JPS63296174A - Abnormality detector using image pickup device - Google Patents

Abstract

PURPOSE:To reduce the burden on a guard and to find the abnormality of a plant or the like in an early stage by picking up the image of a monitoring object by an image pickup device and detecting the change with time of hue in accordance with the obtained video signal and discriminating the abnormal state based on the variation of hue. CONSTITUTION:An image processing computer 2 consists of an input processing part 21 which takes in the video signal periodically from a color television camera 1 and decomposes this video signal to three primary colors of light, a storage part 22 in which three primary colors decomposed by the input processing part 21 are stored, a hue variance detector 23 which periodically takes out the video signal of three primary colors stored in the storage part 22 from the storage part 22 and calculates the time-series variance of hue from the preceding video signal, and an abnormality discriminating part 24 which discriminates whether abnormality occurs in a plant device or not in accordance with the variation of hue calculated by the hue variance detecting part 23. Thus, the burden on the guard is reduced and the abnormal state of a plant or the like as the monitoring object is quickly and accurately detected.

Description

[Detailed Description of the Invention] [Objective of the Invention] (Industrial Application Field) The present invention captures an image of a monitoring location using an imaging device, and detects, for example, steam leakage from a boiler, fire, or human intrusion into a hazardous area. The present invention relates to an abnormality detection device using an imaging device that detects an abnormality based on a video signal.

(Conventional technology) Generally, in large plants such as power plants and chemical factories,
Operating controls and various monitoring instruments.

All displays, etc. are centrally installed in a central control room, and can be controlled remotely.
Operation is performed through remote monitoring and central control. Furthermore, in order to monitor various types of equipment in a plant, a method is used in which a supervisor is not permanently stationed at the site, but rather the supervisor periodically patrols the plant equipment.

(Problem to be solved by the invention) However, in this method, if the patrol cycle is lengthened, even if an abnormality occurs in the plant, there is a high possibility that the abnormality will be discovered late; conversely, if the patrol cycle is shortened, There were problems in that the burden on the observers increased and the number of personnel needed to be increased.

On the other hand, as an alternative to single patrols or in combination with patrols, TV cameras are installed at various locations in one plant, and the video signals from the TV cameras are input to a monitor TV in the central control room, and supervisors monitor the images on the monitor TV. Methods have also been adopted to detect abnormalities in plants. However, with this method, the monitor must always keep an eye on the monitor television, which again poses a problem in that the burden on the monitor is heavy.

In view of the above points, it is an object of the present invention to provide an abnormality detection device using an imaging device that can reduce the burden on a supervisor and detect abnormalities in a plant or the like at an early stage.

[Structure of the Invention] (Means for Solving the Problems) Therefore, the present invention captures an image of a monitoring target with an imaging device,
A temporal change in hue is detected from the obtained video signal, and an abnormal state is determined based on the amount of change in hue.

(Function) When images are taken of the monitored target, if there is a steam leak or ignition in plant equipment, or people entering or exiting a specific location,
Since a hue change is detected in the video signal, an abnormal state can be determined.

(Example) Examples of the present invention will be described in detail below.

FIG. 1 is a block diagram of an abnormality detection device according to an embodiment of the present invention. In the figure, this anomaly detection device includes a color television camera 1 that photographs the status of various equipment in the plant, an image processing computer 2 that periodically receives video signals from the color television camera 1, and determines an abnormality. A display bag to show the results! i! It is composed of 3.

The image processing computer 2 includes an input processing unit 21 that periodically inputs a video signal from a color television camera l and separates the video signal into three primary colors of light; and a storage unit 22 for storing.

The 3H color video signal stored in the storage unit 22 is stored in the storage unit 22.
and a hue variation detection unit 23 that periodically extracts the video signal from the video signal and calculates a time-series change in hue from the previous video signal.

From the hue change amount calculated by the hue change detection unit 23,
Determine whether an abnormality has occurred in plant equipment,
It is comprised of an abnormality determination section 24 that outputs the determination result to the display device 3.

With the above configuration, a place to be monitored, such as plant equipment, is imaged by the color television camera 1. As a result, a video signal is input from the color television camera 1 to the image processing computer 2. This video signal is processed in the input processing section 21.
The 0 hue fluctuation is processed every minute constant period Δ, and is divided into image information of three elements of the three primary colors R, G, and H for color painting and stored in the storage unit 22, as shown in FIG. 2(a). In the detection section 23.

Using the image information stored in the storage unit 22, temporal changes in hue are calculated in a linear manner.

