CN112767362A - Sludge bulking prediction method based on activated sludge phase difference microscopic image - Google Patents

Sludge bulking prediction method based on activated sludge phase difference microscopic image Download PDF

Info

Publication number
CN112767362A
CN112767362A CN202110088840.8A CN202110088840A CN112767362A CN 112767362 A CN112767362 A CN 112767362A CN 202110088840 A CN202110088840 A CN 202110088840A CN 112767362 A CN112767362 A CN 112767362A
Authority
CN
China
Prior art keywords
image
sludge
activated sludge
phase difference
phase
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Pending
Application number
CN202110088840.8A
Other languages
Chinese (zh)
Inventor
赵立杰
刘金池
左越
邹世达
刘健
范文玉
黄明忠
张宇红
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Shenyang University of Chemical Technology
Original Assignee
Shenyang University of Chemical Technology
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by Shenyang University of Chemical Technology filed Critical Shenyang University of Chemical Technology
Priority to CN202110088840.8A priority Critical patent/CN112767362A/en
Publication of CN112767362A publication Critical patent/CN112767362A/en
Pending legal-status Critical Current

Links

Images

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T7/00—Image analysis
    • G06T7/0002—Inspection of images, e.g. flaw detection
    • G06T7/0004—Industrial image inspection
    • G—PHYSICS
    • G01—MEASURING; TESTING
    • G01B—MEASURING LENGTH, THICKNESS OR SIMILAR LINEAR DIMENSIONS; MEASURING ANGLES; MEASURING AREAS; MEASURING IRREGULARITIES OF SURFACES OR CONTOURS
    • G01B11/00—Measuring arrangements characterised by the use of optical techniques
    • G—PHYSICS
    • G01—MEASURING; TESTING
    • G01B—MEASURING LENGTH, THICKNESS OR SIMILAR LINEAR DIMENSIONS; MEASURING ANGLES; MEASURING AREAS; MEASURING IRREGULARITIES OF SURFACES OR CONTOURS
    • G01B11/00—Measuring arrangements characterised by the use of optical techniques
    • G01B11/24—Measuring arrangements characterised by the use of optical techniques for measuring contours or curvatures
    • G—PHYSICS
    • G01—MEASURING; TESTING
    • G01N—INVESTIGATING OR ANALYSING MATERIALS BY DETERMINING THEIR CHEMICAL OR PHYSICAL PROPERTIES
    • G01N21/00—Investigating or analysing materials by the use of optical means, i.e. using sub-millimetre waves, infrared, visible or ultraviolet light
    • G01N21/84—Systems specially adapted for particular applications
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T5/00—Image enhancement or restoration
    • G06T5/50—Image enhancement or restoration using two or more images, e.g. averaging or subtraction
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T5/00—Image enhancement or restoration
    • G06T5/70—Denoising; Smoothing
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T7/00—Image analysis
    • G06T7/10—Segmentation; Edge detection
    • G06T7/11—Region-based segmentation
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T7/00—Image analysis
    • G06T7/10—Segmentation; Edge detection
    • G06T7/136—Segmentation; Edge detection involving thresholding
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T7/00—Image analysis
    • G06T7/60—Analysis of geometric attributes
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T2207/00—Indexing scheme for image analysis or image enhancement
    • G06T2207/10—Image acquisition modality
    • G06T2207/10024—Color image
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T2207/00—Indexing scheme for image analysis or image enhancement
    • G06T2207/10—Image acquisition modality
    • G06T2207/10056—Microscopic image
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T2207/00—Indexing scheme for image analysis or image enhancement
    • G06T2207/20—Special algorithmic details
    • G06T2207/20024—Filtering details
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T2207/00—Indexing scheme for image analysis or image enhancement
    • G06T2207/20—Special algorithmic details
    • G06T2207/20036—Morphological image processing
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T2207/00—Indexing scheme for image analysis or image enhancement
    • G06T2207/20—Special algorithmic details
    • G06T2207/20048—Transform domain processing
    • G06T2207/20052—Discrete cosine transform [DCT]
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T2207/00—Indexing scheme for image analysis or image enhancement
    • G06T2207/30—Subject of image; Context of image processing
    • G06T2207/30108—Industrial image inspection

Landscapes

  • Engineering & Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • General Physics & Mathematics (AREA)
  • Theoretical Computer Science (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Geometry (AREA)
  • Health & Medical Sciences (AREA)
  • Quality & Reliability (AREA)
  • Chemical & Material Sciences (AREA)
  • Analytical Chemistry (AREA)
  • Biochemistry (AREA)
  • General Health & Medical Sciences (AREA)
  • Immunology (AREA)
  • Pathology (AREA)
  • Activated Sludge Processes (AREA)

