Limiting Factors of Simultaneous Measurement Method for Turbidity and Total Suspended Solids Based on Image Processing Approaches

Image processing is one of the computational approaches that can be applied to measure fluctuations in surface water pollutant concentrations. Identifying factors that can affect and become a limitation in the development of image processing-based simultaneous measurement methods is the focus of the discussion in this article. Nineteen variables have been identified from the three-factor categories of hardware configuration, software settings, and the measured suspension characteristics. Measurements were made on thirty images extracted from video captured with a USB Mechanic-DX-230 camera with a 23-megapixel resolution Panasonic CMOS sensor equipped with a macro lens in 130X magnification on the S-EYE-1.6.0.11 interface. Image frame extraction was performed with VirtualDub2 build 4428/release, followed by digital image processing and analysis with ImageJ 1.46r. The lens’s focal length to the sample cell is 5 cm, setting the brightness parameter minimum of 90 and maximum of 255 and minimum threshold settings of 209 and maximum of 255, which is a controlled factor to achieve the best repeatability rate. It is indicated by a relative standard deviation of up to 6% in the measurement chamber with a dark background. Complying with the stated factors is essential to ensure measurement results’ reliability and validity.


Introduction
Measurement of turbidity level based on image processing is a hot topic of discussion.Measurement of color intensity differences and edge detection techniques (particle counting) are two approaches studied in previous studies [1,2].A 3D image recognition-based approach has also been carried out to differentiate turbidity caused by abiotic and biotic suspensions [3].Insoluble inorganic particles, including silica, kaolin, and clay, mainly contribute to turbidity [4].Meanwhile, insoluble organic particles in the form of microorganisms are another contributor to the cause of turbidity [5].The existing turbidity measurement method (nephelometric) cannot distinguish the type of turbidity contributed by the two components.In the nephelometric method, all suspended particles that scatter light are totalled as turbidity [6].
Furthermore, dissolved materials that can interact with monochromatic light beams at specific wavelengths interfere with the turbidity measurement results [7].Using monochromatic light sources at infrared wavelengths solves the interference problem [8].However, the presence of air bubbles, particle shape, size heterogeneity, and the carcinogenic nature of formazin are of great urgency for further study [9,10].Developing a multi-detector system at various lighting angles is an option designed to anticipate potential measurement errors caused by some air bubbles in the measured suspended sample [10].
Currently, in terms of the heterogeneity of particle shape and size, the use of formazin suspension as a turbidity standard is considered to be able to overcome this problem.The potential for measurement error is very likely due to differences in the characteristics of the standards used with suspended particles that naturally exist in surface water flows.Standard formazin is made by mixing two types of polymers, which produce a suspension particle shape and size that tends to be uniform [11].So, it is considered less able to represent the actual condition of suspended particle content in surface water.This article discusses the potential implementation of the image processing approach and using high-purity grade kaolin in turbidity measurement by studying factors that can affect the results of measuring the number of particles.In this case, the number of particles counted becomes a parameter to predict the level of turbidity and the concentration of total suspended solids (TSS).In practice, measuring the number of particles based on image processing requires optimization of software and hardware configurations to standardize the proposed measurement method.This research aims to identify several limiting factors for image processing-based measurement techniques as a reference basis for further development.

Materials and Method
Development of measurement chamber, determination of light source, camera resolution, measurement chamber background, angle and directions of a light source, shape and volume of the sample cell, camera-lense to sample distance, software parameter settings, isolation of ambient light interference and characterization of suspended solution are the series of activities included on this research.

