A method for monitoring the state of a floe based on texture features

Abstract

The application discloses a flocculation state monitoring method based on texture features, which comprises the following steps: taking flocculation water as a sample, dividing the flocculation water into several equal parts, placing the flocculation water in transparent containers respectively, making the flocculation water in a suspended state, taking multiple sample images by using a camera, selecting sample images for pretreatment, extracting a gray level co-occurrence matrix of the images, calculating texture feature values of the images by using a texture extraction algorithm, taking average values of the texture feature values of multiple images of the same sample as results, repeating the above operations to obtain texture feature values of flocculation images in different time periods, calculating relative change amounts of the texture feature values respectively by using a formula, and using the relative change amounts of the texture feature values to judge the state of the flocculation.

Description

TECHNICAL FIELD

[0001] The present application belongs to the technical field of water treatment, and particularly relates to a floc state monitoring method based on texture features. BACKGROUND

[0002] In the daily operation and management of some waterworks, it is often difficult to make a reasonable assessment of the floc state in the water treatment unit due to factors such as old process equipment, poor experimental conditions, and uneven professional knowledge levels of the staff. Changes in the floc state caused by sudden conditions such as a large increase in the inflow turbidity, flow rate in a short period of time, and leakage of coagulant from the dosing pipe cannot be grasped in a timely manner, so that appropriate adjustment measures cannot be taken in a timely manner, resulting in the inability to generate large and dense flocs for sedimentation in the flocculation stage, and further affecting the water quality.

[0003] In the existing methods for monitoring the floc state, most operators use visual observation, mainly based on visual observation of the floc morphology and density and other related conditions. This method is easily affected by subjective factors and experience, and thus has low reliability.

[0004] Patent application No. CN200710144647 proposes a floc state recognition method based on morphological features. This method recognizes the floc state by taking a floc image and calculating the morphological feature values of the floc. However, morphological features are individual characteristics of a single floc, and there is a large random error and a large amount of calculation. SUMMARY

[0005] In order to overcome the shortcomings of the prior art, the purpose of the present application is to provide a floc state monitoring method based on texture features. In the water treatment process, texture feature values such as entropy and correlation are extracted from the floc image. Through analysis of the floc texture features, it is found that the entropy and correlation in the texture feature values have good correlation with the compactness, morphology size, and sedimentation of the floc. The present application uses the texture feature values extracted from the floc image to quantitatively analyze the floc state, making the results more accurate and objective. The present application has the advantages of being simple, intuitive, accurate, efficient, and low cost.

[0006] In order to achieve the above purpose, the technical solution adopted by the present application is as follows:

[0007] A floc state monitoring method based on texture features, comprising the following steps:

[0008] Step 1: Take floc-containing water as a sample and divide it into an equal number of floc-containing water samples;

[0009] Step 2: Place the floc-containing water samples obtained in step 1 in transparent containers respectively, so that they are distributed in a suspended state in the transparent containers, to obtain a plurality of floc water samples distributed in a suspended state.

[0010] Step 3, taking pictures of several samples of floc water in suspension state distribution obtained in step 2 by using a camera, and taking multiple sample images;

[0011] Step 4, selecting sample images from the multiple sample images taken in step 3 for preprocessing, and extracting the gray level co-occurrence matrix of the images by using image analysis software or open source vision library;

[0012] Step 5, calculating the texture feature values of the images by using the gray level co-occurrence matrix extracted in step 4 through a texture extraction algorithm, wherein the texture feature values include entropy and correlation;

[0013] Step 6, taking the average value of the texture feature values of multiple images of the same sample as the result;

[0014] Step 7, repeating steps 1 to 6 to obtain the texture feature values of the floc images at different time stages, and calculating the relative change amounts of the texture feature values respectively by using the formula ; wherein R1 and R2 respectively represent the texture feature values of the previous stage and the next stage of the two consecutive monitoring stages;

[0015] Step 8, using the relative change amounts of the texture feature values respectively obtained in step 7 to judge the state of the floc:

[0016] When the relative change amount of the entropy of the two adjacent stages is greater than 0 and the relative change amount of the correlation is less than 0, it indicates that the floc density increases, the equivalent particle size increases, and the floc state becomes better;

[0017] When the relative change amount of the entropy of the two adjacent stages is less than 0 and the relative change amount of the correlation is greater than 0, it indicates that the floc density decreases, the equivalent particle size decreases, and the floc state becomes worse.

