Handheld water quality bacteria interpretation device and water quality bacteria inspection box

Abstract

The invention discloses a handheld water quality bacteria interpretation device and a water quality bacteria inspection box. The handheld water quality bacteria interpretation device comprises a control module, a fluorescence auxiliary module, an imaging system, a display module and a power supply module, the fluorescence auxiliary module is positioned below the imaging system; an ultraviolet mistaken touch prevention module is arranged below the fluorescent auxiliary module; and a water quality bacteria interpretation system is embedded in the control module. The water quality bacterial inspection box comprises a box body, and a handheld interpretation device, an inspection reagent consumable module and an auxiliary instrument module which are arranged in the box body, according to the application, conventional indexes of water quality microorganisms can be detected, intelligent interpretation of a detection result is realized, quantitative interpretation of a bacterial colony total number enzyme substrate method porous quantitative disc result and quantitative / qualitative interpretation of a total coliform / Escherichia coli detection result can be realized; and meanwhile, quantitative interpretation of test piece results such as the total number of bacterial colonies, coliform / escherichia coli and escherichia coli O157 can be realized.

Description

Technical Field

[0001] The present invention relates to the technical field of water quality detection, and particularly to a handheld water quality bacteria interpretation device and a water quality bacteria inspection box. Background Art

[0002] Detection is a key measure to prevent the spread of pathogenic microorganisms. To ensure the accuracy of detection results and the efficiency of the detection process, testers not only need to have a solid foundation of professional knowledge but also rich practical experience. The enzyme substrate detection methods for total bacterial count, total coliforms, and Escherichia coli in "Standard Test Methods for Drinking Water" GB / T 5750-2023 are clearly defined. Because of its simple operation, it significantly reduces the burden of detection work, so it is adopted by many water treatment companies and testing institutions. However, the detection means based on the enzyme substrate method need to be verified by professionals, which undoubtedly increases the difficulty of detection. And the existing water quality bacteria detection result interpretation devices usually set up dark rooms / chambers to avoid the interference of sunlight on fluorescence detection results, and generally have the characteristics of large volume and cumbersome operation procedures, which are not convenient for use in the field environment.

[0003] In view of this, the demand for a portable, efficient, and accurate water quality bacteria detection device is becoming increasingly urgent. To improve the accuracy and efficiency of detection, the wide application of artificial intelligence image recognition technology provides an innovative way to solve this problem. With advanced image processing algorithms and machine learning technologies, automatic recognition and classification of colony characteristics can be achieved, thereby reducing the dependence on professionals and improving the detection efficiency. Therefore, the purpose of the present invention is to provide a handheld water quality bacteria interpretation device and a water quality bacteria inspection box that can achieve intelligent visual interpretation of detection results. Summary of the Invention

[0004] The present invention discloses a handheld water quality bacteria interpretation device and a water quality bacteria inspection box, which can simultaneously achieve quantitative / qualitative interpretation of detection results such as total coliforms / Escherichia coli, and achieve quantitative interpretation of test strip results such as total bacterial count, coliforms / Escherichia coli, and Escherichia coli O157.

[0005] In the first aspect of the embodiments of the present invention, a handheld water quality bacteria interpretation device is disclosed, which includes a control module, and an imaging system, a fluorescence assistance module, an ultraviolet anti-misoperation module, a display module, and a power supply module electrically connected to the control module; the fluorescence assistance module is located below the imaging system; the ultraviolet anti-misoperation module is arranged below the fluorescence assistance module; the control module is embedded with a water quality bacteria interpretation system; the fluorescence assistance module is used to emit light of a specific wavelength to excite colonies to produce fluorescence; the imaging system is used to clearly image the colonies in the multi-well quantitative plate by using a high-resolution lens and an image sensor; the water quality bacteria interpretation system is used to judge the positive colony fluorescence hole image obtained by the imaging system and obtain an interpretation result. The handheld water quality bacteria interpretation device can realize the quantitative interpretation of the results of the colony total number enzyme substrate method multi-well quantitative plate and the quantitative / qualitative interpretation of the detection results of total coliforms / Escherichia coli, and at the same time can realize the quantitative interpretation of the results of test strips such as colony total number, coliforms / Escherichia coli, and Escherichia coli O157.