Abstract

The invention discloses a sludge bulking prediction method based on an activated sludge phase difference microscopic image, and relates to sewage treatment intelligent detectionThe field of the technology. The specific implementation scheme comprises image acquisition, image fusion, image segmentation, feature extraction and dynamic state
Figure DEST_PATH_IMAGE002
And performing model and sludge bulking early warning. Acquiring images, namely acquiring phase difference microscopic images of the activated sludge samples; image fusion, namely fusing different focal plane images of the activated sludge by adopting discrete cosine transform; image segmentation, namely after removing phase difference by adopting a phase consistency method, realizing the segmentation of the activated sludge filamentous fungi and flocs by low-pass filtering, phase stretching transformation, threshold segmentation, starting operation and gyration radius characteristics; and (4) feature extraction, namely extracting morphological feature parameters of flocs and filamentous fungi. The invention fuses morphological characteristic parameter information of the activated sludge image into the volume index of the activated sludge
Figure 896486DEST_PATH_IMAGE002
The detection provides an on-line, reliable and efficient measurement means.

Description

Sludge bulking prediction method based on activated sludge phase difference microscopic image
Technical Field
The invention relates to the field of intelligent detection of sewage treatment, in particular to a sludge bulking prediction method based on an activated sludge phase difference microscopic image.
Background
Sludge bulking is one of the most common and troublesome problems of activated sludge sewage treatment systems. The sludge expansion not only seriously affects the sewage treatment effect, but also even causes the collapse of a sewage biological system, and the sewage biological system can not be recovered for several months. The sedimentation performance of the activated sludge directly influences the water quality and the operation efficiency of effluent of a biological denitrification sewage treatment plant, and the sludge volume index SVI is an important index for evaluating the characteristics and the sedimentation performance of the sludge. Therefore, the temperature of the molten metal is controlled,
Figure 95410DEST_PATH_IMAGE001
prediction is to predict sludge bulking and realize sewage treatmentOn the premise of optimizing the operation of the treatment process, the working efficiency of the sewage plant can be improved, and the water quality index of the discharged water is ensured.
The general SVI conventional detection method relies on laboratory assay analysis, and the whole process is complicated and time-consuming. Aiming at the problems, at present, many scholars at home and abroad research the measurement of the sludge volume index and obtain good effect, for example, people of Beijing industry university Korean cinnamon and the like establish the sludge volume index based on an integrated neural network prediction method (CN 102778548B) and a recursive RBF neural network (CN 107025338B) respectively
Figure 678838DEST_PATH_IMAGE001
Soft measurement model for realizing sludge volume index
Figure 866237DEST_PATH_IMAGE001
The real-time prediction of the concentration also provides a filamentous bacterium sludge bulking index
Figure 163095DEST_PATH_IMAGE001
The characteristic model construction method (CN 103605882A) extracts the growth dynamics characteristics of the filamentous fungi, corrects the model parameters by adopting a data statistical method, and realizes the characteristic model by the related process variables and the filamentous fungi sludge bulking mechanism
Figure 799087DEST_PATH_IMAGE001
The prediction solves the problem that the sludge bulking model is difficult to establish, and can quickly and effectively predict the sludge volume index
Figure 869811DEST_PATH_IMAGE001
The value is obtained.
Sludge microscopic image processing and analysis is another potential tool for effective sedimentation property prediction. In the past, the floc characteristics are researched and the content of the filamentous fungi is estimated mainly by a method of manual technology under a microscope, and the whole process is tedious and time-consuming and depends on subjective experience of operators. With the rapid development of digital image processing and analysis technology and microscope technology, the sludge bulking index change is forecasted in real time by automatically extracting the morphological characteristics of floccules and filamentous bacteria of the activated sludge microscopic image and combining the physical and chemical parameters of sewage, so that the method is an important means for realizing the early monitoring of the activated sludge sedimentation characteristic and the early warning of filamentous bacteria sludge bulking.
The invention provides a sludge bulking prediction method based on an activated sludge phase difference microscopic image, which is characterized in that microscopic images at the same position and different focal planes are fused based on a discrete cosine transform method; the separation of the activated sludge filamentous fungi and flocs is realized based on a phase stretching transformation, edge detection and threshold value separation method; morphological characteristic parameters of the flocs and the filamentous fungi are extracted from the images divided by the flocs and the filamentous fungi; the morphological characteristic parameters are compared with online measurement data MLSS, DO, pH and