[0018] In step 2, the floc water sample in the transparent container is operated by stirring or aeration to be distributed in a suspended state in the transparent container.

[0019] In step 3, at least 5 sample images are taken for each sampling of the same batch of floc water samples in a suspended state distribution, and the camera shooting parameters are kept consistent.

[0020] In step 4, at least 3 similar sample images are selected for preprocessing.

[0021] In step 4, when the sample images are selected and analyzed, a sampling frame with the same size and position of the floc region in the sample image is selected for analysis.

[0022] The preprocessing in step 4 is to sequentially perform grayscale, median filtering, image segmentation, and contrast enhancement on the sample images.

[0023] The extraction algorithm of the texture feature value in the step 5 is realized by an image processing software or an open source vision library.

[0024] The average value calculation formula of the texture feature value in the step 6 is as follows: Wherein, Xi is each sample value, N is the sample quantity, and X is the final obtained texture feature value.

[0025] The state of the floc in the step 8 includes compactness, morphology size and sedimentation performance.

[0026] Compared with the prior art, the present application has the following beneficial effects:

[0027] 1. In the prior art, the morphological feature value is used to identify the floc state, which only represents the individuality feature of a single floc, has a large random error and a large amount of calculation; while the texture feature used in the present application is the overall feature obtained by traversing the floc image, has good stability and is easy to calculate, the floc state in the water treatment unit is monitored by analyzing the change of the texture feature of the floc sample, since the texture feature is a simple and convenient monitoring index, the present application can be applied to most running water treatment plants, the water treatment plant represents the floc state according to the texture feature value of the floc, evaluates and monitors the coagulation effect, objectively understands the current floc state and coagulation effect, adjusts the operation parameters in time, provides guiding opinions for the optimized operation and energy saving and consumption reduction of the water treatment plant, the present application uses a low-cost and objective method to monitor the floc state in the water supply treatment process, and is a method capable of effectively improving the operation and management effect of the water supply treatment.

[0028] 2. Accurate and objective. Unlike visual inspection method, the present application applies machine vision method, which uses the texture feature value extracted from the floc image to quantitatively analyze the floc state, so that the result is more accurate and objective, and does not depend on the rich experience of the staff.

[0029] 3. Beneficial to realize automation. The operation steps required by the present application such as image acquisition, image processing and texture feature value extraction can be realized by automatic operation with certain instrument equipment, and can be integrated in an online monitoring device, which is beneficial to realize the automation and unattended operation of the water plant.

[0030] 4. Economical and practical. The parameter used in the present application is based on the texture feature value extracted from the floc image, and the image acquisition only needs to use a low-priced image shooting device without other expensive detection devices, so that the cost is reduced.

[0031] In summary, the present application has the advantages of simplicity, intuition, accuracy, high efficiency and low cost. BRIEF DESCRIPTION OF DRAWINGS

[0032] Figure 1Fig. 1 is a diagram of the variation of the floc image features and the turbidity of the settled water under different dosages of the flocculant in the embodiment of the present application; wherein: Figure 1 (a) is a diagram of the variation of the texture feature values and the turbidity of the settled water under different dosages of the flocculant; Figure 1 (b) is a diagram of the variation of the morphological feature values and the turbidity of the settled water under different dosages of the flocculant.

[0033] Figure 2 Fig. 2 is a diagram of the floc state after the addition of the flocculant in the embodiment of the present application.

[0034] Figure 3 Fig. 3 is a pretreatment diagram of the floc state after the addition of the flocculant in the embodiment of the present application.

[0035] Figure 4 Fig. 4 is a diagram of the relative variation of the texture feature values under different dosages of the flocculant in the embodiment of the present application.

[0036] Figure 5 Fig. 5 is a pretreatment diagram of the floc state under different dosages of the flocculant in the embodiment of the present application; wherein: Figure 5 (a) is a pretreatment diagram of the floc state when the dosage of the flocculant is insufficient in the embodiment of the present application; Figure 5 (b) is a pretreatment diagram of the floc state under the optimal dosage of the flocculant in the embodiment of the present application; Figure 5 (c) is a pretreatment diagram of the floc state when the dosage of the flocculant is excessive in the embodiment of the present application. DETAILED DESCRIPTION

[0037] The technical solutions of the present application are further described below through specific embodiments.