[0006] As an optional implementation manner, in the first aspect of the embodiments of the present invention, the water quality bacteria interpretation system is used to judge the positive colony fluorescence hole image obtained by the imaging system and obtain an interpretation result, including:

[0007] S1. Use the fluorescence assistance module and the imaging system to obtain a multi-well quantitative plate image; the multi-well quantitative plate image is an image with positive colony fluorescence holes; in the embodiments of the present invention, an 84-well quantitative plate is used for the multi-well quantitative plate;

[0008] S2. Perform feature extraction processing on the positive colony fluorescence hole image in the multi-well quantitative plate image to obtain a positive colony fluorescence hole feature map;

[0009] S3. Use a preset positive colony fluorescence hole interpretation model to identify and count the positive colony fluorescence hole feature map to obtain the colony type and the number of colonies;

[0010] S4. Call the colony total number MPN comparison table to compare the colony type and the number of colonies to obtain the colony total number detection result.

[0011] As an optional implementation manner, in the first aspect of the embodiments of the present invention, the performing feature extraction processing on the positive colony fluorescence hole image in the multi-well quantitative plate image to obtain a positive colony fluorescence hole feature map includes:

[0012] S21. Perform preprocessing on the positive colony fluorescence hole image in the multi-well quantitative plate image to obtain a primary fluorescence hole image;

[0013] S22. Perform detection processing on the primary fluorescence hole image to obtain a positive colony fluorescence hole feature map.

[0014] As an alternative implementation, in the first aspect of the embodiments of the present invention, the preprocessing of the positive colony fluorescence hole image in the porous quantitative disk image to obtain a primary fluorescence hole image includes:

[0015] S211. Using a flat-field correction algorithm, perform non-uniform illumination correction on the positive colony fluorescence hole image to obtain a first fluorescence hole image; through flat-field correction, reduce image artifacts and noise, and improve image uniformity;

[0016] S212. Using a rolling ball algorithm, perform background fluorescence elimination on the first fluorescence hole image to obtain a second fluorescence hole image;

[0017] S213. Use a band-pass filter to remove the reflected light of the UV wavelength (365 nm) in the second fluorescence hole image, and enhance the signal in the fluorescence emission wavelength range (400 - 700 nm) to obtain a third fluorescence hole image;

[0018] S214. Use Gaussian blur and median filtering to perform noise reduction processing on the third fluorescence hole image to obtain a fourth fluorescence hole image;

[0019] Through the processing of the above steps S211 - S214, the colony fluorescence hole image obtained excludes the interference of non-direct sunlight in the excitation light region of 300 - 400 nm.

[0020] S215. Perform image enhancement processing on the fourth fluorescence hole image to obtain a primary fluorescence hole image.

[0021] As an alternative implementation, in the first aspect of the embodiments of the present invention, the detection processing of the primary fluorescence hole image to obtain a positive colony fluorescence hole feature map includes:

[0022] S221. Perform color space conversion processing on the positive colony fluorescence hole image in the porous quantitative disk image to obtain a first detection image; specifically: convert the image from the RGB color space to the HSV (hue, saturation, value) or LAB (luminance, a-axis, b-axis) color space. These two color spaces are more suitable for color recognition and are used to better separate chromaticity and luminance;

[0023] S222. Perform threshold segmentation on the first detection image to obtain a second detection image; specifically: use a preset color threshold to segment the target color area, and adopt a threshold segmentation method to set a threshold to determine whether a pixel belongs to the target color; it is used to set a threshold range according to the known fluorescence color for luminance analysis;

[0024] S223. Perform connected component analysis on the second detection image to obtain a third detection image. Potential colony fluorescence pores can be identified through connected component analysis, and contour detection and blob analysis methods can be used.

[0025] S224. Perform morphological processing on the adhered colonies in the third detection image to obtain a fourth detection image. Specifically: use erosion and dilation to remove noise or connect adjacent regions, and apply opening / closing operation methods for processing to remove small noise points or connect broken regions.

[0026] S225. Perform screening processing on the adhered colonies in the fourth detection image to obtain a fifth detection image. This is used to complete the determination of colony fluorescence pores based on the expected colony fluorescence pore size and circular filtering.