Figure 860901DEST_PATH_IMAGE001
combining historical measurement data, adopting a method of randomly configuring a network to construct a dynamic state
Figure 638364DEST_PATH_IMAGE001
A model; obtaining morphological characteristics and measurement data of new sample, and utilizing constructed dynamic state
Figure 801580DEST_PATH_IMAGE001
Early warning model implementation
Figure 31704DEST_PATH_IMAGE001
The accurate prediction is carried out, so that the sedimentation performance of the sludge is judged, and whether the sludge is expanded or not is predicted.
Disclosure of Invention
The invention provides a sludge bulking prediction method based on an activated sludge phase difference microscopic image. Specifically comprises the steps of activated sludge microscopic image acquisition, image fusion of different focal planes, image segmentation of flocs and filamentous fungi, feature extraction, and,
Figure 888802DEST_PATH_IMAGE001
Constructing a dynamic model and early warning of sludge bulking. Mixing the extracted morphological characteristic parameters with online measurement data MLSS, DO, pH and
Figure 520771DEST_PATH_IMAGE001
combining historical measurement data as dynamics
Figure 486847DEST_PATH_IMAGE001
The input and output of the early warning model are
Figure 532163DEST_PATH_IMAGE001
. The invention fuses morphological characteristic parameter information of the activated sludge image into the volume index of the activated sludge
Figure 865055DEST_PATH_IMAGE001
The detection provides an on-line, reliable and efficient measurement means.
The invention adopts the following technical scheme:
a sludge bulking prediction method based on an activated sludge phase difference microscopic image is characterized by comprising the following steps:
[01] step 1: image acquisition: the activated sludge sample is taken from the outlet of an aeration tank of a sewage treatment plant, and when the sludge sample is collected, the instantaneous water sample is collected at the water outlet of an aerobic tank, and then the water quality index is measured in a laboratory. The interval time between the sample collection, the image collection and the water quality index measurement is not more than 3 hours. Scanning a sample on the glass slide from the upper left to the lower right by adopting an optical microscope, an industrial digital camera and an image acquisition system to obtain a plurality of phase difference microscopic images under upper, middle and lower different focal planes;
[02] step 2: image fusion: fusing different focal plane phase difference microscopic images by adopting an activated sludge microscopic image fusion method of discrete cosine transform, and finally, further correcting the fused image by using bilateral filter consistency check;
[03] and step 3: image segmentation: the method comprises two parts of filamentous fungus segmentation and floc segmentation, wherein RGB images (red, green and blue) are converted into gray images after phase differences are removed, Gaussian low-pass filtering is carried out on the gray images, and random noise is reduced through blurred images. The PST converts an image from a spatial domain to a frequency domain by utilizing two-dimensional Fourier transform, then multiplies an obtained frequency domain image by a phase kernel, and converts the obtained image from the frequency domain to the spatial domain by utilizing two-dimensional inverse Fourier transform to obtain an output phase image. And (3) binarizing the output phase image by adopting a threshold value method, and segmenting flocs from the gray level image by adopting the threshold value method in the phase contrast image. After detection of floes, morphological dilation filling was performed. Subtracting the obtained floc mask from the segmented binary image obtained using the PST, leaving only filamentous bacteria in the output image;
[04] and 4, step 4: feature extraction: once the flocs and the filaments in the image are distinguished from the background, morphological characteristic parameters such as the sizes and the shapes of the flocs and the filamentous fungi can be calculated. The extracted morphological characteristic parameters and online detection data MLSS, DO, pH and SVI historical measurement data are combined, so that not only can flocs and filaments be distinguished, but also the monitoring purpose can be achieved, the change of sludge characteristics and the like can be detected, and the change of the sedimentation characteristics of the sludge can be monitored;
and 5: dynamic state
Figure 351531DEST_PATH_IMAGE002
Model: extracting morphological characteristic parameters and online detection data MLSS, DO, pH and
Figure 139707DEST_PATH_IMAGE002
historical data is combined to be used as input to construct a dynamic random configuration network model
Figure 344424DEST_PATH_IMAGE002
An early warning model, output is
Figure 215428DEST_PATH_IMAGE002
A value;
