JP2550604B2 - Method of detecting monitored object of safety monitoring device - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION [Industrial field of application] The present invention relates to a safety monitoring device having a visual function added to an automatically operable facility such as an industrial robot or an NC machine tool. The present invention relates to a method of detecting a surveillance object.

[Conventional technology]

Conventionally, this type of safety monitoring device with a visual function sets a monitoring area so as to surround a dangerous area of an automatic machine, and installs a camera above the monitoring area to capture and analyze an image.

As a method of detecting the monitored object, if the monitored object can be detected as a silhouette image by transmitted light illumination, it can be detected regardless of the brightness characteristics of the monitored object. The illumination cannot be installed, and the background is black (or something close to black), and detection is performed by reflected light.

[Problems to be solved by the invention]

In this case, the monitored object is identified by the difference in brightness between the background and the monitored object, but if the background is black, for example, black safety shoes cannot be detected, or if white gloves are used. However, there is a problem that it becomes difficult to detect if it becomes black when working, and the problem was addressed by putting white (yellow) lines on safety shoes or frequently changing gloves.

It is an object of the present invention to provide a method capable of detecting a monitored object regardless of the brightness characteristics of the monitored object, overcoming the drawbacks of the conventional method.

[Means for solving problems]

Therefore, in the present invention, the surveillance area of one camera arranged above is divided into at least two, and the background is made of a conventional black type and a background of higher brightness, and each of the divided surveillance areas is divided into two areas. Set the window to, and on the black background side, set the pixels with brightness level slightly higher than the background level to white “1”, and on the background side with high brightness, the background brightness. Pixels at brightness levels that are slightly lower than the brightness level that are lower than the level are white "1"
It is a method of detecting a monitored object in a safety monitoring device, which binarizes as described above and counts the total number of white "1" pixels to detect the monitored object.

[Work]

By doing this, it is possible to detect, for example, conventional acquisition, helmets, etc. in the black background area, and it is possible to detect, for example, safety shoes, dirty gloves, etc., which were difficult to detect in the background area of high brightness, Therefore, it is possible to detect whatever brightness characteristic the monitored object has.

〔Example〕

Hereinafter, a method of detecting a monitored object of a safety monitoring device according to the present invention will be specifically described based on an embodiment shown in the drawings.

FIG. 1 shows that after the worker sets the welding part 9 on the jig base 8 and then moves back to a safe area and presses the welding start button (not shown), the spot welding robot 7 operates to perform welding. Start. At this time, the safety monitoring device added to the equipment operates.

The safety monitoring device is a device having an image processing function,
The circumference of the welding robot 7 is divided into four areas _6a , _6b , _6c , _6d ,
Cameras 1 to 4 are installed above each area,
There are four cameras in total.

In this embodiment, as shown in FIG.
It is divided into 61 and 62, and windows 1 and 63, so that they fit into each area
Window 2/64 is set.

The monitoring area 61 is a black background on the equipment side 65, and the monitoring area 62 is a green background on the non-equipment side 66. It is better for the latter to be bright, but it is easy to get dirty with shoes of workers, for example, so it is made green.

Then, for each monitoring area _{6 a, 6 b, 6 c} , 6 d, subjected to similar means.

Now, FIG. 3 is a histogram showing the number of pixels corresponding to the brightness.

FIG. 3a shows an example of a histogram of window 1.63. The binarization level 6a2 is set to be slightly higher than the background level 6a1, and the pixels having the brightness level of LEV1 · 6a2 or higher are set to white “1”. Then, except for safety shoes 6a3 etc.
5, work clothes 6a4 etc. will remain as images.

Also, the histogram of window 2.64 is shown in FIG. The binarization level 2b2 is set to a level LEV2 · 6b2 which is slightly lower than the background level 6b1, and the pixels of the lightness level below this level are set to white “1”. This time, safety shoes 6b3 etc. will remain as images.

In this way, the monitored object in the field of view is captured in the image regardless of its brightness.

First, the safety monitoring device inputs the image of the camera 1 and counts the number of white "1" pixels in the windows 1.63 and 2.664. Next, the image of the camera 2 is taken in and the number of pixels is counted similarly. Next, the camera 4 is similarly processed.

If any of the detected pixel numbers of the cameras 1 to 4 is greater than or equal to the determined reference value, an operation disable signal is output to the welding robot 7, and the pixel numbers of all the cameras are determined. If it is smaller than the reference value, the operation enable signal is output.

By repeating the above operation, the safety monitoring device can perform safety monitoring regardless of the brightness of the monitored object.

 A flow chart of the image processing of the present invention is shown in FIG.

A grayscale image is input from camera 1 to camera 4 in step 50, binarized in step 511, window 1 is obtained in step 512, and pixels are counted in step 513. In parallel with this, binarization is performed in steps 521 to 5n1. Is performed, window 2 to window n are obtained, and step 52 is performed.
The pixels are counted in steps 3 to 5n3, and a histogram is generated in step 53.

〔The invention's effect〕

As described above, according to the present invention, by dividing the background into two regions, setting a window for each region, and performing binarization processing with a different binarization level for each window, it is possible to obtain any brightness characteristic. The monitored object can also be detected with certainty, which eliminates the need for the white (yellow) wire on the safety shoes or the replacement of gloves as in the past, which increases the reliability of the safety monitoring device and improves the work efficiency. The efficiency is significantly improved.

[Brief description of drawings]

FIG. 1 is a schematic perspective view showing an embodiment of the present invention, FIG. 2 is a view for explaining background division and window setting in this embodiment, and FIG. 3 is a possibility of extracting a monitored object. FIG. 4 is a flowchart for image processing of the safety monitoring device. 5 ... Safety monitoring device (image processing device), _6a , _6b , _6c , _6d ...
… Monitoring area (background), 7… Welding robot, 8… Jig stand, 9… Welding part, 61… Monitoring area 1 (black), 62
…… Monitoring area 2 (green), 63 …… Window 1, 64…
... Window 2.

Claims (1)

(57) [Claims] 1. A plurality of windows are set in the field of view,
Binarization is performed with different brightness levels for each window.
A method for detecting a monitored object in a safety monitoring device having a visual function provided with a function of counting white (“1”) pixels after binarization, wherein the monitoring area is divided into at least two parts, and each area is set to a brightness value. Different backgrounds are set up, and windows are set in each area, and the monitored object with higher brightness than the background is detected on the background side with low brightness, and the monitored object with lower brightness than the background is detected on the background side with high brightness. And a method for detecting a monitored object in a safety monitoring device.