JPS6279330A - Apparatus for recognizing floc image in water purifying plant - Google Patents

Abstract

PURPOSE:To enhance the recognition accuracy of a floc image, by a method wherein an irradiation apparatus to a floc group is provided in the floc forming basin of a water purifying plant and the number of flocs and the average area thereof are calculated when the obtained image information is binarized and a values, when a max. average area is obtained for the max. number of flocs, is set as a threshold value. CONSTITUTION:A hermetically closed container 20 having a TV camera 30 mounted therein and an illumination apparatus 40 for a floc group are sunk is a floc forming basin 10 and the image information obtained by the camera 30 is sent to an image recognition control apparatus 60 through the image recognition processing part 50 provided outside the floc forming basin 10. In this constitution, an automatic threshold value determining apparatus 90 is provided in the processing part 50 and the output therefrom is binarized by a binarizing circuit 103 through a threshold value input circuit 102 and processing is subsequently performed by a particle size measuring circuit 106 and a particle size comparing circuit 107 while the processed value is sent to a particle size distribution operational display part 100. Thereafter, a particle size is calculated by the display part 100 and the threshold values of the max. average area with the max. number of flocs are obtained to form a highly accurate floc image.

Description

DETAILED DESCRIPTION OF THE INVENTION [Field of Application of the Invention] The present invention relates to a flock image recognition device for a water purification plant, and particularly to a flock image recognition device that automatically determines a threshold value for binarizing a flock image.

[Field of application of the invention]

At water treatment plants, since the suspended particles in raw water are small in size, the process is to agglomerate them into flocs, and then let the flocs settle. For this reason, it is essential to monitor flocs in the floc formation pond (mixing pond).

Conventionally, flocs have been visually monitored several times a day by water treatment plant maintenance managers. Since it relies on visual inspection, the judgment criteria are subjective and qualitative, and the monitoring results have the disadvantage of being difficult to reflect in driving operations.

Furthermore, since the monitoring frequency is discontinuous, countermeasures against agglomeration failures are delayed, which increases troubles.

In contrast, recently industrial television cameras (ITV)
A method has been adopted in which the flocs in the floc formation pond are monitored using However, even in this case, monitoring remains subjective and discontinuous because it relies on human vision.

In order to improve these drawbacks, Japanese Patent Application Laid-Open No. 54-1432
As described in No. 96, a method has also been proposed in which a photoelectric conversion device is used to extract an electrical signal according to the shape of the floc. However, if there are changes in water quality (turbidity), changes in ambient light, etc., it is difficult to stably extract the size of flocs, and unless these problems are solved,
Unmanned monitoring using only monitoring equipment cannot be put to practical use.

[Purpose of the invention]

An object of the present invention is to provide a floc image recognition device that increases the accuracy of floc image recognition by appropriately setting a binarization threshold when recognizing floc groups in a floc formation pond (mixing pond). It is.

[Summary of the invention]

In the present invention, the threshold value for binarization is automatically determined according to the number of flocs and the average area of the flocs in the binarized image.

The details will be explained below.

Generally, when performing image processing using an image processing device,
The problem is how to set the threshold for binarization.

Fixed binarization can be considered, in which a threshold value is set by comparing the grayscale image recognized by the underwater camera and the binary image, but in the case of fixed binarization, after setting the threshold value, the ambient light of the underwater camera is If there is a change in turbidity or a change in turbidity, 2.
Value images may not be binarized well when compared to grayscale images, and unless the threshold settings are changed, correct image processing calculations will not be performed. The present invention automatically determines the threshold value for binarization by determining the number of objects and the average area of the objects in the binary image, so that correct image processing operations can always be performed. be.

[Embodiments of the invention]

Figure 2 shows a diagram of the floc formation pond at the water treatment plant. The floc formation pond 10 has three floc formation ponds 10A and IOB.
, IOC, and the inflow water 10a containing minute flocs flowing from the rapid mixing ponds indicated by arrows flows into each floc formation pond 1.
It sequentially passes through 0A-+10-+10C and flows out as outflow water 10b.