step 6: sludge bulking early warning: carrying out morphological feature extraction and data measurement on a new activated sludge sample, and conveying the sample to the constructed dynamic state
Figure 884307DEST_PATH_IMAGE002
Early warning model, output-based
Figure 520081DEST_PATH_IMAGE002
And (5) judging whether the sludge is expanded or not within the value interval range.
The activated sludge phase difference microscopic image-based sludge bulking prediction method is characterized in that an activated sludge sample is obtained from an aeration tank outlet of a sewage treatment plant through image acquisition, an optical microscope, an industrial digital camera and an image acquisition system are adopted, the sample on a glass slide is scanned from the upper left to the lower right, and a plurality of phase difference microscopic images under upper, middle and lower three different focal planes are obtained.
The activated sludge phase difference microscopic image-based sludge bulking prediction method is characterized in that after edge enhancement sharpening pretreatment is carried out on microscopic images with the same position and different focal planes by an activated sludge microscopic image fusion method based on discrete cosine transform, transformation coefficients of DCT domains of block images are respectively calculated, subblocks with large coefficient variance are selected as subblocks of a fusion image, image fusion is realized through DCT inverse transformation, and finally, a bilateral filter consistency check is applied to further modify the fusion image.
The sludge bulking prediction method based on the activated sludge phase difference microscopic image is characterized in that the image segmentation is realized by respectively segmenting filamentous fungi and flocs. First, the RGB images (red, green, and blue) are converted into grayscale images after phase differences are removed. The grayscale image is gaussian low-pass filtered to reduce random noise by blurring the image. The PST converts an image from a spatial domain to a frequency domain by utilizing two-dimensional Fourier transform, then multiplies an obtained frequency domain image by a phase kernel, and converts the obtained image from the frequency domain to the spatial domain by utilizing two-dimensional inverse Fourier transform to obtain an output phase image. And (3) binarizing the output phase image by adopting a threshold value method, and segmenting flocs from the gray level image by adopting the threshold value method in the phase contrast image. After detection of floes, morphological dilation filling was performed. The resulting mask was subtracted from the split binary image obtained using PST, leaving only filamentous fungi in the output image.
The activated sludge phase difference microscopic image-based sludge bulking prediction method is characterized in that the characteristic extraction is used for sequentially extracting the characteristics of each target of flocs and filamentous bacteria in the image, and the morphological characteristic parameters extracted for each target are as follows:
Figure 8831DEST_PATH_IMAGE003
the activated sludge phase difference microscopic image-based sludge bulking prediction method is characterized in that a dynamic SVI model and a modeling method adopt a dynamic random configuration network model, and model input comprises activated sludge microscopic image flocs and filamentous bacterium morphological characteristic parameters
Figure 388253DEST_PATH_IMAGE005
On-line measurement of data
Figure 100002_DEST_PATH_IMAGE007
And
Figure 537737DEST_PATH_IMAGE009
historical data
Figure 639685DEST_PATH_IMAGE011
The output of the model is
Figure 818994DEST_PATH_IMAGE013
。
Figure 359696DEST_PATH_IMAGE015
The dynamic SCN model is expressed as
Figure 924539DEST_PATH_IMAGE016
Wherein
Figure 952714DEST_PATH_IMAGE018
The sludge bulking prediction method based on the activated sludge phase difference microscopic image is characterized in that the sludge bulking early warning is carried out, a new activated sludge sample is obtained, and the activated sludge sample is subjected to special predictionExtracting characteristic parameters and measuring data, transmitting the characteristic parameters and the measured data to a constructed dynamic SVI early warning model, and predicting new activated sludge samples
Figure 353739DEST_PATH_IMAGE020
Then according to
Figure 698133DEST_PATH_IMAGE022
And judging whether the sludge is expanded or not in the interval of the values. Based on
Figure 133793DEST_PATH_IMAGE020
And dividing the value into 5 grades, and judging the sedimentation performance of the activated sludge to predict whether the sludge is expanded or not.
Figure 577544DEST_PATH_IMAGE020
When the value is less than 60, the sludge is in a first grade, almost enough filamentous bacteria are not contained in the sludge, and the sludge sedimentation rate is high;
Figure 793762DEST_PATH_IMAGE020
the value is 60-120, the second grade is obtained, and the activated sludge settling performance is good;
Figure 112485DEST_PATH_IMAGE020
the value is 120-180, the grade is the third grade, and the activated sludge settling performance is general;
Figure 402652DEST_PATH_IMAGE020
the value is the fourth grade when the value is 180-250, the sedimentation performance of the activated sludge is poor, a large number of filamentous bacteria exist, and the sludge bulking is easy to occur;