[0027] S226. Perform edge elimination processing on the fifth detection image to obtain a feature map of positive colony fluorescence pore cavities. Specifically: crop the recognition area from the fifth detection image in a 1:1 ratio of length to width, and scale the image to a unified size of 224×224 pixels.

[0028] As an optional implementation manner, in the first aspect of the embodiments of the present invention, the method for constructing the preset positive colony fluorescence pore cavity interpretation model includes:

[0029] S31. Construct an initial positive colony fluorescence pore cavity interpretation model.

[0030] S32. Obtain a positive colony fluorescence pore cavity interpretation sample set. The positive colony fluorescence pore cavity interpretation sample set includes N colony feature map samples. The positive colony fluorescence pore cavity interpretation sample set can be obtained by repeatedly executing steps S1 to S2.

[0031] S33. Perform recognition processing on the positive colony fluorescence pore cavity interpretation sample set to obtain an identified positive colony fluorescence pore cavity interpretation sample set. Each sample colony in the identified positive colony fluorescence pore cavity interpretation sample set corresponds to identification information.

[0032] S34. Use the identified positive colony fluorescence pore cavity interpretation sample set to train the initial positive colony fluorescence pore cavity interpretation model to obtain the preset positive colony fluorescence pore cavity interpretation model.

[0033] As an optional implementation manner, in the first aspect of the embodiments of the present invention, the performing recognition processing on the positive colony fluorescence pore cavity interpretation sample set to obtain an identified positive colony fluorescence pore cavity interpretation sample set includes:

[0034] S331. Perform feature extraction processing on the samples in the positive colony fluorescence pore cavity interpretation sample set to obtain a first positive colony fluorescence pore cavity interpretation sample set.

[0035] S332. Classify the samples in the first positive colony fluorescence hole reading sample set by color to obtain a second positive colony fluorescence hole reading sample set;

[0036] S333. Mark the samples in the second positive colony fluorescence hole reading sample set, associate the identification information with the sample to obtain an identified positive colony fluorescence hole reading sample set. The annotation information can include multi-frame analysis such as the number and position of the lamp beads, and the detection stability is improved by using time continuity.

[0037] As an optional implementation manner, in the first aspect of the embodiments of the present invention, training the initial positive colony fluorescence hole reading model by using the identified positive colony fluorescence hole reading sample set to obtain a preset positive colony fluorescence hole reading model includes:

[0038] S341. Divide the identified positive colony fluorescence hole reading sample set into a training set and a test set;

[0039] S342. Select any sample from the training set, and use the initial positive colony fluorescence hole reading model to identify whether the any sample contains multiple colony targets to obtain a first judgment result;

[0040] If the first judgment result is yes, execute step S343; otherwise, execute step S344;

[0041] S343. Classify the colony types contained in the any sample, and execute step S344 for each type of colony;

[0042] S344. Use the initial positive colony fluorescence hole reading model to identify the varieties and numbers of all types of colonies in the any sample to obtain the colony type and number information corresponding to the any sample;

[0043] S345. Compare and process the colony type and number information corresponding to the any sample with the identification information associated with the sample to obtain the recognition accuracy rate of the initial positive colony fluorescence hole reading model;

[0044] S346. Loop to execute steps S342 to S345 to complete the recognition and statistics of all samples in the identified positive colony fluorescence hole reading sample set to obtain recognition result statistical information;

[0045] S347. Judge whether the recognition result statistical information is lower than a preset first recognition success rate to obtain a second judgment result;

[0046] When the second judgment result is yes, the parameters of the initial model for interpreting positive colony fluorescence holes are optimized and step S342 is executed; otherwise, step S348 is executed;

[0047] S348. Use the test set to test the initial model for reading positive colonies with fluorescent holes to obtain statistical information of the test recognition results; when the statistical information of the test recognition results is lower than the preset second recognition success rate, perform parameter tuning on the initial model for reading positive colonies with fluorescent holes, and execute step S342; otherwise, use the trained initial model for reading positive colonies with fluorescent holes as the preset positive colony fluorescence hole reading model.