In each of the floc formation ponds 10A to 10C, a stirring paddle I
It has IA to 11C, stirs the inflowing flock group, and coagulates the flocs by colliding with each other. A rectifying wall 13A is provided between the flocculation ponds 10A and IOB, and a rectification wall 13B& is provided between the flocculation ponds 10B and 10C to effect rectification. Each time the flocs 12 pass through each floc formation pond 10A→IOB→10C, the particle size gradually increases.

To monitor the flocs, use one of the floc formation ponds 10A to 10. A flock imaging device including a TV camera is installed in C. Flock formation pond 10A ~ IOC
Where should I install it?

Depends on the purpose of floc recognition. If the purpose is to monitor the final degree of floc formation, it is installed in the final floc formation tank 10C (i.e., before the settling tank), and if the purpose is to monitor the process of floc formation, it is installed in the floc formation tank 10A or IOH. .

FIG. 3 shows an embodiment of a flock image recognition device including a flock imaging device. In FIG. 3, an airtight container 20, a TV camera 30 provided inside the container 20, and a lighting device 1ff40 are installed in the floc formation pond 10.

A priest window 21 is provided in the front part of the airtight container 20, and a wiper 22 for preventing the flock from adhering to the window 21 is provided in front of the window 21. The wiper 22 is driven by a motor, but the mechanism is omitted in the figure. The back screen 24 is a support member 24A supported by the airtight container 20.

24B, this screen 24 is black and serves to improve the contrast for imaging the flocs present in the space between the AM window 21 and the screen 24 with the TV camera 30 with good visibility. In particular, the color of the flock 12 is white, and by making the back screen 24 black, the contrast of the flock 12 when viewed from the TV camera 30 becomes extremely high.

Illumination device 4o as an auxiliary means to increase contrast,
and a light shielding cover 42. The reason for this is as follows.

The floc formation pond 10 is opened to the atmosphere for the purpose of constantly monitoring flocs. Therefore, the amount of light incident on the floc formation pond 10 changes over time and is strongly influenced by the weather. A lighting device 40 is usually installed to constantly monitor the flock. If the purpose is simply monitoring that relies on the eyes of a maintenance manager, it is sufficient to install the lighting device 40, and it goes without saying that the lighting device 40 may be installed above the floc formation pond 10.

However, even if the illumination device R40 is installed, changes in ambient illuminance strongly affect the accuracy of image recognition of the flock group. For example, if the illuminance is low, the flocs will be perceived as small, and conversely, if the illuminance is high, the flocs will be perceived as large. In order to eliminate this influence, it is necessary to avoid being influenced by changes in illuminance as a natural phenomenon.

In this embodiment, a light-shielding cover 42 is provided to darken the surroundings, and the illuminance is provided under a certain condition only by the illumination device 40.

The ITV 30 fixed in the airtight container 20 is a close-up lens 3
1, an observation window 21 made of transparent material such as glass
The image of the flocs 12 in the floc formation pond 10 is enlarged and recognized through the lens. The wiper 22 operates periodically to remove dirt from the surface of the observation window 21.

Note that baffle plates 25A, 25B, and 25C were installed to prevent light scattering. Furthermore, T of the pack screen 24
wtI from V left camera 0! On the third side.

A reference white area for checking the lighting device is provided. While this reference white area is observed with a TV camera, the lighting device 4o is judged to be normal, and if wt cannot be detected, it is judged to be abnormal.

In FIG. 3, the image capture circuit 32, the illumination control circuit 41, the image recognition processing section 5o, and the image recognition processing section [i! ! It is 60. The control circuit 41 instructs the lighting control device 40 to perform lighting, and the TV control circuit 32 takes in the captured image by the TV left camera 0. The output of this circuit 32 becomes a video signal.

FIG. 1 shows an embodiment of an image recognition processing section 50 and an image recognition control device 60 according to the present invention.

The image recognition processing unit 50 extracts information useful for water quality management at the water purification plant from the image information obtained by the TV left camera 0.
A control device 60 that performs various processes such as particle size and distribution of the floc group performs timing control thereof.

In FIG. 1, the image recognition processing section 50 has the following configuration.

Automatic threshold value determination device 90 is a device that automatically determines a threshold value, and receives information for this purpose from the memory 108.