Figure 79621DEST_PATH_IMAGE020
when the value is more than 250, the sludge is in the fifth grade, the sedimentation performance of the activated sludge is deteriorated, the number of filamentous bacteria is in absolute dominance, and the sludge bulking phenomenon occurs.
The invention is mainly characterized in that:
the invention provides an activated sludge microscopic image fusion method based on discrete cosine transform for constructing a clear activated sludge phase difference microscopic image. Aiming at the inherent phenomena of inconsistent phase, halo and artifact of the activated sludge phase difference microscopic image, the filamentous bacterium and floc image segmentation method based on phase stretching transformation, edge detection and threshold segmentation is provided.
The invention constructs a dynamic SVI measurement model based on the morphological characteristic parameters of the flocs and the filamentous fungi extracted and on-line detection data MLSS, DO, pH and SVI historical measurement data by adopting a method of randomly configuring a network, thereby realizing SVI calculation and sludge bulking early warning, wherein the model inputs the morphological characteristic parameters and the measurement data and outputs the morphological characteristic parameters and the measurement data as SVI. The method integrates the morphological characteristic parameter information of the activated sludge image, provides an online and reliable measuring means for the SVI detection of the activated sludge volume index, and realizes the accurate prediction of the SVI and the accurate judgment of the sludge bulking.
Drawings
[01] FIG. 1 is a flow chart of a sludge bulking prediction method based on an activated sludge phase contrast microscopic image;
[02] FIG. 2 is a flow chart of phase difference microscopic image acquisition of activated sludge;
[03] FIG. 3 is a flow chart of activated sludge phase contrast microscopy image segmentation.
Detailed Description
[04] The technical solutions in the embodiments of the present invention are clearly and completely described below with reference to the drawings in the embodiments of the present invention, and it is obvious that the described embodiments are only a part of the embodiments of the present invention, and not all embodiments.
[05] As shown in fig. 1, it is a flow chart of a sludge bulking prediction method based on an activated sludge phase contrast microscopic image according to the present invention. The invention relates to a sludge bulking prediction method based on an activated sludge phase difference microscopic image, which is characterized by comprising the following steps of:
[06] step 1: image acquisition: the sludge sample is taken from the outlet of an aeration tank of a sewage treatment plant, and when the sludge sample is collected, the instantaneous water sample is collected at the water outlet of an aerobic tank, and then the water quality index is measured in a laboratory. 500 mL of sludge sample and effluent sample are taken each time, and the interval time between sample collection, image collection and water quality index measurement is not more than 3 hours. A calibration pipette with a tip having a tip end in cross section was used to place 10. mu.L of a sample on a slide, covered with a 24 mm by 24 mm square cover slip, and scanned from top left to bottom right using an optical microscope, an industrial digital camera, and an image acquisition system to obtain a plurality of phase-contrast microscopic images at three different focal planes, top, middle, and bottom. A total of 108 pictures (12 × 3 × 3) were obtained per sample to improve the representativeness of the sludge information. During the test, 36 groups of samples were obtained, and a total of 3888 pictures (108X 36). The flow of acquiring the phase-contrast microscopic image of the activated sludge is shown in figure 2;
[07] step 2: image fusion: performing edge enhancement sharpening pretreatment on microscopic images at the same position and different focal planes by adopting an activated sludge microscopic image fusion method of discrete cosine transform, respectively calculating the transformation coefficients of DCT domains of the block images, selecting subblocks with large coefficient variance as subblocks of the fused image, realizing image fusion through DCT inverse transformation, and finally further correcting the fused image by applying bilateral filter consistency check;
[08] and step 3: image segmentation: comprises two parts of filamentous fungi and floc segmentation. The activated sludge microimage segmentation process is shown in fig. 3, and firstly, RGB images (red, green and blue) are converted into gray images after phase differences are removed. The grayscale image is gaussian low-pass filtered to reduce random noise by blurring the image. PST extracts image edges by applying more phase to higher frequency domain features to emphasize edge information in the image. The PST converts an image from a spatial domain to a frequency domain by utilizing two-dimensional Fourier transform, then multiplies an obtained frequency domain image by a phase kernel, and converts the obtained image from the frequency domain to the spatial domain by utilizing two-dimensional inverse Fourier transform to obtain an output phase image. And (3) binarizing the output phase image by adopting a threshold value method, and segmenting flocs from the gray level image by adopting the threshold value method in the phase contrast image. After detection of floes, morphological dilation was performed to fill. Subtracting the obtained mask from the divided binary image obtained using the PST, leaving only filamentous bacteria in the output image;