[0048] It should be noted that the positive colony fluorescence hole interpretation model was tuned and optimized based on actual results to improve accuracy and robustness. Verification was performed using different test images to ensure that the algorithm could correctly identify colors in various situations.

[0049] As an optional implementation, in the first aspect of the embodiment of the present invention, the fluorescence auxiliary module is a 365nm ultraviolet excitation module.

[0050] The second aspect of an embodiment of the present invention discloses a water quality bacteria testing box, comprising: a box body, and a handheld water quality bacteria interpretation device, a testing reagent and consumables module, and an auxiliary equipment module arranged in the box body; the reagent and consumables are used for detecting conventional microbial indicators of water quality.

[0051] As an optional embodiment, in the second aspect of the embodiment of the present invention, the test reagent consumables module includes a total colony count enzyme substrate detection reagent, a porous quantitative plate, a total coliform group / Escherichia coli enzyme substrate detection reagent, a test bag and a total colony count reagent water.

[0052] As an optional embodiment, in the second aspect of the embodiment of the present invention, the auxiliary equipment module includes a portable alcohol lamp, matches, a pipette, a 10mL centrifuge tube, a 5mL pipette, an electronic balance, a weighing cup, a compressed towel, sterilized latex gloves, and disinfectant wipes; a charger, a charging cable, and a marker pen for a handheld water quality bacteria reading device.

[0053] Compared with the prior art, the embodiments of the present invention have the following beneficial effects:

[0054] A handheld water quality bacteria interpretation device and a water quality bacteria inspection box disclosed by the present invention combine ultraviolet excitation fluorescence, porous disk imaging, and a lightweight AI model to achieve fluorescence imaging interpretation under non-direct sunlight interference, and realize the "quantitative + qualitative" dual-mode detection of the detection results of the enzyme substrate method for total colony count, total coliforms, and Escherichia coli; relying on a preset positive colony fluorescence hole interpretation model, through enhanced small target feature extraction, the interpretation and recognition of dense porous fluorescence imaging are realized, and mAP≥95%; through practical application, the interpretation time is shortened by 67% compared with the traditional method, which is suitable for applications in the wild and emergency scenarios, and is portable and efficient. It can be seen that by using the technical solution provided by the present application, the quantitative interpretation of the results of the enzyme substrate method for the total colony count in the porous quantitative disk and the quantitative / qualitative interpretation of the detection results of total coliforms / Escherichia coli can be realized. At the same time, the quantitative interpretation of the test strip results of total colony count, coliforms / Escherichia coli, and Escherichia coli O157 can be realized, improving the accuracy and efficiency of detection. BRIEF DESCRIPTION OF THE DRAWINGS

[0055] Figure 1 A handheld water quality bacteria interpretation device disclosed in an embodiment of the present invention;

[0056] Figure 2 A schematic diagram of the LED ultraviolet lamp and circuit diagram used in an ultraviolet excitation fluorescence module disclosed in an embodiment of the present invention;

[0057] Figure 3 A schematic diagram of the composition of a water quality bacteria inspection box disclosed in an example of the present invention;

[0058] Figure 4 A schematic diagram of the initial model structure for interpreting positive colony fluorescence holes disclosed in an embodiment of the present invention;

[0059] Figure 5 A schematic diagram of the blue fluorescence of total colony count and Escherichia coli disclosed in an embodiment of the present invention;

[0060] Figure 6 A schematic diagram of the interpretation of total colony count disclosed in an embodiment of the present invention;

[0061] Figure 7 A schematic diagram of the interpretation of total coliforms disclosed in an embodiment of the present invention;

[0062] Figure 8 A schematic diagram of the interpretation of Escherichia coli disclosed in an embodiment of the present invention.

[0063] Reference Signs and Descriptions:

[0064] 1. Imaging system; 2. Ultraviolet excitation module; 3. Ultraviolet anti-misoperation module; 4. Box; 5. Handheld water quality bacteria interpretation device; 6. Total colony count enzyme substrate reagent; 7. 10 mL centrifuge tube; 8. Total coliform / Escherichia coli reagent; 9. 100 mL test bag; 10. Total colony count reagent water; 11. Electronic balance; 12. Weighing cup; 13. Compressed towel; 14. 84-well quantitative plate for total colony count; 15. 5 mL pipette; 16. Marker pen; 17. Charging cable; 18. Charger; 19. Alcohol lamp. Detailed implementation manners

[0065] The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all the embodiments. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without creative efforts shall fall within the protection scope of the present invention.