Abnormality detection device 91: Obtains data for determining abnormality (lights out) of the lighting device 40.

y-report output determination unit 92: Determines the presence or absence of an alarm output based on the data obtained by the abnormality detection device 91.

Abnormality alarm unit 93: issues an alarm when an abnormality is determined.

Wiper drive section 94...Performs wiper drive. When an abnormality is detected, the number of times the wiper is driven is increased and the R1 test is performed again to determine whether there is an abnormality or not.
g, do ra.

Recognition timing control unit 100...takes in the video signal from the control circuit 32.

AD converter 101: Performs AD conversion of a video signal.

Threshold value input circuit 102: Takes in and outputs the threshold value determined by the automatic threshold value determination circuit 90.

Binarization circuit 103: Binarizes the multi-value output of the AD converter 101 based on the output threshold of the threshold input circuit 102.

Labeling circuit 104 performs labeling (numbering) of flocks within one side.

Particle size measurement circuit 106...Performs particle size measurement for each floc.

Particle size calculation mode designation circuit 105: Designates the particle size calculation mode.

Particle size comparison circuit 107...Classifies each floc by particle size.

Memory 108: has memory elements 108A to 108Z, and one memory element 1 corresponding to each size corresponds to one grain size.
08A to 108Z and stored. The stored data is a number.

Recognition number storage unit 109: performs image recognition multiple times. At this time, the flock images to be recognized each time are flock images captured in a time-division manner within a certain time period.

Particle size distribution calculation/display section 100...Calculates and displays the particle size distribution of flocs.

In FIG. 1, the recognition timing control circuit 100
The time interval (frequency) at which image information of the flock image obtained through the ITV 30 and ITV controller 32 shown in the figure is captured is controlled. Next, the A/D conversion circuit 101 receives the obtained analog signal of luminance information, for example, the image signal of FIG. 4, and converts the signal into a digital signal one by one.

Figure 4 shows the image plane of the floc group recognized by the ITV30.Since the floc group is a grayscale image, the boundary between the floc 12 and the water is not clear in reality, but for the sake of simplicity, the circle of the floc group is Only parts are shown. The brightness level of the flock group is high and white, while the back screen 2 is in the background.
4, the brightness level of the water is low and black.

The converted digital signal is binarized in a binarization circuit 103 based on a threshold specified by a threshold input circuit 102.

For example, FIG. 5 shows a case where the screen of FIG. 4 is scanned along the AA' line to display the luminance level distribution. Here, the brightness level is displayed in 8 bits (256 levels).
The brightness is low toward the top of the vertical axis, while the brightness is high toward the bottom.

Since the flock 12 is white, the luminance is high. In other words, the downwardly valleyed portion represents the floc.

In this brightness distribution, each pixel is binarized by a binarization circuit 103 based on a threshold specified by a threshold input circuit 102, for example, the brightness specified by the BB' line. The threshold value specified by the threshold value input circuit 102 is maintained constant under constant illuminance, but can also be operated by an operator.

In the binarization circuit 103, a pixel at a brightness level higher than a threshold value is set to 1'1'', while a pixel at a brightness level higher than a predetermined value is set to 'O''. Then, as shown in FIG. 6, the portion corresponding to floc becomes "1" and the portion corresponding to water becomes 110 II.

FIG. 7 shows an example of the results of floc recognition performed in this manner.

Once the flocs are recognized, the representative grain size of the flocs is calculated, but before that a number is assigned to each floc. That is, the labeling circuit 104 recognizes each flock independently and assigns a number to each flock. Then, the particle size measuring circuit 106 calculates the representative particle size of each floc in the order of its number.

Typical particle sizes include directional diameter Dc, maximum diameter, and large diameter Dm&K.

The single grain size calculation mode designation circuit 105, which has a minimum diameter Dmin, an area circle equivalent diameter Dctr, an equivalent peripheral length DC1, etc., designates a representative grain size to be adopted from among these representative grain sizes. The particle size of each floc is calculated in accordance with the standard of the representative particle size specified in this way.

The particle size comparison circuit 107 compares the particle sizes of each floc and stores the number of flocs having each particle size in a corresponding storage location, that is, a number storage circuit 108A.