[09] and when the phase contrast image is segmented by adopting a PST method, misclassifying part of the targets into foreground targets. These falsely detected pixels can be of both random noise and boundary noise types, in the first phase objects with pixels smaller than 20 will be removed, in the second phase the flocs and filamentous bacteria are distinguished with reduced radius of rotation and removed;
and 4, step 4: feature extraction: separating flocs and filamentous fungi in the image from the background, calculating morphological characteristic parameters such as sizes and shapes of the flocs and the filamentous fungi, and extracting the morphological characteristic parameters of each target as follows:
Figure DEST_PATH_IMAGE023
Figure 392922DEST_PATH_IMAGE024
the Euler number can be used as the characteristic of an identified object, the Euler number is defined as the number of connected components minus the number of cavities, and the topological characteristic has the characteristics of stability, rotation and unchanged proportion; the shape factor (FF) is mainly used for comparing the deviation between the target and the circle, the shape factor is used for judging whether the boundary is smooth, the value range of the shape factor is 0 to 1, and the shape factor of the circle is 1; roundness, denoted by R, which is determined by the elongation of its target, ranges from 0 to 1, where the roundness of a circle is 1; the radius of rotation, RG, is also determined by the target elongation, the radius of rotation of the longer flocs is larger, where the radius of rotation of the circle is
Figure 79118DEST_PATH_IMAGE026
(ii) a Aspect ratio, which is a feature that indicates the symmetry of the object shape, is denoted by AR.
There may be many targets in each image, the above parameters (except the total area) calculated for the images of the multiple targets are averaged, and 16 morphological features are calculated for each image of the separated flocs and filamentous fungi, that is, 32 morphological parameters are extracted from each sample. We selected 32 basic features in total, and the purpose of this part of the study was to find image analysis information and
Figure 662940DEST_PATH_IMAGE028
the correlation between them;
and 5: dynamic SVI model: the modeling method adopts a dynamic random configuration network model, and the model input comprises activated sludge microscopic image flocs and filamentous fungus morphological characteristic parameters
Figure 714073DEST_PATH_IMAGE030
On-line measurement of data
Figure 576986DEST_PATH_IMAGE032
And SVI historical data
Figure 801294DEST_PATH_IMAGE034
The output of the model is
Figure 564589DEST_PATH_IMAGE036
. The SVI dynamic SCN model is represented as
Figure 786623DEST_PATH_IMAGE038
Wherein
Figure 871253DEST_PATH_IMAGE040
Step 6: sludge bulking early warning: extracting the characteristic parameters of filamentous bacteria and flocs of a phase contrast microscopic image of a new activated sludge sample, and measuring MLSS, Do, pH value and history on line
Figure 836935DEST_PATH_IMAGE042
Using the constructed dynamics
Figure DEST_PATH_IMAGE044
And predicting the SVI value of the new activated sludge sample. Based on
Figure DEST_PATH_IMAGE046
Value, divided into 5 grades, judging sedimentation performance of activated sludge and predicting sludgeWhether the mud expands.
Figure DEST_PATH_IMAGE048
When the value is less than 60, the sludge is in a first grade, almost enough filamentous bacteria are not contained in the sludge, and the sludge sedimentation rate is high;
Figure DEST_PATH_IMAGE050
the value is 60-120, the second grade is obtained, and the activated sludge settling performance is good;
Figure 834497DEST_PATH_IMAGE048
the value is 120-180, the grade is the third grade, and the activated sludge settling performance is general;
Figure DEST_PATH_IMAGE052
the value is the fourth grade when the value is 180-250, the sedimentation performance of the activated sludge is poor, a large number of filamentous bacteria exist, and the sludge bulking is easy to occur;
Figure DEST_PATH_IMAGE054
when the value is more than 250, the sludge is in the fifth grade, the sedimentation performance of the activated sludge is deteriorated, the number of filamentous bacteria is in absolute dominance, and the sludge bulking phenomenon occurs.
The method comprises the steps of activated sludge microscopic image acquisition, image fusion, image segmentation, feature extraction and dynamic state
Figure 663650DEST_PATH_IMAGE054
Model and sludge bulking early warning, and the method integrates the morphological characteristic parameter information and the measurement data of the activated sludge image into the volume index of the activated sludge
Figure 563473DEST_PATH_IMAGE054
The detection provides an on-line and reliable measuring means, and provides an efficient method for predicting whether the activated sludge is swelled.