FIG. 3 is a flowchart showing this process, and at each point in time at each fixed period Δτ, first, the video signal of the original image at the current time is input. As a result, for example, the color element R at each pixel divided into 512 vertically and horizontally
, G, and B (for example, 0 to 255) are obtained (101). Next, the pixel information consisting of the three elements R, G, and B of this original image is expressed by the formula X = rR + gG + · b [l As a result, one element information X is obtained. Here, -r, g, and b are weighting coefficients that indicate the importance of each element for difference determination when one element information X is used, and are calculated using a common value for each pixel, such as 0.

To determine steam leakage, consider the white difference and use r=0.3.
3. g=0.33. b = 0.33, and r = 0.8 for firing determination. g=0.1. By using the coefficient b=0.1, which focuses on the red color, it is possible to increase the efficiency of abnormality detection (Fig. 3 102.Fig. 2 (
b)).

Next, 1-element information Xi◆1 image and 1-element information X at 1gA generated in the same way from the color image 61 seconds ago
i, calculate the difference (Xi + s - Xi) for each pixel, and set the absolute value of that value as the difference ΔXi (Fig. 3, 1
03, see Figure 2(c)).

Next, the sum of n differences calculated before that point
For example, if the difference ΔXu is calculated now, the total sum Y is as follows (see FIG. 3 (104) and FIG. 2 (d)).

Using the X calculated here, a determination is made in the abnormality determining section, that is, the number k of pixels larger than a preset threshold value a from the value r just added is determined (105)
, exceeding this threshold value a is the basis for determining that the variation in hue is actually larger than the error component. In other words, it can be interpreted that the steam leak, oil leak, or human intrusion that is being detected is accompanied by a temporal change in hue.

Shigashi.

Since the hue fluctuates due to external disturbances such as flickering due to mechanical errors in television cameras or changes in sunlight, we decided to use y obtained by adding the difference ΔXi in the judgment in order to make the judgment resistant to such disturbances. .

Therefore, the number k of pixels larger than this threshold value a is set in advance, and if it is larger than the threshold value A, it is determined to be abnormal, and in other cases, it is determined to be normal. This is also to avoid erroneous judgments due to external disturbances, and when an abnormal state like the one described above occurs, hue fluctuations occur over a certain extent in the image photographed by the television camera. Therefore, when a change in hue occurs in a plurality of pixels in this way, it is determined that there is an abnormality (see FIG. 3 106 and FIG. 2(e)).

This judgment result is displayed as an abnormality judgment signal! 13, and this notifies the monitor of abnormality and normality. If a supervisor notices an abnormality, he or she will immediately take countermeasures such as switching the TV monitor to show the scene in question or actually flying to the scene. This results in
The burden on the supervisors will be reduced because they will be able to detect abnormalities in the plant early and take prompt action without having to constantly watch the monitoring location.

In addition, in the embodiment described above, three color elements R
, G, and B to calculate l element information X, and abnormality determination is made by comparing this difference value with a threshold value. The calculation method can be set arbitrarily, such as comparing the data with a threshold value and determining an abnormality when all three elements or any two elements exceed the threshold value.

Furthermore, the present invention is not limited to on-site monitoring of plants, etc. as in the above-described embodiments, but is naturally applicable to a wide range of abnormality detection in devices and objects that undergo a hue change in the event of an abnormality.

Further, in one or more embodiments, a television camera is used to extract the video signal; however, a solid-state image sensor such as a CCD image sensor may also be used. Furthermore, it is also possible to temporarily record the video signal extracted in this way on, for example, a video tape recorder or an optical disk, thereby detecting an abnormality when reproducing the video signal.

[Effects of the Invention] As described above, according to the present invention, an abnormal state is determined by capturing an image of a monitoring target using an imaging device and detecting a change in hue based on the obtained video signal. While the burden on personnel is reduced, abnormal conditions in plants to be monitored can be detected quickly and accurately, greatly improving the ability to monitor abnormalities in plants compared to the past. Safe driving becomes possible.

[Brief explanation of the drawing]

FIG. 1 is a block configuration diagram of an abnormality detection device using an imaging device according to an embodiment of the present invention, FIG. 2 is an explanatory diagram showing processing of the device, and 311! ! 1 is a flowchart showing abnormality determination processing. DESCRIPTION OF SYMBOLS 1... Color television camera, 2... Image processing computer, 3... Display device, 21... Input processing section, 22...
- Storage unit, 23... Hue variation detection unit, 24... Abnormality determination unit. (7317) Agent Patent Attorney Aide Kensuke (8g6
9) Ken Daishimaru Figure 1 Figure 2

Claims (1)

[Claims] An imaging device that photographs a place or object to be monitored and extracts a video signal, an image processing means that detects a temporal change in hue based on the extracted video signal, and an abnormal state based on the detected amount of change in hue. What is claimed is: 1. An abnormality detection device using an imaging device, comprising: an abnormality determining means for determining an abnormality; and an abnormality display device for displaying an alarm of an abnormal state.