JPS63133060A - Fish living condition monitoring instrument - Google Patents

Abstract

PURPOSE:To quickly and automatically judge whether a poison has flowed into water by measuring the moving speed of the position of the center of gravity and the number of times of the inversion movement of a fish. CONSTITUTION:A water tank 1 is supplied with water to be inspected and a fish 10 is raised in the water tank 1. The fish 10 is illuminated by an illuminator 7 and the image of the fish 10 is sensed by an image sensing device 3 like, for example, an industrial television camera. The center of gravity and the direction of the fish 10 are obtained by an image processor 4 and a result is fed to an arithmetic unit 6. The unit 6 measures the moving speed and the presence of the inversion movement of the fish 10 from the informations on the center of gravity and the direction of the fish 10. Above-described processing is repeated for a prescribed time to obtain the distributions of the positions and the moving speeds and the number of times of inversion movement of the fish 10. By comparing the values of those factors with those stored in the arithmetic unit 6 in advance, whether the movement of the fish 10 is abnormal is judged. If abnormal, an alarm is emitted from an alarm device.

Description

[Detailed Description of the Invention] [Industrial Application Field] The present invention relates to a close monitoring device for monitoring and determining the behavior of aquatic animals kept in the raw water of a water treatment plant or the inflowing sewage of a sewage treatment plant. .゛ [Conventional technology] Conventionally, in water treatment plants, in order to monitor whether or not poisonous substances have been mixed into the raw water, a portion of YK water is introduced into the water layer and treated with crucian carp, carp, Japanese grass, and grass. If you are raising aquatic animals such as oysters, and a poisonous substance is mixed into the raw water, you can monitor the inflow of toxic substances into the raw water by taking advantage of the phenomenon that the fish behave abnormally or die. There is. In addition, sewage treatment plants need to know whether legally prohibited toxic substances have entered the incoming sewage, and for this reason they conduct intermittent manual water quality analysis. However, such monitoring of toxic substances in water that relies on manual visual inspection of fish and water quality analysis has a problem in that continuous monitoring and early detection are difficult, resulting in delays in taking measures such as stopping water distribution to customers.

Also, as a method of monitoring fish, there is a method of detecting fish in the water layer from the upper part of the water layer using an industrial television camera (ITV) and processing the image, for example, as described in the lecture collection of the 36th National Water Supply Research Conference, p.
464-466, and according to this method, when a fish floats on its belly on the surface of the water, the fish can be recognized as an independent bright spot of a certain size or larger, and is present near the water surface. It is stated that by extracting only the high brightness areas of fish and the changes in light due to irregularities on the water surface, the background can be sorted out and the behavior of the fish can be determined.

Further methods of monitoring fish include monitoring the movement of a plurality of organisms within one or more tank devices using video equipment and analyzing the movement of the organisms using computer equipment to make predictions corresponding to the statistical distribution of expected movement patterns. A method of comparing parameter sets is described in, for example, Japanese Patent Laid-Open No. 61-46294.

[Problem that the invention seeks to solve]

In the conventional method described above, in which fish in the water layer are detected from the upper part of the water layer using ITV and image processed, abnormalities in fish behavior cannot be detected early because fish cannot be detected until they rise to the surface of the water. The method of monitoring and analyzing the movement of living organisms mentions position, shape, and orientation as characteristics of fish movement, but does not describe a specific method for determining abnormal behavior of fish from these characteristics. do not have.

SUMMARY OF THE INVENTION An object of the present invention is to provide a close-up monitoring device that can quantitatively and continuously monitor the movement of fish in a water layer and quickly determine the presence or absence of toxic substances in the water.

[Means for solving problems]

The above purpose is to provide an aquarium in which aquatic animals (fish) are raised in order to detect the inflow of toxic substances into water, an imaging device that converts image information of the fish into electrical signals, and a system that detects the position of the fish from the image information obtained from the imaging device. Image recognition device i! that detects orientation!
(image processing device) and means (computing device) for detecting the speed and reversal of the fish from the position and orientation of the fish and determining abnormality of the fish from the rotation of the reversal, etc.
? This is accomplished with an M vision device.