The numbers are stored in 108B, 108C, and 108z. Since the image of the flocs is binarized, the minimum unit for measuring the size of one particle is one pixel, which is predetermined on the CRT. Therefore, if the number of flocs corresponding to each grain size is N1, then, for example, the number storage circuit 108A stores the number Nl of floes whose grain size corresponds to one pixel, and the number storage circuit 108B stores the number Nl of floes whose grain size corresponds to one pixel. The number N2 of flocks corresponding to the number N2 is stored. The minimum unit of particle size is 1
However, whether the minimum unit is two pixels or not depends on the purpose of floc monitoring, the condition of the floc formation pond, and the content of the image captured by the TV camera.

By the way, in order to accurately determine the particle size distribution of flocs,
It is better to capture the screen multiple times and obtain the particle size distribution from a large amount of information. The number of times the flock image is recognized is specified in the recognition number control circuit 109, and if the number is less than this number, the process returns to the recognition timing control circuit 100. The recognition timing control circuit captures the image at the specified timing, repeats the operations described above, calculates the particle size of the flocs, and stores the number of each floc in the number storage circuits 108A, 108B, 108C, -1087. do.

On the other hand, when the number of recognition times control circuit 109 reaches a specified number of times, the particle size distribution calculation and display circuit 110 outputs number storage circuits 108A, 108B, and 108G.

...Based on the value of 108Z, calculate the number concentration distribution for each particle size and output the result.

Next, automatic threshold value determination, which is a feature of the present invention, will be explained.

The automatic threshold value determination is performed by operating the above-described image recognition processing unit 5° in an automatic threshold value determination mode. In order to determine the automatic threshold value, the threshold value is changed one after another with respect to the reference image captured by the TV camera 30, and the relationship between the floc particle size and number is determined for each threshold value. Next, this relationship is searched for each threshold value, and a threshold value determination process is performed to determine the optimal threshold value.

In this automatic determination operation, the automatic threshold value determination circuit 9o changes the threshold value and sends this to the binarization circuit 103 via the input circuit 102. The relationship between the floc particle size and the number of flocs for each threshold value is stored in the memory 108. The search for the contents of the memory 108 and the optimum threshold value determination process are carried out by the automatic threshold value determination circuit 9.
0 does.

The threshold value determination method will be explained more specifically.

To simplify the explanation, consider the case where the grayscale image shown in FIG. 8 is binarized. The flock has a spherical shape,
Therefore, the center of the flock has the highest brightness level, and the brightness distribution resembles contour lines.

The brightness level spreads out from the center and becomes lower outside.

To simplify the explanation, eight equal levels l1A201~
Considering 208, equal level, ff1201 is level 4, equal level line 202 is level 5, equal level, 11203 is level 6, equal level line 204 is level 7, equal level line 20
5 to level 81 equal level [206 to level 9, equal level 4!207 to level 10, equal level line 208 to level 2
shall be. C and D overlap with flock. Next, if the threshold value for binarization is set to 1 and those with a brightness level of 1 or more are considered to be flocks, the grayscale image 200 in FIG. 8 will be binarized like the binary image 210 in FIG. 9. Ru. The shaded part is 2
This is the part that is judged to be white when converted into a value. Next, when the binarization threshold is selected as 2, the grayscale image 2oO in FIG.
The image is binarized like a binary image 210 in the figure. If the binarization threshold is set to 3, the grayscale image 200 in FIG.
It is binarized as shown in the binary image 211 in Figure 0, and the backboard (water) and flock can be distinguished. Next, even when the binarization threshold is selected as 4, the grayscale image 200 in FIG.
It is binarized like the binary image 211 in Figure 0. Next, when the binarization threshold is selected as 5, the grayscale image 200 in FIG.
The image is binarized like a binary image 212 in FIG. Next 2
When the value threshold is selected as 6, the grayscale image 200 in FIG.
is binarized like a binary image 213 in FIG. Next, when the binarization threshold is selected as 7, the grayscale image 2 in Figure 8
00 is binarized like the binary image 214 in FIG.
The number of objects is two. Next, when the binarization threshold value is selected to be 8, the grayscale image 200 in FIG. 8 is binarized like the binary image 215 in FIG. 14. Next, set the binarization threshold to 9
If selected, the grayscale image 200 in FIG. 8 is binarized like the binary image 216 in FIG. 15. Next, when the binarization threshold is selected as 10, the grayscale image 200 in FIG.
It is binarized as shown in the binary rain image 217 in FIG. 6, and the number of flocs is reduced to one. Next, when the binarization threshold is selected to be 11 or more, the grayscale image 200 in FIG.
It is binarized as shown in the value image 218, and the number of flocs becomes 0. As explained above, when the binarization threshold is changed,
The number of objects (flocks) and the average area of the objects vary. First, looking at the number of objects, the number of objects increases as the threshold value is raised, but decreases after a certain threshold value is exceeded, and there is a range of threshold values in which the number of objects is maximum. Also, if we pay attention to the average area of objects, the average surface area decreases as the threshold value is increased. Furthermore, among the binary images 210 to 218 in FIGS. 9 to 17, if we consider the binary image that is accurate as the information that the most flocs exist, the number of flocs is 2, and the number of flocs is 2. 2 becomes 2
It is considered that the binary image 214 in FIG. 13, which has the largest average area among the individual screens, is accurate.

