JPH0671599B2 - Image recognition method for activated sludge - Google Patents

Description

Description: TECHNICAL FIELD The present invention relates to an image recognition method for activated sludge, and more particularly to an aeration tank in equipment for purifying wastewater by using the activated sludge method in sewage treatment and the like. The present invention relates to an image recognition processing method for recognizing and evaluating the properties of activated sludge in the soil.

[Conventional technology]

The activated sludge method is a water treatment process used in a wide range of fields such as sewage treatment. The requirements for carrying out the method are to decompose organic pollutants by microorganisms in the presence of oxygen, to ensure the assimilation reaction, and to ensure the effective precipitation separation of the activated sludge flocs produced. Due to fluctuations, a situation in which good precipitation separation cannot always be ensured sometimes occurs, and it has been regarded as one of the water treatment processes in which conventional operation management is relatively difficult.

In order to ensure good sedimentation separation, the activated sludge flocs formed must be dense and have a large particle size, that is, a high sedimentation rate. In the past, in general (macroscopically), sedimentation was evaluated using indicators such as SVI and used as information for operation management. However, since the SVI index itself is a result of manual analysis, it lacks real-time performance, and there is no clear threshold for that value, and it is empirical to deal with the quality of precipitation separation. It had a character that made it difficult to automate and go online, such as requiring judgment intervention.

Against this background, in recent years, attempts have been made to capture an enlarged image of activated sludge with a microscope or the like into an image processing device using a television camera and perform processing to obtain microscopic information on the activated sludge. . However, these were on the one hand the recognition counts of protozoa and on the other hand the recognition counts of filamentous bacteria that interfere with sedimentation and did not deal with the denseness of the activated sludge flocks themselves.

[Problems to be Solved by the Invention]

As described above, in the prior art, no technique has been developed for handling the denseness of the activated sludge flocs.

Therefore, an object of the present invention is to provide an image recognition method for quantifying and evaluating the denseness of the activated sludge flocks and enabling stable operation control of the activated sludge flocs.

[Means for Solving the Problems] The inventors of the present invention have earnestly studied on the basis that the factor controlling precipitation separation is not only the amount of filamentous bacteria but also the denseness of the activated sludge flocks themselves. As a result of research, the inventors have invented a new image recognition processing method capable of quantitatively expressing denseness and achieved the above object.

That is, the present invention relates to a method of enlarging and capturing an image of activated sludge and performing image processing, (1) a first step of calculating a difference between a volatility value of original image data and a luminance value of background image data, (2) A second step of binarizing the image obtained in the first step;
And (3) after performing line detection processing on the image obtained in the first step,
Third step of binarization processing, (4) Fourth operation of calculating a sum of luminance values of the image data obtained in the second step and the image data obtained in the third step to obtain a mask image (5) A fifth step of obtaining the emphasized image by calculating the sum of the brightness values of the image data obtained in the first step and the image data obtained in the fourth step, (6) ) A sixth step of binarizing the image obtained in the fifth step, and (7) Reduction and enlargement of the image obtained in the sixth step to obtain only the image of the activated sludge flocs. 7th step of obtaining, (8) 8th step of integrating the number of pixels of the image obtained in the 7th step, and calculating the area and / or volume of the activated sludge flocks, (9) the 1st step A ninth step of binarizing the image obtained in the step with a plurality of brightness values, and (10) reduction processing and enlargement of the image obtained in the ninth step. (11) For each of the plurality of images obtained in the tenth step, the number of pixels is integrated to calculate the area and / or volume of the activated sludge flocks, and (12) The ratio of the area of the activated sludge flocs obtained in the 11th step to the area of the activated sludge flocs obtained in the 8th step, or the activated sludge flocs obtained in the 11th step. For each of the volumes, the 12th step of calculating at least one of the ratio with the volume of the activated sludge flocs obtained in the 8th step, and (13) the activated sludge by the calculated value obtained in the 12th step. A thirteenth step for judging the density of flocs, and an image recognition method for activated sludge, which comprises:

In the above, the activated sludge is a generic term for those formed by floc-forming bacteria, filamentous bacteria and protozoa, and the floc-shaped target formed by the floc-forming bacteria is referred to as activated sludge floc. The filamentous bacterium is a filamentous bacterium that is independently released or radially extends from the floc, and in the present invention, the activated sludge floc is quantitatively recognized.