[Effect]

In the above-mentioned Jiomai monitoring device, an image of a fish kept in an aquarium for detecting the inflow of toxic substances into the water is converted into a brightness signal by an imaging device, and the brightness signal is digitized and imported into the image memory of an image processing device, and the image information is After binarizing and extracting the outline, the position and direction of the center of gravity of the fish in the fish image are calculated, and the speed and reversal of the fish are detected by a calculation device from the position and direction of the fish, and the fish image is obtained by measuring these over a predetermined period of time. By comparing fish behavior measurement patterns such as fish position and speed and number of reversals from measurement information with normal fish behavior patterns, it is possible to monitor and quantify abnormal fish behavior, which generally includes reversals, frenzy, nose raising, etc. It can be judged accurately.

〔Example〕

An embodiment of the present invention will be described below with reference to FIGS. 1 to 5.

FIG. 1 is an overall configuration diagram showing an embodiment of a close eye monitoring device according to the present invention. In Figure 1, 1 is an aquarium in which aquatic animals (fish) are raised to detect the inflow of toxic substances into the water, 2 is a back screen, 3 is an imaging device that converts image information of fish into electrical signals, and 4 is an imaging device! ! ! 3 is an image processing device (image recognition device) that detects the position and orientation of the fish from the image information obtained from the image information, 5 is a monitor, 6 is a means for detecting reversal of the fish from the direction of the fish, and determining an abnormality of the fish from the number of reversals. 7 is a lighting device, 8 is a lighting control device, 9 is an alarm device, and 10 is an aquatic animal (fish).

Figure 1 Water layer 1 for fish rearing to detect inflow of toxic substances into water
Water is always supplied to the system, such as raw water from a water treatment plant, influent sewage from a sewage treatment plant, or river water in the case of toxic substance monitoring in a river. Normally, one or more fish 10 are kept, but in this example, for ease of explanation and understanding, the case of one fish will be explained as an example. , Japanese warbler, Japanese tanago, and Oikawa are bred here. The lighting device 7 that illuminates the fish 10 in the water layer 1 requires uniform illumination in order to apply the image processing technology, so it is controlled by the lighting control device W8, and there is a frosted glass between the lighting device 7 and the water layer 1. A translucent back screen 2 corresponding to a light scattering plate made of white acrylic or the like is provided. This back screen 2 has a white background and a black color for the fish 10, so that it also has the role of recognizing the fish 10 with good contrast. The imaging device 3 that converts the image of the fish 10 in the water layer 1 into an electric signal (video signal) uses, for example, an industrial television camera (ITV), and decomposes the two-dimensional space to be imaged into, for example, 256 x 256 pixels. It outputs an electric signal with a voltage corresponding to the brightness of each pixel (111 degrees). Image processing device (image recognition device iri) 4
outputs horizontal and vertical synchronization signals to the imaging device 3 to control the timing of imaging, and the video signal output from the imaging device 3 is sent to the image processing device 4.

Further, the image processing device 4 is configured to handle the fish 10 obtained from the imaging device 3.
The fish 10 is recognized based on the image information, and the center of gravity position G and orientation D of the fish 10 are determined. In addition, image processing bag fl! f4
The configuration and operation of will be described later. A monitor 5 is also connected to the image processing device 4, and displays fish images captured by the imaging device 3, the results of image processing, and the like. Next, the signals of the center of gravity G and direction D of the fish 10 obtained by the image processing device 4 are sent to the arithmetic device 6, and the arithmetic device 6 uses the sent information about the center of gravity G and the direction D of the fish 10 to determine the moving speed V of the fish 10. The presence or absence of reversal is measured, and this process is repeated for a predetermined period of time to obtain the position and velocity distribution of the fish 10 and the number of reversals. This arithmetic device 6 stores the position and speed distribution and the number of reversals when the fish 10 is in a normal state, and by comparing these normal values with the above online measurement value, it is possible to determine whether the fish 10 is in a normal state or not, and by comparing these normal values with the above-mentioned online measurement value. If a difference occurs, the movement of the fish 10 is determined to be abnormal. This judgment result is sent to the alarm device 9, and the alarm bag is sent! ! ! When 9 receives this abnormality detection signal, it sounds an alarm and outputs a voice message to urge the supervisor to investigate the water quality.
Image processing bag [i! In addition to monitoring the fish 10, the monitor 5 connected to the fish 10 displays measured values such as the position and speed of the fish 10 and the number of reversals, and also changes the color of the monitor screen to red, for example, when the movement of the fish 10 is abnormal. etc. to visually notify abnormalities. Furthermore, the arithmetic device i! The normal values of the position and velocity distribution of the fish 10 and the number of reversals stored in the fish 10 can be corrected or changed according to the type of the fish 1o and environmental conditions such as water temperature and season. Note that details of the configuration and operation of the arithmetic unit 6 will be described later.