Claims (7)

1. A sludge bulking prediction method based on an activated sludge phase difference microscopic image is characterized by comprising the specific steps of image acquisition and image fusionMerging, image segmentation, feature extraction, and motion
Figure DEST_PATH_IMAGE001
Model and sludge bulking early warning six steps:
step 1: image acquisition: collecting an activated sludge sample at an outlet of an aeration tank of a sewage treatment plant, and capturing an activated sludge phase difference microscopic image by adopting an optical microscope, an industrial digital camera and an image acquisition system;
step 2: image fusion: fusing the activated sludge phase difference microscopic images of different focal planes by adopting an activated sludge microscopic image fusion method of discrete cosine transform, and finally further correcting the fused image by using bilateral filter consistency check;
and step 3: image segmentation: after phase difference is removed by adopting a phase consistency method, activated sludge filamentous bacteria and flocs are segmented by low-pass filtering, phase stretching transformation, threshold segmentation, starting operation and gyration radius characteristics;
and 4, step 4: feature extraction: calculating morphological characteristic parameters such as the sizes and the shapes of the separated flocs and the separated filamentous fungi;
and 5: dynamic state
Figure 746939DEST_PATH_IMAGE001
Model: combining morphological characteristic parameters with MLSS, DO, pH and SVI historical data of online measurement data, and constructing a dynamic random configuration network model by taking a sludge volume index as output
Figure 903114DEST_PATH_IMAGE001
An early warning model;
step 6: sludge bulking early warning: and (3) carrying out morphological feature extraction on a new activated sludge sample, fusing online physicochemical measurement data, conveying the data to a constructed dynamic SVI early warning model, judging the settling performance of the sludge according to the interval range of the volume index value of the output sludge, and predicting whether the sludge is expanded.
2. The activated sludge phase difference microscopic image-based sludge bulking prediction method according to claim 1, wherein the image collection is to scan a sample on a slide glass from the upper left to the lower right by using an optical microscope, an industrial digital camera and an image collection system on an activated sludge sample obtained from an aeration tank outlet of a sewage treatment plant to obtain a plurality of phase difference microscopic images at upper, middle and lower three different focal planes.
3. The activated sludge phase difference-based microscopic image sludge bulking prediction method according to claim 1, wherein the image fusion is characterized in that after edge enhancement sharpening pretreatment is carried out on microscopic images at the same position and different focal planes, discrete cosine transform is carried out on the images, the transform coefficients of DCT domains of the block images are respectively calculated, subblocks with large coefficient variance are selected as subblocks of the fused image, image fusion is realized through DCT inverse transformation, and finally, bilateral filter consistency check is applied to further modify the fused image.
4. The activated sludge phase difference microscopic image-based sludge bulking prediction method according to claim 1, wherein in the image segmentation, RGB images (red, green and blue) are converted into gray level images after phase difference influence is removed by adopting a phase consistency method, Gaussian low-pass filtering is carried out on the gray level images, and random noise is reduced by blurring the images; performing phase stretching transformation on the image, converting the image from a spatial domain to a frequency domain by utilizing two-dimensional Fourier transformation, multiplying the obtained frequency domain image by a phase kernel, and converting the obtained image from the frequency domain to the spatial domain by utilizing two-dimensional inverse Fourier transformation to obtain an output phase image; binarizing the output phase image by adopting a threshold value method, and segmenting flocs from a gray level image by adopting the threshold value method in the phase contrast image; after detecting the flocs, performing morphological dilation filling; the obtained mask is subtracted from the divided binary image obtained using the PST, and the result is output as a filamentous fungus image.
5. The activated sludge phase difference microscopic image-based sludge bulking prediction method according to claim 1, wherein the feature extraction comprises morphological features of sludge flocs and filamentous bacteria, specifically, the area of a total target region, the area of an average target region, the length of a long axis of a second-order central moment ellipse, the length of a short axis of the second-order central moment ellipse, the eccentricity of the second-order central moment ellipse, an intersection angle of the long axis of the second-order central moment ellipse and an x axis, the number of on pixels in a filling region convex polygonal image, the number of on pixels in a filling region image, an Euler number, a perimeter, a contour line perimeter, an equivalent circle diameter, a shape factor, a roundness, a rotation radius and an aspect ratio.
6. The activated sludge phase difference microscopic image-based sludge bulking prediction method according to claim 1, wherein the dynamic SVI model and the modeling method adopt a dynamic random configuration network model, and model inputs comprise activated sludge microscopic image flocs and filamentous fungus morphological characteristic parameters
Figure 762486DEST_PATH_IMAGE002
On-line measurement of data
Figure DEST_PATH_IMAGE003
And
Figure 620589DEST_PATH_IMAGE001
historical data
Figure 794082DEST_PATH_IMAGE004
The output of the model is
Figure DEST_PATH_IMAGE005
；
Figure 680129DEST_PATH_IMAGE001
The dynamic SCN model is expressed as
Figure DEST_PATH_IMAGE007
Wherein
Figure 950881DEST_PATH_IMAGE008
。
7. The activated sludge phase difference microscopic image-based sludge bulking prediction method according to claim 1, wherein the sludge bulking pre-warning is performed to obtain new activated sludge sample morphological characteristic parameters and online measurement data MLSS, Do and pH value, and the constructed dynamic state is utilized
Figure 109330DEST_PATH_IMAGE001
Early warning model of sludge volume index of activated sludge
Figure 555354DEST_PATH_IMAGE001
Make a forecast based on
Figure 499171DEST_PATH_IMAGE001
And judging whether the sludge is expanded or not in the interval of the values.
CN202110088840.8A 2021-01-22 2021-01-22 Sludge bulking prediction method based on activated sludge phase difference microscopic image Pending CN112767362A (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202110088840.8A CN112767362A (en) 2021-01-22 2021-01-22 Sludge bulking prediction method based on activated sludge phase difference microscopic image

