JPS63307358A - Method for recognizing position of plural fishes - Google Patents

Abstract

PURPOSE:To obtain a method for accurately detecting the behavior states of plural living bodies by labeling the living bodies as one of image processing methods and extracting an image of each living body solid body. CONSTITUTION:An image monitoring device 30 picks up an image of fishes in a water tank by an image pickup device 20 and a signal which is outputted at constant intervals is A/D-converted 32 and stored in a multilevel image memory 32M. A threshold determining circuit 34 determines a binary coding threshold value with a calculation result from a brightness frequency distribution calculating circuit 33 and a signal from a fish body area setting circuit 34S. The signal of the memory 32M and a circuit 34 is stored into a binarization memory 35M with a binarization circuit 35. Then binarization fish images are numbered according to the area and a gravity center calculating circuit 37 calculates the centers of gravity of the fishes according to the number of the binarization fish. An overlap counting circuit 38 calculates the number of overlaps of fishes from a specific expression and an overlap distributing circuit 39 distributes the found number of overlaps in the area increasing order. Here, an abnormal position judging circuit 40 judges abnormal water quality and sends out a warning signal when the number of fishes nearby the water surface is larger than a set value.

Description

[Detailed Description of the Invention] [Industrial Application Field] The present invention relates to image monitoring of a plurality of organisms kept in inflow water and treated water of a water purification plant or sewage treatment plant. Concerning how to accurately understand behavior.

[Conventional technology]

At water treatment plants, in order to determine whether or not toxins have been mixed into the raw water, a portion of the raw water or purified water is introduced into an aquarium where fish such as crucian carp, carp, dace, tanago, oysterfish, and goldfish are raised. ing. Similarly.

Fish are sometimes kept in order to monitor the presence of toxic substances in treated water from sewage treatment plants, discharged water, river water, and lakes. If poisonous substances are mixed into the water, the fish may behave abnormally or die, so this is visually monitored. However, since it relies on visual inspection and cannot be detected unless a person is monitoring it, automatic monitoring has been desired.

An automatic method for monitoring fish is to photograph fish in an aquarium from the top of the tank using an industrial television camera and process the images (Reference: 36th Gold National Waterworks Research Presentation Association).

The Mizo Enshu p, 464-466 have been devised. This method uses a camera to capture the phenomenon of fish that have died due to abnormal water quality rising up onto the surface of the water with their bellies facing upwards, detecting them as ``independent bright spots of a certain size or larger'', and detecting them as aK2, which then issues a warning of abnormal water quality. It is. Further, Japanese Patent Laid-Open No. 61-46294 discloses a method of monitoring the behavior patterns of a plurality of living things in order to monitor water quality. The disclosed technique describes that a water quality sensor measures temperature and water quality, and the measured values are used as a reference to determine whether a behavioral pattern is normal or abnormal. It is stated that the behavior pattern includes speed and position. However, even if we simply analyze behavioral patterns using image technology, is this idea already known? j [It is difficult to implement because it does not disclose how to specifically detect or evaluate the speed and position of a few objects.

[Problem that the invention seeks to solve]

Among the above conventional techniques, the first method is to detect dead fish.
Second, there was a problem in that water quality abnormalities could not be detected until the fish died and floated to the surface.Secondly, there was a problem in that no measures were taken when the fish died at the bottom of the tank and did not float to the surface. Moreover, since JP-A No. 61-46294 does not disclose a specific method for detecting abnormal water quality, it is difficult to implement. moreover,
If only one fish is tested, it is difficult to distinguish between disease and water quality abnormalities based on its behavior.

An object of the present invention is to provide a method that can accurately detect the behavioral states of multiple living things.

[Means for solving problems]

The above purpose is one of the image processing methods for multiple living things.
This is achieved by performing labeling, extracting images for each individual creature, and calculating the position of the individual creature in the aquarium.

[Effect]

When fish have abnormal water quality, they exhibit a behavior called nose raising, in which they stick their mouths out to the surface of the water. By sequentially detecting the position of each individual fish in the aquarium, we can accurately understand this abnormal behavior.
The behavioral status of fish can be evaluated effectively.

〔Example〕

Examples will be described below with reference to the drawings.

The configuration of the embodiment will be explained using FIG.

The water to be tested is transferred to a water tank 10 by a water supply pipe 11 and a water supply pump 12.
supplied to The water introduced into the water tank 10 is drained through a drain pipe 13. In the aquarium 10, a breeding space 1 is partitioned by partition plates 18A and 18B such as wire mesh or perforated plates.
There is 9, and here fish 14A, 14B, 14C are raised 1
In this embodiment, a case where there are three fish will be explained, but a similar embodiment can be applied to a case where there are more fish. lighting device ill 5
A.