FIG. 2 is a side view of the detailed configuration of the image processing device 4 shown in FIG. 1. In FIG. 2, 401 is an A/D conversion device, 402 is a grayscale image memory, 403 is a timer, 404 is a binarization device, 405 is a binary memory, 406 is a contour extraction device, 407
408 is a position calculation device, 408 is a direction calculation device, 409 is a selector, 410 is an input/output device, and 411 is a selector.

This image processing device 4 is a system for processing fish 1o obtained by the imaging device 3.
This is a means for detecting the center of gravity 1tYG and direction D of the fish 10 from the image.

The video signal from the imaging device (ITV) 3 in Fig. 2 is an A/D
The converted image is digitized by the conversion device 401 and stored in the grayscale image memory 402. This grayscale image memory 402 has, for example, 256 x 256 pixels x 8 bits (each pixel has 256 floors
R), and the digitized images are stored continuously.

First, the detection of the center of gravity G of the fish 10 will be described. The timer 403 sends image information of the fish 10 in the grayscale image memory 402 to the binarization device 404 via the selector 409 at preset time intervals Δt. Binarization device 4
04 is image information (luminance information H) of the grayscale image memory 402
Then, all pixels brighter than the threshold value T are set to 0'', and all pixels darker than the threshold value T are set to 11''.

The data is stored in the binary memory 405. This binary memory □405
For example, has a capacity of 256 x 256 pixels x 1 bit,
Luminance information H(
xyj) and the brightness B(IIJ) of the pixel in the i row and j column of the binary memory 405.
The calculation of the value is performed using the following formula.

When H(i, J)≧T, B (i, j) = O(1)
When H(i* j)<T, B (i, j) = 1
(2) Here, the image information (brightness information H) is that the fish 1o is black (low brightness) and the background is white (high brightness), so the brightness B (i.

The set of pixels (], IIJ where J) has a value of 1'' is fish 1o
, and the pixel (i,
j) represents the background. Here, the position calculation device 407 calculates the brightness B (i) of the binary memory 405.

j), the coordinates B (Xt, Yt) of the pixel (11J) whose brightness B (IIJ) is 1''', and the total number of pixels are N, then the center of gravity of the fish 10 is h'ZG (Xg ,
Yt ) is calculated using the following formula.

X yo = Σ X 1 / N (
3) i=1 i=1 Next, to explain the detection of the direction D of the fish 10, the black image information in the grayscale image memory 407 output at every time interval Δt via the timer 403 is used by the contour extraction device. 40
Send to 6. This contour extraction device 406 has a function of emphasizing the luminance difference in black image information (luminance information H), and even if
Using known image processing techniques such as Laplacian and median filters, point images are created with fish 10 being black and background power 1.
Since it is a white color, the part b) where the brightness difference between the background and the fish 10 is large is emphasized as an outline, and the brightness power of the part b) where the brightness difference is small inside the boundary between the background and the fish 10 is reduced. , to this contour-enhanced information power, a contour extraction device 406
The signal is output from the selector 409 to the binary input device 404 . Next, the binary output device 404 inputs this contour-enhanced information P'(ly j).

Binarize it using the following formula and write it as a binary memo i-no-405.
Information B (i, j) is stored in which the boundary between 0 and the background, that is, the outline, has a brightness of "1" and the rest has a brightness of It OI+.

When P(IIj)≧T, B (it j) = 1 (
5) When P (11j)<T, B (i, j) = O
(6) Here, the direction calculation bag 5t1408 is a binary value T=l
In order to calculate the direction of the fish 10 based on the information from J405, first, the pixels of luminance tt l u in the binary memory 405 are thinned to make the outline a line with a width of 1 pixel, and then the line of this outline is Calculate the connection directions for all pixels representing the contour in four directions: vertical, horizontal, and 45° directions x 2, then classify all pixels representing the outline into the four directions above and count each pixel in each direction. Among the four directions, the direction with the largest number of pixels is defined as the direction of the fish 10. As described above, the position calculation device 407 and the orientation calculation bag [4
The center of gravity of the fish 10 calculated by 08 [G and direction D
Is the information about the input/output device? ! 410 to the arithmetic bag fit6, and the video signal from the A/'D converter 401;
Signals from the grayscale image memory 402, binary memory 405, position calculation device 407, and orientation calculation device 408 are sent to the selector 4.
It can be output to the monitor 4 via 11.