[0038] A floc state monitoring method based on texture features, comprising the following steps:

[0039] Step 1: Take the floc-containing water in the water treatment unit of a water treatment plant as a sample, and divide it into an equal number of floc-containing water samples.

[0040] Step 2: Place the floc-containing water samples in transparent containers respectively, and distribute the flocs in the transparent containers in a suspended state by stirring. When stirring, the stirring intensity should not be too large to prevent the flocs from being broken.

[0041] Step 3: Take pictures of the floc-containing water samples distributed in a suspended state by using a camera, i.e. take more than 5 pictures of the floc-containing water samples distributed in a suspended state of the same sample each time. When taking pictures, control the shooting parameters to be consistent, so that the images can truly reflect the texture information of the flocs.

[0042] Step 4: Select at least three images with similar floc distribution. Use sampling frames of the same size and position to crop the area where flocs exist and there is no interference from other objects. Use image analysis software to perform preprocessing steps such as grayscale conversion, median filtering, image segmentation, and contrast enhancement on each image before extracting its grayscale co-occurrence matrix.

[0043] Step 5: Using image analysis software, calculate the entropy and correlation in the texture feature values ​​using the gray-level co-occurrence matrix of the image extracted in Step 4 through a texture extraction algorithm;

[0044] Step 6: Take the average of the texture feature values ​​of multiple images of the same sample as the result. The formula for calculating the average of each texture feature value is as follows: Where X i For each sample value, N is the number of samples, and X is the final texture feature value.

[0045] Step 7: Repeat steps 1 to 6 to obtain the texture feature values ​​of floc images in the water treatment plant at different time stages, and use the formula... The relative changes of the texture feature values ​​are calculated respectively; where R1 and R2 represent the texture feature values ​​of the previous and next stages of continuous monitoring, respectively. The sampling time interval is 1 day, which can also be adjusted according to actual needs.

[0046] The texture feature values ​​in step 7 include correlation and entropy.

[0047] Step 8: Use the relative changes of the obtained texture feature values ​​to determine the state of the flocs. The state of the flocs includes density, shape and size, and settling performance.

[0048] When the relative change in entropy between two adjacent stages is greater than 0 and the relative change in correlation is less than 0, it indicates that the density of the flocs increases, the equivalent particle size increases, and the state of the flocs improves.

[0049] When the relative change in entropy between two adjacent stages is less than 0 and the relative change in correlation is greater than 0, it indicates that the density of the flocs decreases, the equivalent particle size decreases, and the state of the flocs deteriorates.

[0050] Furthermore, in step 1, the floc-containing water can be taken from a water treatment plant, such as a waterworks or a sewage treatment plant; the floc-containing water can be taken from different types of flocculation tanks, clarification tanks, flocculation equipment, and other water treatment units that contain floc-containing water.

[0051] Furthermore, when analyzing images using professional image analysis software, it is necessary to select areas where flocculent matter exists but without interference from other objects.

[0052] The beneficial effects of the present invention will be further illustrated below through specific embodiments.

[0053] Embodiment

[0054] Step 1, taking raw water in a pretreatment tank of a waterworks in Gansu Province as a sample, which was divided into 3 equal parts in beakers, each with a volume of 1 L;

[0055] Step 2, using polyaluminum chloride (PAC) as a flocculant, the flocculant was added to each water sample at a certain dosage, and a stirrer was used to stir rapidly at a stirring speed of 200 r / min for 1 min and slowly at a stirring speed of 40 r / min for 9 min;

[0056] Step 3, immediately after stopping stirring, the camera was used to continuously take 5 images of each water sample with the same shooting parameters, as shown in FIG. 1, a total of 3x5=15 flocculation images were obtained; Figure 2

[0057] Step 4, 3x3=9 flocculation images with similar and clear distribution of flocculation were selected from the images taken from each group of water samples, a sampling frame with the same size and position was used to intercept the area where flocculation existed and was not disturbed by impurities, and then the image analysis software was used to sequentially perform preprocessing operations such as graying, median filtering, image segmentation, and contrast enhancement on each image, and the preprocessed image is shown in FIG. 2; Figure 3