15B illuminates the fish 14 in the aquarium 10. Lighting device 15A
, 15B and the aquarium 1o are provided with a semitransparent plate 16 made of a semitransparent substance such as ground glass or paper. Lighting bag [
The semi-transparent plate 16 receives the lights 15A and 15B and scatters the light, and the light emitted from the entire semi-transparent plate 16 illuminates the aquarium 10. An imaging device 20 such as an industrial television camera (ITV) is arranged on the opposite side of the water tank 1o when viewed from the lighting devices 15A and 15B. That is, the imaging device 20 includes the lighting device 15A,
The light emitted from 15B and passing through the semi-transparent plate 16 is imaged.

Here, the imaging device 20 images the rearing space 19. Furthermore, an aeration device 17 is provided in the tank to prevent fish from coming to the water surface due to lack of dissolved oxygen in the water.

The signal from the imaging device 20 is guided to the image monitoring device 30. A detailed explanation of the configuration and operation of the image monitoring device 30 will be given later.

To briefly explain the functions of the image monitoring device 30, the one-image monitoring device 30 first captures captured images at preset time intervals of 1, performs image recognition of the fish body, and calculates the center of gravity of the fish body. The center of gravity of the fish is calculated sequentially at each time interval and the results are stored in memory. Based on this memory information, the behavior of the fish 14 is monitored by calculating a statistical pattern of the fish's position (center of gravity) at a preset measurement time T, and an alarm is issued in the event of an abnormality based on the monitoring results. Emitted from the device 70.

A monitor television 50 displays the captured image.

The image monitor 60 receives the signal from the image monitoring device 30 and
Display image recognition results.

Next, the configuration of the one-image monitoring device 30 will be explained in detail. The timer 31S has an initial setting time of 1'JJll? A trigger signal is output to the A/D converter 32 every 'lh. The A/D converter 32 receives the image signal output from the imaging device 20 in synchronization with this trigger signal, converts it from an analog value to a digital value, and stores it in the multivalued image memory 32M.

The brightness frequency distribution calculation circuit 33 receives the signal from the multivalued image memory 32M and calculates the brightness frequency distribution (histogram) of the multivalued image.

Here, the brightness frequency distribution represents the frequency of multivalues (brightness) of a multivalued image. The threshold value determination circuit 34 receives the calculation result of the brightness frequency distribution, and also receives the signal from the fish body area setting circuit 34S, and determines the threshold value for binarization based on both signals. The binarization circuit 35 receives the signal from the multilevel image memory 32M and the signal from the threshold value determination circuit 34, and stores them in the binarization memory 35M. In the image numbering circuit 36, the binarized fish 14A and 14B are
.

The portions 14C are numbered in descending order of area. The center of gravity calculation circuit 37 calculates the center of gravity of the binarized parts of the fish in numerical order.

In the overlap number calculation circuit 38, from the point setting circuit 38S,
From the input of the total number of fish N and the numbered total number m, the number P of overlapping fish is calculated. This value is expressed by the following formula.

P=N-m...(1
) The overlap distribution circuit 39 distributes the number of overlaps P obtained by equation (1) in the order of the fish with the largest area. The abnormal position determination circuit 40 determines from the abnormal setting circuit 40S and the position of the center of gravity of the fish.
If the number of fish near the water surface is greater than the set number of fish, it is determined that the water quality is abnormal, and a warning signal is output to the alarm device 70.

Next, the operation of the image monitoring device 30 will be explained in detail.

The timer 31T outputs A to the A/D converter 32 at each time interval.
/D conversion trigger signal is output. This h is 0.1
The time interval is about 1 to 2 seconds, and the following image processing is executed at this time interval.

The A/D converter 32 converts the multivalued image signal from the imaging device 20 from analog values to digital values in synchronization with the trigger signal output from the timer 31T.

The digital multi-value image signal is stored in the multi-value image memory 32M. The multivalued image memory 32M has 256 images in the vertical direction and 2 images in the horizontal direction.
There are 56 memory locations, and the luminance signal of the pixel corresponding to each memory location is stored as a digital value. The signal (luminance) at the i-th row and j-th column (i=1 to 256, j=1 to 256) of this storage location is expressed as G(i, j). If the A/D converter 32 converts analog values into 7-bit digital values, a (1t j) has 128 levels of digital values. FIG. 3 shows an example of a multivalued image stored in the multivalued image memory 32M. FIG. 3 represents an image with multilevel luminance. The brightness frequency distribution calculation circuit 33 calculates the brightness frequency distribution of the multivalued image. The brightness frequency distribution of FIG. 3 is shown in FIG. The threshold value determination circuit 34 determines the threshold value ■ upon receiving the calculation result of the brightness frequency distribution. Next, a method for setting the threshold value will be explained.