FIGS. 3(a) and 3(b) show two parts of the binarization device 404 in FIG.
This is an explanatory diagram of the value conversion process, and FIG. 3(a) shows the grayscale image memory 4.
An explanatory diagram of the point image stored in 02, FIG. 3(b) is the 3rd
FIG. 4 is an explanatory diagram of the brightness distribution on the line A- of the point image in FIG. 3A and the binarization threshold T of the binarizer 404. In FIG. 3(a), B indicates the background part, and F indicates the fish 10 part. In FIG. 3(b), A-A in FIG. 3(a)! The brightness distribution above is the brightness I of the background B part] (B)
is high, and the brightness H(F) of the fish F(10) portion is low. The brightness information H of this point image is converted to a binarization threshold 'r (F<T≦
When the binarization device 404 performs binarization according to the above equations (1) and (2) using B), the part of the fish F(10) with a low m degree is extracted as a binary image with luminance rr 1 n, It is stored in binary memory 405.

Figures 4 (a), (b), and (c) are the contour extraction system shown in Figure 2.
406 and an explanatory diagram of point image contour extraction processing and orientation detection processing of the orientation calculation device 408.

FIG. 4(a) is an explanatory diagram of the point image stored in the grayscale image memory 402, and FIG. 4(b) is an explanatory diagram of the point image stored in the grayscale image memory 402, and FIG. 4(b) shows the point image in FIG. An explanatory diagram of an image that has been binarized by the device 404 and stored in the binary memory 405 and thinned by the orientation calculation device 408. FIG. 4(c) is the binarized thin line after contour extraction in FIG. 4(b). From the converted point image, the orientation calculation device 408 determines the connection direction of pixels in the vertical direction (direction 3), horizontal direction (direction 1), and 45° direction X2 (direction 2).
, 4 directions) is an explanatory diagram of four directions in which the direction of the fish F (10) is determined by classifying them into four directions. In FIG. 4(a), B is the background, Fl is the main body of fish 10, and Fi is fish 1.
Each represents the fin part of fish 0, and the fin part Fx of fish 10
The brightness H(Fz) of is lower than the brightness H(B) of the background part B but higher than the brightness H(Fl) of the main body part Fl of the fish 10 (Ft<Fz<B), and then in FIG. 4(b' ), in
The point image in FIG. 4(a) is extracted by the contour extraction device 406 into the background portion B, the main body portion F1 of the fish F (10). After extracting the contour by emphasizing the portion with a large brightness difference between each boundary of the fin portion F2, the information P with this contour emphasis is converted into the above-mentioned ((
5) and (6) are binarized and stored in the binarized memory, and further thinned by the orientation calculation device 408, the body part F1 and the fin part F of the fish F (10) are obtained as shown in the figure.
Along with the boundary between 2 and the background part B, the boundary between the body part F1 and the fin part F2 of the fish F (10) also has a width of 1, which forms the outline line.
It is detected as a column of pixel brightness 111 +1.

Next, in FIG. 4(C), from the fish image which has been binarized and thinned after contour extraction in FIG. )
, horizontal direction (1 direction).

If the direction of fish F (10) is determined by classifying into four directions of 45° direction X2 (2 and 4 directions) and calculating the connectivity of all pixels, in the example shown The direction is 45° direction D2 (direction 2).

FIG. 5 is a side view of the detailed configuration of the arithmetic unit N6 in FIG. 1. In FIG. 5, 601 is a selector, 602 is a center of gravity storage device, 603 is a reversal detection device, 604 is a speed calculation device, and 6
05 is a reversal storage device, 606 is a center of gravity determination device, 607 is a speed determination device, and 608 is a reversal determination device. This calculation device 6 processes the fish 1 from the image processing device (image Lz recognition device) 4.
Means for calculating the moving speed of the fish 10 and detecting reversals based on the information on the center of gravity of the fish 10 [0 and the direction D, and comparing the center of gravity of the fish 10 and the number of reversals of the speed 9 with normal values to determine abnormality in the behavior of the fish 10 to do.