Development of measurement chamber
The measurement chamber is made to reduce ambient light interference, which can theoretically affect suspended particle quantification results.Sunlu PLA+ 3D printer filament is a measurement chamber material printed using the Creality CR-10 Mini 3D printer with a bed temperature setting of 60 o C and a nozzle temperature of 210 o C. PLA+ 3D printer filament with black color is used to create a measurement chamber with a dark background (BMC), while PLA+ 3D printer filament with white color is used to create a measurement chamber with a light background (WMC).The design of the measurement chamber was made using the FreeCAD open-source software version 0.17.8443.The measurement chamber set is made in such a way as to ensure that the position of each complementary part is in a fixed condition.Detailed measurements were made on the complementary parts, including a USB camera with a resolution of 23 megapixels Mechanic DX-230, a C-mount macro lens, a light emitting diode (LED) light source, and setting the distance between these parts.The measurement results are then poured into a 3-dimensional model to be printed using a 3D printer.The printed measurement chamber prototype will then be used as a measurement medium for various experimental treatment combinations in the laboratory.The light sources used are 5mm LED lamps, SMD5050 LEDs, and SMD2835 3-eye LED modules.A voltage source of ±3.2 Volt is applied to the 5mm and SMD5050 LED with the lighting position perpendicular to the camera lens.Whereas for the 3-eye LED module SMD2835, a voltage source of ±9 Volts and ±12 Volts is applied with the lighting position in the direction of the camera lens.At this stage, the surface water sample used as the measurement subject came from the Cilemahabang River at coordinates -6.27617843, 107.17796724.The selection of surface water samples as measurement objects ensures the developed system can detect suspended particles in natural water samples.
In addition to the treatment combinations in Table 1, other experiments were carried out to determine the camera resolution that could produce the best image capture results for the various treatment combinations.The three types of CMOS cameras being compared are non-branded 2megapixel USB cameras, 23-megapixel USB cameras Mechanic DX-230 and 51-megapixel USB cameras Mechanic RX-510.Thirty digital image frames from each treatment combination resulting from video extraction with VirtualDub were further processed with ImageJ and then statistically analyzed using the ANOVA method to see the effect of the lighting source and the type of camera used on the suspended particle quantification results.The relative standard deviation (%RSD) was calculated for five replicate measurements of 40 mg/L kaolin suspension solution in an experiment to determine the position of the light source and the measurement results' repeatability.The best treatment combination was determined based on the lowest %RSD value.The HANNA HI731331 has a circular cuvette base to produce a convex surface on the cuvet wall.Determining the best cuvette shape that suits the needs of this study cannot be separated from the use of the measurement chamber being developed.At this stage, images are taken of the same surface water samples in various treatment combinations of various factors, including the type of light source, lighting position, the amount of voltage source, and also the type of sample cell in order to obtain images with the best resolution which can be processed further with a processing approach and digital image analysis using ImageJ 1.46r open source software.The level of contrast between the background and the appearance of suspended particles in digital images is one of the important criteria as a reference for selecting the best image.Noise in the image due to light leakage is another thing to consider in connection with its effect on the determination of RoI, which will indirectly affect the quantification results of suspended particles caused by a processing threshold that is not optimal at a predetermined threshold setting value.
The artificial water sample used as the measurement object at this stage was prepared by dissolving high-purity grade kaolin SigmaAldrich K7375 at a concentration of 40 mg/L, which was determined randomly.The choice of a 40mg/L kaolin suspension solution as the object of measurement is to minimize the presence of interfering matrices that may be contained in surface water so that it is possible to affect the stability of the quantification value in the image captured by the camera.The relative standard deviation (%RSD) was calculated for five replicate measurements of 40 mg/L kaolin suspension in an experiment to determine the type of sample cell and the measurement results' repeatability.The best treatment combination was determined based on the lowest %RSD value.

Determination of software-parameter settings
Processing and analyzing digital images captured by a USB camera with the S-EYE 1.6.0.11interface, with the position of the sample cell placed in a customized measurement chamber, was performed using ImageJ 1.46r open source software.The principle of image processing that is carried out is based on the ability of the software to distinguish the brightness of suspended solid objects from the background in the image.The processing steps include converting the image from the RGB color format to 8-bit grayscale and adjusting the brightness level value.To then differentiate the object and the background by setting the threshold value in such a way as to obtain the best image processing results, which can be further quantified.The brightness and threshold setting values are determined using a trial and error approach to obtain the processed image with the best quality, which can be further quantified.

Characterization of suspended solution
The selection of a reference standard for the turbidity parameter is based on the similarity of the surface water sample characteristics with the artificial suspension solution used in the experiment.The similarity in question includes the content of the main components making up the turbidity and the size of the suspended particles in each measured solution.The effect of dilution on the concentration of the suspension solution is also considered to ensure that the developed method can detect changes in the concentration of suspended particles in the sample solution.The three types of samples being compared include the surface water of the west tarum canal taken from the sampling point at coordinates -6.31416347, 107.15377310; formazine suspension solution 4000NTU HACH246149; and high purity grade kaoline powder SigmaAldrich K7375.Separating suspended particles in surface water samples and formazin suspension solutions was carried out by vacuum filtration on Whatman 934-AH TSS grade microfiber filter paper.7.5 L of surface water samples were filtered using a vacuum filtration unit to produce five filter papers containing residual solids retained on their surface.As for the formazin suspension solution, the filtration process was carried out on 500 ml of 1000 NTU formazin suspension solution, which resulted from diluting the 4000 NTU formazin stock solution.The solid residue retained on the filter paper was dried at a temperature of 105 o C for 2 hours, then separated carefully in such a way as to obtain suspended particulate matter in powder form.The powder samples were then analyzed with SEM-EDS JEOL JSM-6510 LA to determine the composition of turbidity constituents in each suspension solution.Determination of suspended particle size in 20 NTU formazin suspension solution, 20 mg/L kaolin suspension solution, and surface water samples was carried out using the Analyze Particles feature in ImageJ 1.46r open source software.Variations in kaolin concentration in the study of the effect of the dilution level of the suspension solution were determined randomly at values of 0.04 mg/L, 5 mg/L, 20 mg/L, 20 mg/L, 40 mg/L, 70 mg/L, 100 mg/L, 200 mg/L, 800 mg/L and 1600 mg/L with two replicate treatments at each dilution concentration.

Results and discussion
The nineteen limiting factors identified are divided into three categories, namely eleven factors in the category of hardware configuration, five in the category of software parameter settings, and three in the category of suspended solution characteristics.