[0066] Embodiment 1

[0067] Please refer to Figure 1 . Figure 1 It is a schematic structural diagram of a handheld water quality bacteria interpretation device disclosed in an embodiment of the present invention.

[0068] As Figure 1 shown, a handheld water quality bacteria interpretation device disclosed in an embodiment of the present invention includes a control module, and an imaging system 1, a fluorescence assistance module 2, an ultraviolet anti-misoperation module 3, a display module and a power supply module that are electrically connected to the control module.

[0069] The control module is embedded with a water quality bacteria interpretation system; the water quality bacteria interpretation system is used to judge the positive colony fluorescence hole images obtained by the imaging system 1 and obtain an interpretation result; the fluorescence assistance module is used to emit light of a specific wavelength to excite the colonies to produce fluorescence; the ultraviolet anti-misoperation module 3 is used to prevent the operator from being accidentally injured by ultraviolet rays; the imaging system 1 is used to clearly image the colonies in the multi-well quantitative plate by using a high-resolution lens and an image sensor. The handheld water quality bacteria interpretation device can realize the quantitative interpretation of the results of the total colony count enzyme substrate method multi-well quantitative plate and the quantitative / qualitative interpretation of the total coliform / Escherichia coli detection results, and at the same time can realize the quantitative interpretation of the results of test strips such as total colony count, coliform / Escherichia coli, and Escherichia coli O157.

[0070] It should be noted that in the multi-well quantitative plate, each well contains a specific culture medium and chromogenic agent, and bacteria grow and reproduce in the well and react with the chromogenic agent. In the embodiment of the present invention, the multi-well quantitative plate adopts an 84-well quantitative plate.

[0071] In another optional embodiment, the water quality bacteria interpretation system is used to judge the positive colony fluorescent hole image obtained by the imaging system and obtain an interpretation result, including:

[0072] S1, using the fluorescence auxiliary module 2 and the imaging system 1 to obtain a multi-well quantitative plate image; the multi-well quantitative plate image is an image of fluorescent holes with positive bacterial colonies;

[0073] S2, performing feature extraction processing on the positive colony fluorescence hole image in the multi-hole quantitative plate image to obtain a positive colony fluorescence hole feature map;

[0074] S3, using a preset positive colony fluorescence hole interpretation model to identify and count the positive colony fluorescence hole characteristic diagram to obtain the colony type and colony number;

[0075] S4. Call the MPN comparison table for the total colony count, compare the colony types and colony numbers, and obtain the total colony count test result.

[0076] It should be noted that the water quality bacteria identification system uses advanced image processing algorithms and machine learning technology, and adopts artificial intelligence image recognition technology to realize automatic identification and classification of colony characteristics, thereby reducing dependence on professionals and improving the accuracy and efficiency of detection.

[0077] In another optional embodiment, the performing feature extraction processing on the positive colony fluorescent hole image in the multi-well quantitative plate image to obtain the positive colony fluorescent hole feature map includes:

[0078] S21, pre-processing the positive colony fluorescent hole image in the multi-well quantitative plate image to obtain a primary fluorescent hole image;

[0079] S22, performing detection processing on the primary fluorescent hole image to obtain a positive colony fluorescent hole characteristic map.