[Action]

In the image recognition method according to the present invention, an enlarged image of an activated sludge floc generated in an activated sludge process such as sewage treatment is obtained.
By performing a predetermined image arithmetic processing, it is possible to quantify and evaluate the denseness, thereby diagnosing the quality of the activated sludge flocks, effective operation management,
It acts to enable automatic control.

〔Example〕

Hereinafter, the present invention will be described in more detail with reference to the drawings, but the present invention is not limited to the embodiments.

First, description will be made with reference to FIG. FIG. 1 is an explanatory diagram showing an operation example of an image recognition processing method for activated sludge according to the present invention.

The raw water 1 is introduced into the aeration tank 2 after undergoing a pretreatment if necessary, and is mixed there with the activated sludge 12 returned from the settling tank 4. Air is supplied from the blower 3 into the aeration tank 2 in order to supply oxygen necessary for the activity of the activated sludge. The activated sludge decomposes and assimilates organic substances in the wastewater while staying in the aeration tank. Outflow water from the aeration tank 2 is sent to a settling tank 4, where activated sludge is separated by settling to obtain clear treated water 7. A part of the activated sludge that has settled is returned sludge 12, and the rest is discharged outside the system as excess sludge 13.

An underwater microscope 8 suitable for obtaining an enlarged image of activated sludge is immersed in the aeration tank 2, and the image of activated sludge is periodically captured and transmitted to the image recognition system 9 according to the present invention. . As the underwater microscope 8, for example, a device suitable for this purpose is selected, as described in Japanese Patent Application No. 62-279056 previously filed by the present applicant.

The image recognition processing system 9 is composed of an image processing unit, a computer, a monitor, a television, etc.
Necessary arithmetic processing is performed on the video signal from the above to calculate an index for evaluating and determining the density of the activated sludge flocs. The calculation result in the system 9 is transmitted through the controller 10,
The aeration blower 3, return sludge pump 5, and excess sludge pump 6 are controlled as needed.

Next, the procedure of processing image data in the image recognition system 9 will be described with reference to the process diagram of FIG. 2 showing an example of the flow of the image recognition processing method of the present invention.

The original image data 21 input from an underwater microscope, etc., is smoothed by multiple integration inputs if necessary, and does not include the object (activated sludge flocks, etc.) in order to remove illumination unevenness etc. of the input image. A difference 23 between the brightness value and the background image data 22 is calculated (first step). Next, the image obtained in the first step is subjected to image processing in accordance with two processing procedures (A) and (B).

(A) In the processing of the route, the image obtained in the first step is binarized by a predetermined luminance value for the image obtained in the first step (second step), and the image obtained in the first step. Image obtained by performing line detection processing 25 using the Laplacian operator to emphasize the intermediate luminance part (filamentous bacteria, peripheral area of activated sludge flocs, etc.) and then performing binarization processing 26 (third step) And the sum 27 of the brightness values is calculated (step of FIG. 4).

Here, the size of the Laplacian operator is 3 × 3 matrix or more, preferably 9 × 9 matrix,
The weighting coefficients are preferably arranged in about 12 kinds so that the weighting coefficient can be emphasized regardless of the direction of a line segment having a random direction.
After performing filtering processing using 12 types of operators, a logical sum operation is performed to obtain the brightness value of each pixel.
An image serving as a mask is created by taking the sum of the obtained image and the image obtained in the second step. If the sum of the obtained brightness values exceeds the maximum brightness value (255 in the case of 8 bits), the maximum brightness value is set.

