there is danger. Warning device.

2. The aforementioned danger marker is configured as a rectangle containing multiple marks arranged on the same line along the direction of the danger line. Each mark contained within the aforementioned danger marker is configured as an asymmetrical figure indicating the direction of danger with respect to a direction perpendicular to the direction of the danger line. The hazard area determination unit determines the hazard line at the detection position of the hazard marker as the first side, and the second side by moving the first side by a predetermined distance toward the hazard direction, and determines the rectangle formed by connecting the first side and the second side as constituent elements as the hazard area. The danger determination unit is characterized in that it determines that the camera's shooting position is dangerous when the shooting point in the captured image indicating the camera's shooting position is included in the warning start area which includes the danger area. The warning device according to claim 1.

3. The warning device further includes a safety marker detection unit that detects safety markers different from the danger markers from the captured images of the danger markers, and detects safety lines from the detected safety markers. The hazard area determination unit is characterized in that, if the safety marker is included within the hazard area determined from the hazard marker, it moves the safety line of the safety marker by a second distance in the hazard direction, and excludes the area between the safety line before and after the movement from the hazard area. The warning device according to claim 1 or claim 2.

4. The warning device further includes a safety marker detection unit that detects safety markers different from the danger markers from the captured images of the danger markers, and detects safety lines from the detected safety markers. The hazard area determination unit is characterized in that, if the safety marker is not included within the hazard area determined from the hazard marker, and the safety marker is located between the image position corresponding to the camera's shooting position and the image position of the hazard line, the hazard line that is obscured by the safety line of the safety marker from the image position corresponding to the camera's shooting position is excluded from the determination of the hazard area. The warning device according to claim 1 or claim 2.

5. The aforementioned danger marker detection unit detects a first danger marker and a second danger marker from the same captured image. The hazard area determination unit, when the second hazard marker is included within the first hazard area determined from the first hazard marker, moves the hazard line of the second hazard marker by a third distance in the hazard direction of the first hazard marker, and excludes the area between the hazard line before and after the movement from the first hazard area. The warning device according to claim 1 or claim 2.

6. The danger determination unit detects the direction of movement of the camera from the image captured by the camera, and controls the warning issuing unit to issue a stronger warning than when the camera's shooting position is dangerous, if the direction of movement of the camera is toward the danger area. The warning device according to claim 1 or claim 2.

7. The danger determination unit acquires the camera's turning information, and if the camera's turning information indicates that the camera turned during a period when the danger line is not detected in the camera's captured images in the time-series changes of the camera's captured images, the unit determines that the camera's shooting position did not change during that undetected period. The warning device according to claim 1 or claim 2.

8. The danger marker detection unit is characterized in that, when it detects multiple danger markers whose respective danger lines lie on the same extension line, it considers those multiple danger markers as a single danger marker. The warning device according to claim 1 or claim 2.

9. A danger marker detection unit that detects danger markers from images captured by the camera, A safety marker detection unit detects a safety marker from the captured image of the aforementioned danger marker, A hazard area determination unit determines the area indicated by the hazard marker detected by the hazard marker detection unit as a hazard area, A hazard determination unit determines whether the camera's shooting position is hazardous or not based on the positional relationship between the camera's shooting position, which captured the image of the hazard marker, and the hazardous area. The system includes a warning issuing unit that issues a warning when the aforementioned danger determination unit determines that there is danger. The hazard area determination unit is characterized in that, if the safety marker is visible closer than the hazard marker, it deletes the hazard area, thereby preventing the warning issuing unit from issuing a warning. Warning device.

10. A danger marker detection unit that detects danger markers from images captured by the camera, A safety marker detection unit detects a safety marker from the captured image of the aforementioned danger marker, A hazard area determination unit determines the area indicated by the hazard marker detected by the hazard marker detection unit as a hazard area, A hazard determination unit determines whether the camera's shooting position is hazardous or not based on the positional relationship between the camera's shooting position, which captured the image of the hazard marker, and the hazardous area. The system includes a warning issuing unit that issues a warning when the aforementioned danger determination unit determines that there is danger. The hazard area determination unit is characterized in that, if the safety marker is visible at a distance greater than the hazard marker, the hazard area beyond the safety marker is deleted. Warning device.

Images