[0080] In yet another optional embodiment, the pre-processing of the positive colony fluorescent hole image in the multi-well quantitative plate image to obtain a primary fluorescent hole image includes:

[0081] S211. Using a flat-field correction algorithm, perform non-uniform illumination correction on the positive colony fluorescent hole image to obtain a first fluorescent hole image; through flat-field correction, image artifacts and noise are reduced and image uniformity is improved;

[0082] S212, using a rolling ball algorithm to eliminate background fluorescence from the first fluorescent hole image to obtain a second fluorescent hole image;

[0083] S213. Use a band-pass filter to remove the reflected light of the UV wavelength (365 nm) in the second fluorescence cavity image, enhance the signal in the fluorescence emission wavelength range (400 - 700 nm), and obtain the third fluorescence cavity image;

[0084] S214. Use Gaussian blur and median filtering to denoise the third fluorescence cavity image and obtain the fourth fluorescence cavity image;

[0085] S215. Perform image enhancement processing on the fourth fluorescence cavity image to obtain the primary fluorescence pore image;

[0086] Specifically: Use adaptive histogram equalization (CLAHE), gamma correction (to enhance weak fluorescence signals), and illumination compensation based on the Retinex theory for image enhancement.

[0087] It should be noted that through the above steps of processing, the obtained colony fluorescence pore image excludes the interference of non-direct sunlight in the excitation light region of 300 - 400 nm, and solves the influence of sunlight interference on colony recognition.

[0088] In another optional embodiment, the detection processing of the primary fluorescence pore image to obtain the positive colony fluorescence cavity feature map includes:

[0089] S221. Perform color space conversion processing on the positive colony fluorescence cavity image in the multi-well quantitative plate image to obtain the first detection image; Specifically: Convert the image from the RGB color space to the HSV (hue, saturation, value) or LAB (lightness, a-axis, b-axis) color space. These two color spaces are more suitable for color recognition and are used to better separate chrominance and luminance;

[0090] S222. Perform threshold segmentation on the first detection image to obtain the second detection image; Specifically: Use a preset color threshold to segment the target color region, adopt the threshold segmentation method, and set a threshold to determine whether a pixel belongs to the target color; It is used to set the threshold range according to the known fluorescence color for luminance analysis;

[0091] S223. Perform connected component analysis on the second detection image to obtain the third detection image; Identify potential colony fluorescence pores through connected component analysis. Contour detection and blob analysis methods can be used;

[0092] S224. Perform morphological processing on the adhered colonies in the third detection image to obtain the fourth detection image; It is used to remove small noise or connect broken regions, and the opening / closing operation method is applied for processing; Specifically: Use erosion and dilation to remove noise or connect adjacent regions;

[0093] S225. Screen the adherent colonies in the fourth detection image to obtain a fifth detection image; used to complete the determination of colony fluorescence pores according to the expected colony fluorescence pore size and circular filtering;

[0094] S226. Perform edge elimination processing on the fifth detection image to obtain a positive colony fluorescence pore feature map. Specifically: Crop the recognition area from the fifth detection image in a 1:1 ratio of length to width, and scale the image to a unified size of 224×224 pixels.

[0095] It should be noted that by performing the above detection processing on the image, the influence of blur and adhesion on colony recognition is excluded, effectively improving the recognition rate of fluorescence pore images.

[0096] In another optional embodiment, the method for constructing the preset positive colony fluorescence pore interpretation model includes:

[0097] S31. Construct an initial positive colony fluorescence pore interpretation model; preferably, the above initial positive colony fluorescence pore interpretation model uses an improved model based on YOLO (You Only Look Once), Faster R-CNN or SSD (Single Shot MultiBox Detector) to identify positive colony fluorescence pore images;

[0098] S32. Obtain a positive colony fluorescence pore interpretation sample set; the positive colony fluorescence pore interpretation sample set includes N colony feature map samples; the positive colony fluorescence pore interpretation sample set can be obtained by repeatedly executing steps S1 to S2; N is not less than 1000;

[0099] S33. Perform recognition processing on the positive colony fluorescence pore interpretation sample set to obtain a labeled positive colony fluorescence pore interpretation sample set; each sample colony in the labeled positive colony fluorescence pore interpretation sample set corresponds to labeling information;

[0100] S34. Use the labeled positive colony fluorescence pore interpretation sample set to train the initial positive colony fluorescence pore interpretation model to obtain the preset positive colony fluorescence pore interpretation model.