Next, the sum of the brightness values of this mask image and the image obtained in the first step is calculated by the same operation as in the fourth step.
29 to obtain an emphasized image (fifth step). At this time, the mask image is inverted in black and white 28, and the brightness is converted by using all the brightness values of the part without the object after the combination as the background. By binarizing 30 the image obtained in the fifth step at a predetermined level (sixth step), noise, uneven illumination, etc. are eliminated, and unclear filamentous bacteria and rings around the flocks are removed. It is possible to obtain an accurate binarized image of the original image in which spaces and the like are clarified.

Then, in the route (A), the filamentous bacteria are removed from the image and only the blocks are extracted, so that the reduction process and the enlargement process 31 are performed the necessary number of times (seventh step). This number is
Considering the expansion factor of the sample, the thickness of the filamentous bacteria, and the like, the number of times necessary and sufficient for removing them may be performed. The number of pixels of the obtained image is integrated and the area of the activated sludge flocs is calculated 32 (eighth step).

This may be calculated, for example, based on the number of pixels in the block part and the unit pixel area (area of a square formed by four pixels), and the number of pixels in the block part with respect to the total number of pixels on the screen may be calculated. It may be calculated as a ratio.

Next, the processing procedure of (B) route will be described.

In the route (B), the image obtained in the first step is binarized 33 with a plurality of brightness values (the ninth step).
The setting of the brightness value of the binarization process may be arbitrarily set based on the knowledge about the distribution of the brightness value of the activated sludge.
For example, the maximum diameter is calculated for the target activated sludge flocks, the maximum luminance value and the minimum luminance value are obtained from the luminance value distribution obtained by scanning along the maximum radial direction, and the maximum luminance value and the minimum luminance value are calculated. It is preferable to set as a value and at least one luminance value in between. As a result, a plurality of binarized images having different brightness values in the binarization process can be obtained.

For each of the plurality of binarized images, in order to remove filamentous bacteria and the like, the image is reduced and enlarged as many times as necessary.
Go to step 34 to extract only the activated sludge flocs (10th step). The number of times may be sufficient and sufficient to remove the sample in consideration of the magnification of the sample, the thickness of the filamentous bacteria, and the like. For each of the extracted images,
The area of the activated sludge flocs is calculated and measured 35 by the same operation as in the eighth step (11th step).

Finally, for the multiple images obtained by route (B),
For each calculated value related to the block area,
The ratio 36 to the calculated value of the block area obtained by the route (A) is calculated (12th step). Using the obtained ratio value, the degree of dispersion / aggregation of the activated sludge flocs, that is, the denseness is determined (thirteenth step).

The above contents will be described more specifically with reference to FIG. FIG. 3 is a process diagram showing the principle of the image recognition processing method of the present invention.

Sample S1 is an activated sludge floc in which microorganisms are sparsely aggregated from the periphery to the center of the floc, and sample S2 is an activated sludge floc that is densely aggregated.

FIG. 3-1 is a schematic diagram showing a cross section of a microscope slide whose both sides are narrowed by glass or the like. The brightness obtained by scanning a cross section of these samples (for example, a cross section in the maximum radial direction as described above) The value distribution curve is shown in Figure 3-2. In this way, the activated sludge flocs can be grasped as an object having a ring with a high luminance value from both the central portion toward the peripheral portion, and how to process the intermediate luminance value in the peripheral portion. Therefore, it depends on whether or not the target object can be accurately recognized.

Therefore, in the route (A), an intermediate luminance value corresponding to this peripheral portion is subjected to an emphasis process using a Laplacian man filter so that the peripheral ring gap is attached to obtain an accurate image of the target. The procedure to obtain is adopted. That is, the Flock image obtained by the procedure of route (A) is obtained as the image closest to the true image.