[0058] Step 5, the texture feature values of the preprocessed flocculation images were extracted using the texture extraction algorithm based on the gray level co-occurrence matrix in the image processing software, and then the average value of each texture feature value was calculated as the final result of the batch sample;

[0059] Step 6, changing the dosage of the flocculant, repeating steps 1 to 5, and obtaining the flocculation image texture feature values under different dosages of the flocculant, as shown in FIG. 3(a); Figure 1 (a), each texture feature index was calculated by , where R1 and R2 represent the texture feature values of the previous stage and the next stage of the two consecutive stages, respectively, and the relative change diagram of the texture features under different dosages of the flocculant was obtained, as shown in FIG. 3(b); Figure 4

[0060] Step 7, after stopping stirring each time, the flocculation suspension was allowed to stand for 20 min, and then the supernatant was taken and the turbidity of the supernatant was measured using a turbidimeter, which was the turbidity of the settled water after flocculation of the raw water under each dosage, as shown in FIG. 4(a); Figure 1

[0061] Step 8, the flocculation images obtained under each dosage were sequentially subjected to graying and binarization processing using the image processing software, and then the morphological characteristics such as flocculation particle size, flocculation quantity, and fractal dimension were measured, and the morphological characteristic data under each dosage stage were obtained as shown in FIG. 4(b).​​​​Figure 1 (b) shows.

[0062] Example analysis:

[0063] From Figure 1 (b) can be seen that in the monitoring stage of insufficient coagulant dosage (dosage < 25 mg / L), with the addition of coagulant, the fractal dimension of flocs, floc size gradually increases, and the turbidity of settled water gradually decreases, which shows that the flocs in water gradually increase and become more compact, the boundary between the flocs in the image and the image background becomes more obvious, and the floc state gradually improves. The floc image in the stage of insufficient dosage is shown in Figure 5 (a). At this time, from Figure 1 (a), Figure 4 It can be seen that the correlation of the floc image gradually decreases, and the relative change of the correlation is less than 0, because with the increase of the dosage, a large number of small particles in the water gradually collide and aggregate to form flocs, and the floc image features change from being full of small particles to being shaped flocs with a small amount of uneven distribution, thereby reducing the correlation of the image. The entropy of the floc image gradually increases, and the relative change of the entropy is greater than 0, because the flocs gradually increase and become compact, which leads to the increase of image information complexity, thereby increasing the entropy.

[0064] From Figure 1 (b) can be seen that when the coagulant dosage is the optimal dosage (25 mg / L), the fractal dimension of flocs, floc size reaches the maximum, and the turbidity of settled water reaches the minimum, which shows that the floc state is the largest and most compact among the multiple stages, and has the best settling property. The floc image under the optimal dosage is shown in Figure 5 (b). Figure 1 (a) can be seen that the entropy in the texture feature value of the floc image reaches the peak value, and the correlation reaches the minimum value.

[0065] From Figure 1 (b) can be seen that in the monitoring stage of excessive coagulant dosage (dosage > 25 mg / L), with the addition of coagulant, the fractal dimension of flocs, floc size gradually decreases, and the turbidity of settled water gradually increases, which is due to the increasing repulsion between the flocs, the re-stabilization of colloidal particles, the gradual loosening of large flocs under the shear force of water flow, and the formation of multiple small flocs, resulting in the decrease of floc density and particle size, and the gradual deterioration of floc state. The floc image in the stage of excessive dosage is shown in Figure 5 (c). At this time, from Figure 1 (a), Figure 4It can be seen that the correlation of the flocculation image gradually increases, and the relative change of the correlation is greater than 0, because the fine flocculation makes the flocculation distribution more uniform, thereby increasing the correlation; and the entropy of the flocculation image gradually decreases, and the relative change of the entropy is less than 0, because the small and loose flocculation reduces the complexity of the image, thereby reducing the entropy.

[0066] In addition, any one of the flocculation images after the pretreatment in step 4 is taken, and on the basis of the image, the proportion of the image analysis area is gradually reduced to obtain five flocculation images with different area sizes, and the morphological characteristic values and the texture characteristic values of each image are extracted, as shown in Table 1.