The center of gravity Ii of the fish 10 calculated from the image processing device 4 in FIG. 5! G (X g , Y t ) and direction D information are sent to the center of gravity storage device 60 via the selector 601.
2 and the reversal detection device 603, and the center of gravity G (Xi
, Yt) is stored in the barycenter position storage device 602 for a predetermined time T at every time interval Δt, and this barycenter G
Xi − Yw ) forms the appearance frequency distribution of the center of gravity G.

Information on the frequency distribution of the center of gravity G during this predetermined time T is sent to the center of gravity determination device 606 and compared with the frequency distribution of the center of gravity during normal times.
If the deviation is larger than the predetermined value, alarm bag F! 17
Sends an abnormality signal to. In addition, the speed calculating device 604 calculates the moving speed V of the fish 10 at each time interval Δt from the center of gravity a (XJI, Yll) sent during a predetermined time T at every time interval Δt from the center of gravity storage device 602. The moving speed V forms a frequency distribution of the speed V.

Information on the frequency distribution of the speed V during the predetermined time T is sent to the speed determination device 607 and compared with the frequency distribution of the speed during normal times.
If the deviation is larger than a predetermined value, an abnormality occurrence signal is sent to the alarm device 17. One reversal detection device 603 stores information on the direction D of the fish 10 sent from the selector 601, and also stores the center of gravity G of the fish 10 sent from the center of gravity storage device 602 for a predetermined period T at every time interval Δt.
(Xi.

The moving direction (distribution) of the fish 10 can be calculated and stored from the information of the fish 10 (Y wealth), and the reversal of the fish 10 can be detected from the information of the moving direction of the fish 10 and the direction D of the fish 10. The information is sent to and stored in the reversal storage device 605 for a predetermined period of time T. 6 Furthermore, the number of reversals of the reversal information of the fish 10 during the predetermined period of time T stored in the reversal storage device 605 is sent to the reversal determination device 608. The number of reversals is compared with the normal number of reversals, and if the deviation is larger than a predetermined value, an abnormality occurrence signal is sent to the alarm device 7.

〔Effect of the invention〕

According to the present invention, abnormalities in the fish can be determined by continuously and quantitatively monitoring the behavior of the fish (aquatic animal) based on the position of the fish's center of gravity, moving speed, and number of reversals.

It is possible to quickly and automatically determine whether poisonous substances have entered the water, and the monitoring of blue water such as incoming raw water at water treatment plants can be carried out accurately and labor-savingly, thereby ensuring the safety of water quality.

[Brief explanation of the drawing]

FIG. 1 is a side view of the overall configuration of an embodiment of the image processing device according to the present invention, FIG. 2 is a detailed configuration diagram of the image processing device shown in FIG. 1, and FIGS. 2 Fish images and 2
Explanatory diagram of valorization threshold, Fig. 4 (a), (b), (c)
2 is an explanatory diagram of the fish image in FIG. 2, the binarized thinned image after outline extraction, and the four directions of the direction of the fish, and FIG. 5 is a side view of the detailed configuration of the arithmetic device in FIG. 1. DESCRIPTION OF SYMBOLS 1... Water layer, 2... Back screen, 3... Imaging device, 4... Image processing device (image recognition device for detecting the position and orientation of fish, etc.), 5... Monitor, 6・・・
- Arithmetic device (means for detecting reversal of fish, means for determining abnormality of fish from the number of reversals, etc.), 7... Lighting, 8.
...Lighting control device, 9...Alarm device, 1o...Fish (
aquatic animals).

Claims (1)

[Claims] 1. An aquarium in which aquatic animals are raised to detect the inflow of toxic substances into water, an imaging device that converts image information of the aquatic animals into electrical signals, and a position and orientation of the aquatic animals based on the image information obtained from the imaging device. A fish condition monitoring device comprising: an image recognition device for detecting a change in the aquatic animal; a device for detecting a reversal based on the orientation of the aquatic animal; and a means for determining an abnormality in the aquatic animal based on the number of times the aquatic animal turns.