Next, processing is performed according to the procedure according to the route (B). As shown in Fig. 3-2, the image of the flock can be binarized by the brightness values A to D because the degree of sparseness and denseness can be obtained as the difference in the brightness value distribution obtained by scanning the flock cross section. When the processing is performed, the area of the flocks is differently calculated as shown in FIG. 3-3. FIG. 3-4 shows plots for each luminance value as a ratio with respect to the block area obtained by calculating these areas according to the procedure of the route (A). In other words, the sparser activated sludge flocs inside have a greater change in the area of the extracted flocks due to the difference in the brightness value of the binarization process, and the density of the flocs can be determined by using the degree of this change as an index. become.

When using volume instead of area, it is sufficient to add the thickness between the preparations to the obtained area.

For example, as shown in FIG. 3-4, the degree of change may be regarded as a straight line (calculation of the least-squares correlation straight line), and a method of obtaining the inclination may be adopted. That is, the sample S2 represents that the change in the recognition rate of the area or the volume due to the difference in the brightness value is smaller horizontally than that of the sample S1, that is, that the degree of microbial aggregation is dense up to the inside of the flock, It is understood that the denseness of the flocks can be evaluated by this method.

It is possible to accumulate a large amount of such data, and refer to the database each time the observation is performed to determine the block state.

〔The invention's effect〕

As described above, according to the present invention, it becomes possible to quantify and evaluate the denseness of the activated sludge flocks, which cannot be performed conventionally, and the effect of achieving stable operation control of the activated sludge process is produced.

[Brief description of drawings]

FIG. 1 is a schematic explanatory view showing an operation example of the image recognition processing method for activated sludge of the present invention, FIG. 2 is a process chart showing an example of the flow of the present invention, and FIG. 3 shows the principle of the present invention. It is a process drawing. 1 ... Raw water, 2 ... Aeration tank, 3 ... Aeration blower, 4 ...
Sedimentation tank, 5 ... Return sludge pump, 6 ... Surplus sludge pump, 7 ... Treated water, 8 ... Underwater microscope, 9 ... Image recognition system, 10 ... Controller, 11 ... Sedimentation sludge, 12 ...
… Returned sludge, 13 …… Excess sludge,

Claims (1)

[Claims] 1. A method of enlarging and capturing an image of activated sludge and performing image processing, comprising: (1) a first step of calculating a difference between a luminance value of original image data and a luminance value of background image data; and (2) a first step. The second step of binarizing the image obtained in the step of
And (3) after performing line detection processing on the image obtained in the first step,
Third step of binarization processing, (4) Fourth operation of calculating a sum of luminance values of the image data obtained in the second step and the image data obtained in the third step to obtain a mask image (5) A fifth step of obtaining the emphasized image by calculating the sum of the brightness values of the image data obtained in the first step and the image data obtained in the fourth step, (6) ) A sixth step of binarizing the image obtained in the fifth step
Step (7), a seventh step of reducing and enlarging the image obtained in the sixth step to obtain only an image of the activated sludge floc, and (8) the image obtained in the seventh step 8th step of integrating the number of pixels of and calculating the area and / or volume of the activated sludge flocks, and (9) the 9th step of binarizing the image obtained in the 1st step with a plurality of luminance values. And (10) a tenth step of reducing and enlarging the image obtained in the ninth step, and (11) the number of pixels for each of the plurality of images obtained in the tenth step. The eleventh step of integrating and calculating the area and / or volume of the activated sludge flocks, and (12) the activity obtained in the eighth step for each of the areas of the activated sludge flocs obtained in the eleventh step. For each ratio to the area of sludge flocs or the volume of activated sludge flocs obtained in the 11th step, Then, the twelfth step of calculating at least one of the ratio with the volume of the activated sludge flocs obtained in the eighth step, and (13) the denseness of the activated sludge flocks based on the calculated value obtained in the twelfth step 13. A method for recognizing activated sludge, which comprises a thirteenth step of determining