Therefore, the threshold value for binarization is changed one after another via the input circuit 102, the number of flocs is determined for each threshold value, and the area of each floc is determined, and the results are stored in the memory 108 to automatically determine the threshold value. The circuit 90 searches the contents of the memory 108, issues a threshold value when the number of flocks is maximum and the area of the flocks is maximum, and determines this as the official threshold value.

The determination mechanism will be explained with reference to FIGS. 18(a) to 18(c). In FIG. 18(a), both the horizontal and vertical axes are expressed as the threshold value V t
18(b) shows the threshold value Vth on the horizontal axis and the number M of flocs on the vertical axis, and FIG. 18(c) shows the threshold value Vth on the horizontal axis and the flocs on the vertical axis. Average area S
shows. The threshold value Vih takes a value from 0 to 127. In FIG. 18 (b), section A is the section where the number of flocs is maximum, and in this section A, threshold value Vz≦Vth≦Vz
This is an interval in which the same maximum number occurs consecutively. on the other hand.

In FIG. 18(c), the average area S of flocs for each threshold value Vth is calculated and displayed. This area curve S,
Vt v Vz (1) intersection at 81. When Sl is S, )St, Sl is the maximum average area in section A.

Therefore, the optimal threshold value V th to be found is V th
" V 5 (=Pz). After determining the optimum threshold value, the automatic threshold value determination mode is switched to the floc recognition mode, and floc recognition is performed according to the process shown in FIG.

The automatic threshold value determination mode may be activated in advance, or may be activated as appropriate during the floc formation process.

Incidentally, the automatic threshold value determination circuit 90 itself has the functions of the circuits 100 to 100.
You may have 108. In that case, it is possible to independently automatically determine the threshold value.

〔Effect of the invention〕

According to the present invention, by appropriately determining the threshold value, it is possible to prevent deviations in the binarization threshold setting due to changes in the ambient light of the underwater camera, changes in turbidity, etc., and highly reliable image processing is possible. Enables continuous unattended processing.

[Brief explanation of drawings]

Fig. 1 is an embodiment of the present invention, Fig. 2 is a block diagram of a floc formation pond to which the present invention is applied, Fig. 3 is an overall block diagram of the present invention for a floc formation pond, and Fig. 4 is a floc image. Figures 5 and 6 are explanatory diagrams of binarization, Figure 7 is an example of a binary image, Figure 8 is a diagram of the relationship between flocs and isoline levels, and Figures 9 to 17 are FIG. 18, an example of a binary image obtained by changing the threshold value, is an explanatory diagram of determining the optimum threshold value. 50... Image recognition processing unit, 60... Image recognition control device, 90... Automatic darkness value determination circuit.

Claims (1)

[Claims] 1. An illumination device installed in a water treatment plant floc formation pond for the floc group, and image information of the floc group that received light from the illumination device are taken in and a binarized image is obtained using a threshold value, and the binarized image is The purified water floc image consists of an image recognition processing unit that performs processing and floc recognition, and when determining a threshold value, the number and average area of flocs are determined, and the value when the maximum number of flocs results in the maximum average area is used as the threshold value. recognition device.