Hardware configuration
Eleven factors studied in this category include camera resolution, power supply, measurement chamber background, light source, sample cell shape, sample cell volume, light source, camera lens to sample cell distance, angle and direction of light source, magnification level, and ambient light interference.

3.1.1.
Light source and camera resolution.The combination of the light source and camera resolution must be optimized to produce the best quality digital image captured by the camera to be processed further.The experimental results show that using a 5050 SMD LED in combination with a 23 MP USB camera gets the best results based on visual observations and initial testing of digital image processing and analysis with ImageJ.The captured images of various combinations of light sources and camera resolutions are presented in Figure 1.It can be observed that of the six treatment combinations, only two produce digital images that can be further processed.Concerning the brightness level of the digital image, it can be distinguished between background and suspended particles, which will be quantified further.A combination of 2MP and 23MP camera resolution with an SMD5050 LED light source showed good contrast between the background and suspended particles in the digital image.In the third combination, the camera's resolution with 5mm LED and the combination of 51MP camera with SMD5050 LED cannot produce good image contrast, so the background cannot be distinguished from suspended particles.The results of the ANOVA test showed that the camera resolution variable and the type of light source used affect the calculated particle value shown by the P-value of the two independent variables <0.05 and show the interaction between interrelated variables at an alpha value of 0.05.The %RSD value obtained is still affected by "very random" movement at the beginning of image capture due to the influence of agitation at the beginning of the treatment.The experimental results show that a combination of a dark chamber background and 90 o perpendicular angle proven as the best combination that may produce the images' more consistent pattern.

Shape and volume of the sample cell.
Observations on measurements using two types of cubicle and cylindrical cell samples show that cubicle cell samples provide a more consistent particle distribution pattern compared to cylindrical cell samples due to differences in the face shape of the sample cells from one another.The "curvy" surface of the cylindrical sample cell distorts the form of suspended particles.In contrast, the "cubicle" sample cell has a flat surface to reduce "noise" due to the distortion of particle shape.Alternatively, it can be stated that the consistency of the distribution pattern is affected by the shape of the sample cell surface, as shown in Figure 3. Visualization E is an image sequence frame image on a cylindrical sample cell, while visualization F is an image sequence frame image on a cubicle sample cell.Observe the difference in the shape of the particles caught on camera.

Figure 3. Visualization of camera-captured images on (E) cylindrical cell samples and (F) and cubicle cell samples
In the cylindrical cell sample, there is a change in particle shape due to distortion due to the curvy shape of the cylindrical sample cell.Figure 4 presents a plot of the particle distribution patterns in both types of sample cells.

Suspended solution characteristics
Three factors studied in this category include the dilution level of the suspended solution, particle size analysis in suspension solutions, and analysis of the elemental composition of the suspended particles used.The experimental results show that the maximum dilution level that can be quantified using the developed digital image processing and analysis-based method is at a kaolin concentration of 100 mg/L or equivalent to 69.1 NTU from the readings of a commercial Eutech TN100 turbidimeter under experimental conditions.The plot of the results of the quantification of the number of kaolin particles at various concentrations and the results of the turbidity readings are presented in Figure 7.

Conclusions
Eleven factors are included in the hardware configuration category, five in the software-parameter settings, and three on suspended solution characteristics from the nineteen studied variables.The summary of limiting factors is provided in Table 2.The summary generated from various combination treatments was purposively conducted to answer relevant research questions related to the research objectives.It may conclude that some criteria should be complied with to achieve reliable and valid measurement results.Failed in compliance to comply with the stated requirement will affect the results.

Figure 1 .Figure 2 .
Figure 1.Captured images of various combinations of light sources and camera resolution Figure G is the distribution pattern of suspended particles in a "cylindrical" sample cell with an RSD value of 5x repeated measurements of 26%.In comparison, figure H is the distribution pattern of suspended particles in the sample cell "cubicle" with an RSD value of 5x repeated measurements of 8%.The lowest %RSD value indicates high consistency/repeatability/repeatability of measurement results.

Figure 4 .
Figure 4. Particle distribution patterns in (G) sample cylindrical cells and (H) sample cubicle cells

Figure 5 .
Figure 5. (I) 3D measurement chamber models; (J) A series of measurement chambers consisting of a light source (LED SMD5050), 32MP camera with C-mount Lense, DC source

Figure 6 .
Figure 6.Comparison of (K) original image with (L) threshold-image on BMC-LED SMD5050

Figure 7 .
Figure 7. Plot (M) results of quantification of the number of kaolin particles and (N) results of reading the turbidity value of the kaolin suspension solution at various concentrations

Figure 9 .
Figure 9. Profile of particle size in various suspension solution samples

Table 1 .
Combination of treatment on the development of measurement chamber 2.2.Determination of the shape and volume of the sample cellHACH2095000 cuvet square glass 25ml and HANNA HI731331 turbidity meter cuvette 10ml are two types of sample cells used in this study.The two types of cuvettes have different shapes and sizes.The HACH2095000 has a square cuvette base, resulting in a flat surface on the cuvet walls.

Table 2 .
The summary of limiting factors of image processing-based measurement techniques