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202110088840.8A CN112767362A (en) 2021-01-22 2021-01-22 Sludge bulking prediction method based on activated sludge phase difference microscopic image

Publications (1)

Publication Number Publication Date
CN112767362A true CN112767362A (en) 2021-05-07

Family

ID=75705868

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202110088840.8A Pending CN112767362A (en) 2021-01-22 2021-01-22 Sludge bulking prediction method based on activated sludge phase difference microscopic image

Country Status (1)

Country Link
CN (1) CN112767362A (en)

Cited By (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN114581468A (en) * 2022-03-04 2022-06-03 平顶山学院 Activated sludge strain segmentation method based on anisotropic phase stretch transformation

Citations (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN111192236A (en) * 2019-12-13 2020-05-22 沈阳化工大学 SVI (singular value index) measurement method based on activated sludge phase difference microscopic image
CN111724394A (en) * 2020-06-08 2020-09-29 浙江大学 Matlab image analysis and self-optimization-based rapid characterization method for sludge morphology

Patent Citations (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN111192236A (en) * 2019-12-13 2020-05-22 沈阳化工大学 SVI (singular value index) measurement method based on activated sludge phase difference microscopic image
CN111724394A (en) * 2020-06-08 2020-09-29 浙江大学 Matlab image analysis and self-optimization-based rapid characterization method for sludge morphology

Cited By (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN114581468A (en) * 2022-03-04 2022-06-03 平顶山学院 Activated sludge strain segmentation method based on anisotropic phase stretch transformation

Similar Documents

Publication Publication Date Title
CN108564114B (en) Human body fecal leucocyte automatic identification method based on machine learning
CN105825169B (en) A kind of pavement crack recognition methods based on road image
CN101153850A (en) Method and system for detecting asphalt mixture
CN114882040B (en) Sewage treatment detection method based on template matching
CN110766689A (en) Method and device for detecting article image defects based on convolutional neural network
CN110334760B (en) Optical component damage detection method and system based on RESUnet
CN112215819A (en) Airport pavement crack detection method based on depth feature fusion
CN116758071B (en) Intelligent detection method for carbon electrode dirt under visual assistance
CN116630813B (en) Highway road surface construction quality intelligent detection system
CN102214290B (en) License plate positioning method and license plate positioning template training method
CN115311507B (en) Building board classification method based on data processing
CN115272339A (en) Metal mold dirt cleaning method
CN109544513A (en) A kind of steel pipe end surface defect extraction knowledge method for distinguishing
CN112767362A (en) Sludge bulking prediction method based on activated sludge phase difference microscopic image
Benens et al. Evaluation of different shape parameters to distinguish between flocs and filaments in activated sludge images
Jenné et al. Towards on-line quantification of flocs and filaments by image analysis
Kaddah et al. Automatic darkest filament detection (ADFD): a new algorithm for crack extraction on two-dimensional pavement images
CN110533626B (en) All-weather water quality identification method
Shanono et al. Image processing techniques applicable to wastewater quality detection: towards a hygienic environment
Cenens et al. On the development of a novel image analysis technique to distinguish between flocs and filaments in activated sludge images
CN117036346B (en) Silica gel sewage treatment intelligent monitoring method based on computer vision
CN115082379A (en) Activated sludge phase contrast microscopic image floc and filamentous bacterium segmentation method
CN114581468B (en) Activated sludge strain segmentation method based on anisotropic phase stretching transformation
CN115330792A (en) Sewage detection method and system based on artificial intelligence
CN115496716A (en) Single and double micronucleus cell image detection method based on deep learning and related equipment

Legal Events

Date Code Title Description
PB01 Publication
PB01 Publication
SE01 Entry into force of request for substantive examination
SE01 Entry into force of request for substantive examination