[0067] Table 1 - Morphological characteristic values and texture characteristic values of each image

[0068]

[0069] The extraction of the flocculation image features is performed by framing the flocculation in a certain area range of the flocculation image, and for the same image, the total number of flocculation used for analysis decreases with the decrease of the image framing proportion. As shown in Table 1, when the image framing proportion is lower than 60%, the morphological characteristics of the flocculation image gradually distort, and the dispersion coefficient is large, while the texture characteristics are relatively stable, and the dispersion coefficient is small. This is because the texture characteristics are not the characteristics of a single flocculation, but the comprehensive characteristics obtained by analyzing the distribution and density of all flocculations in the image, which has the characteristics of strong integrity and high stability, and will not change greatly due to the different extraction ranges of the flocculation image. As can be seen from the above example, using the texture characteristics of the flocculation to monitor the flocculation state is not only economical and practical, but also stable and objective, and does not depend on the rich experience and subjective judgment of the operator.

[0070] Therefore, the present application is effective in monitoring the flocculation state by the texture characteristics of the flocculation.

[0071] The present application is not limited to the above embodiments, and based on the technical solutions disclosed in the present application, those skilled in the art can make some substitutions and modifications to some technical features without creative labor, and these substitutions and modifications are all within the protection scope of the present application.

Claims

1. A method for monitoring floc state based on texture features, characterized in that: Includes the following steps: Step 1: Take water containing flocculent material as a sample and divide it into several equal portions of water containing flocculent material. Step 2: Place several flocculent water samples obtained in Step 1 into transparent containers, so that they are distributed in a suspended state in the transparent containers, thus obtaining several flocculent water samples distributed in a suspended state. Step 3: Use a camera to photograph several flocculent water samples obtained in Step 2 that are distributed in a suspended state, and take multiple sample images. Step 4: Select sample images from the multiple sample images captured in Step 3 for preprocessing, and extract the gray-level co-occurrence matrix of the images using image analysis software or open-source vision libraries. Step 5: Calculate image texture feature values ​​using the gray-level co-occurrence matrix extracted in Step 4 through a texture extraction algorithm. The texture feature values ​​include entropy and correlation. Step 6: Take the average value of the texture feature values ​​of multiple images of the same sample as the result; Step 7: Repeat steps 1 to 6 to obtain the texture feature values ​​of the floc images at different time stages, and use the formula... The relative changes of the texture feature values ​​are calculated respectively; where R1 and R2 represent the texture feature values ​​of the previous and next stages of continuous monitoring, respectively. Step 8: Use the relative changes in the texture feature values ​​obtained in Step 7 to determine the state of the flocs. When the relative change in entropy between two adjacent stages is greater than 0 and the relative change in correlation is less than 0, it indicates that the density of the flocs increases, the equivalent particle size increases, and the state of the flocs improves. When the relative change in entropy between two adjacent stages is less than 0 and the relative change in correlation is greater than 0, it indicates that the density of the flocs decreases, the equivalent particle size decreases, and the state of the flocs deteriorates.

2. The method for monitoring floc state based on texture features according to claim 1, characterized in that: In step 2, the water sample containing flocculent material in the transparent container is stirred or aerated to make it suspended in the transparent container.

3. The method for monitoring floc state based on texture features according to claim 1, characterized in that: In step 3, at least 5 images of the same batch of suspended flocculent water samples are taken each time, and the camera shooting parameters are kept consistent.

4. The method for monitoring floc state based on texture features according to claim 1, characterized in that: In step 4, at least three similar sample images are selected for preprocessing.

5. A method for monitoring floc state based on texture features according to claim 1 or 4, characterized in that: When performing selection analysis on sample images, a sampling frame of the same size and position as the flocculent region in the sample image is selected to extract the analysis portion.

6. A method for monitoring floc state based on texture features according to claim 1 or 4, characterized in that: The preprocessing in step 4 involves sequentially performing grayscale conversion, median filtering, image segmentation, and contrast enhancement on the sample image.

7. The method for monitoring floc state based on texture features according to claim 1, characterized in that: The texture feature extraction algorithm in step 5 is implemented by image processing software or open-source vision libraries.

8. The method for monitoring floc state based on texture features according to claim 1, characterized in that: The formula for calculating the average value of texture feature values ​​in step 6 is as follows: Where Xi is each sample value, N is the number of samples, and X is the final texture feature value.

9. The method for monitoring floc state based on texture features according to claim 1, characterized in that: The state of the flocs in step 8 includes density, shape and size, and settling performance.

Images