FIG. 4 represents the brightness frequency distribution. With the illumination method of the present invention, the fish school 14 can always be imaged as a dark object, so as shown in FIG. A first threshold value Il is set. Since the area f varies depending on the state, the minimum area is set. This threshold value determination method is particularly effective when the water is cloudy. However, if the water is not cloudy, it is better to use the second threshold 2.

In FIG. 4, the peak Pf represents the fish body, the peak pb represents the background, and the portion represented by Pe represents the gills and outline of the fish. In order to extract only the fish body, a second threshold value 2 is set at the boundary between Pf and Pe. As shown in Fig. 4, the threshold value is set to at least luminance ■1 in advance, and while searching each frequency in the direction of increasing luminance, the boundary (minimum value) between Pf and Pe is set.
If there is, select 2 for this brightness.

Next, the binarization circuit 35 receives the signal from the multivalued image memory 32M and the signal I (If or I2) from the threshold value determination circuit 34,
The multivalued image is binarized and stored in the binary memory 35M. Next, the specific operation of the binarization circuit 35 will be explained. 2
In the value converting circuit 35, the luminance G (IT
J) received.

All pixels brighter than the threshold are set to 1'0" level, and conversely, all pixels darker than the threshold are set to It 1 ++ level and stored in the binarization memory 35M. This set of binarized signals is stored as B( jyj), the binarization calculation is expressed by the following equation.

If G (x r J)≧■, then B(i, j)=O...
・(2) If G(i, j)<I, then B(i, j)=1・
...(3)(2) By calculating equation (3) for each pixel, the background is set to "O" level, and the fish school 14 is set to "1".
The result of binarizing FIG. 3 is shown in FIG. 5. The portion t4111 is shown by hatching.

The image numbering circuit 36 receives the signal from the binarization memory 35M and numbers the images in descending order of area. The images in FIG. 5 are recognized by being numbered in descending order of area as shown in FIG. 6. In addition, the image numbering material circuit 36 is 1 to 256 for a maximum of 256 independent objects in one screen.
can be numbered. Subsequently, the center of gravity calculation circuit 37 calculates the center of gravity of the binarized images in the order of numbering to obtain 1gL*.
Find the center of gravity of g2. The overlap number calculation circuit 38 calculates the number P of overlaps in the image by subtracting the total number m of numbered fish from the number N of reared fish set by the negative number setting circuit 38S. The overlap number distribution circuit 39 distributes the overlap number P in descending order of area based on the weight of the area number. In Figure 6, there is one object each with barycentric coordinates of -gt * gz, but the number of overlapping objects P is
From the area (Sz>St), it is recognized that there are two objects assigned to gl and indicating the center of gravity of gl.

The abnormal position judgment circuit 40 receives the vertical coordinates where fish are expected to be located and the number of fish at that position from the abnormality setting circuit 408 when the water quality is abnormal, and compares them with the calculated center of gravity and the number of fish. If there are more fish than the set number of fish near the coordinates, it is determined that there is an abnormality and a signal is output to the alarm device 70.

The position where fish exist when the water quality is abnormal, which is set by the abnormality setting circuit 408, is determined by the nose-up behavior caused by the fish due to the abnormal water quality. Since the nose-raising behavior occurs near the water surface of the aquarium, the value near the water surface is taken as the position of the fish when the water quality is abnormal.

〔Effect of the invention〕

According to the present invention, the positions of multiple groups of organisms in the aquarium can be calculated, and the number of individuals within a set range can be ascertained. It is effective to conduct more advanced and accurate water quality monitoring based on the proportion of fish representing the total number of fish kept6.

[Brief explanation of drawings]

Fig. 1 is a detailed explanatory diagram of the present invention, Fig. 2 is a detailed explanatory diagram of an image processing device in an embodiment, Fig. 3 is a multilevel image diagram of a school of multiple fish, and Fig. 4 is a brightness frequency distribution. 5 is a binary image diagram, and FIG. 6 is a diagram showing numbering regarding the binary image. 10...Aquarium, 14...Fish, 20...Imaging device,
32... A/D converter, 35... Binarization circuit, 36
...Image number material is circuit, 38...Overlap number calculation circuit, 40...Abnormal position judgment circuit, 5o...Monitor TV, 60...Picture 3/Mao diagram

Claims (1)

[Claims] 1. An aquarium in which a plurality of aquatic creatures are kept for detecting the inflow of toxic substances into water, an imaging device that converts image information of the creatures into electrical signals at regular time intervals, and a lighting device that illuminates the creatures; A creature binarizing means for extracting the living thing as a binary image from a multivalued image captured by the imaging device, a means for recognizing the position of the living thing individually, and a means for calculating the position of the living thing. A method for recognizing the position of a plurality of fish, characterized in that an abnormality in water quality is detected based on the position of the creature in an image monitoring device.