[0101] In another optional embodiment, the performing recognition processing on the positive colony fluorescence pore interpretation sample set to obtain a labeled positive colony fluorescence pore interpretation sample set includes:

[0102] S331. Perform feature extraction processing on the samples in the positive colony fluorescence pore interpretation sample set to obtain a first positive colony fluorescence pore interpretation sample set;

[0103] S332. Classify the samples in the first positive colony fluorescence hole reading sample set by color to obtain a second positive colony fluorescence hole reading sample set;

[0104] S333. Mark the samples in the second positive colony fluorescence hole reading sample set, associate the identification information with the sample, and obtain an identified positive colony fluorescence hole reading sample set. Label each image for subsequent training and evaluation. The labeling information can include multi-frame analysis such as the number and position of the lamp beads, and utilize temporal continuity to improve detection stability.

[0105] In another alternative embodiment, training the initial positive colony fluorescence hole reading model using the identified positive colony fluorescence hole reading sample set to obtain a preset positive colony fluorescence hole reading model includes:

[0106] S341. Divide the identified positive colony fluorescence hole reading sample set into a training set and a test set;

[0107] S342. Select any sample from the training set, and use the initial positive colony fluorescence hole reading model to identify whether the selected sample contains multiple colony targets to obtain a first judgment result;

[0108] If the first judgment result is yes, execute step S343; otherwise, execute step S344;

[0109] S343. Classify the types of colonies contained in the selected sample, and execute step S344 for each type of colony;

[0110] S344. Use the initial positive colony fluorescence hole reading model to identify the varieties and numbers of all types of colonies in the selected sample to obtain the colony type and number information corresponding to the selected sample;

[0111] S345. Compare and process the colony type and number information corresponding to the selected sample with the identification information associated with the sample to obtain the recognition accuracy rate of the initial positive colony fluorescence hole reading model;

[0112] S346. Loop through steps S342 - S345 to complete the identification and statistics of all samples in the identified positive colony fluorescence hole reading sample set, and obtain the identification result statistics information;

[0113] S347. Judge whether the identification result statistics information is lower than a preset first recognition success rate to obtain a second judgment result;

[0114] When the second judgment result is yes, the parameters of the initial model for interpreting positive colony fluorescence holes are optimized and step S342 is executed; otherwise, step S348 is executed;

[0115] S348. Use the test set to test the initial model for reading positive colonies with fluorescent holes to obtain statistical information of the test recognition results; when the statistical information of the test recognition results is lower than the preset second recognition success rate, perform parameter tuning on the initial model for reading positive colonies with fluorescent holes, and execute step S342; otherwise, use the trained initial model for reading positive colonies with fluorescent holes as the preset positive colony fluorescence hole reading model.

[0116] It should be noted that the positive colony fluorescence hole interpretation model was tuned and optimized based on actual results to improve accuracy and robustness. Verification was performed using different test images to ensure that the algorithm could correctly identify colors in various situations.

[0117] In another optional embodiment, the fluorescence auxiliary module 2 is a 365nm ultraviolet excitation module, and the ultraviolet excitation module uses an LED ultraviolet lamp as a light source. The LED ultraviolet lamp and the circuit diagram are as shown in FIG. Figure 2 shown.

[0118] In another optional embodiment, the fluorescence auxiliary module 2 includes a fluorescence excitation light source and a fluorescence filter; the fluorescence excitation light source adopts a high-brightness, narrow-band LED light source, and can be configured with LEDs of multiple wavelengths, such as 365nm, 488nm, etc. according to different detection requirements; the fluorescence filter is used to filter out light of non-target wavelengths and only allow fluorescence of specific wavelengths to pass through, thereby improving the accuracy of detection; the fluorescence auxiliary module 2 is connected to the control module through a PWM control circuit, and the control module controls the brightness of the fluorescence excitation light source by adjusting the duty cycle of the PWM signal.

[0119] In another optional embodiment, the fluorescence auxiliary module 2 adopts a 365nm ultraviolet excitation module, and the ultraviolet excitation module adopts an LED ultraviolet lamp as a light source. The LED ultraviolet lamp and the circuit diagram are as shown in FIG. Figure 2 shown.

[0120] In another optional embodiment, the imaging system 1 uses a high-resolution CMOS image sensor and an optical lens group. The lens group has an autofocus function and can clearly capture images of bacteria in water samples. The imaging system 1 is connected to the control module via a MIPI interface and transmits the collected image data to the control module in real time for processing and analysis.

[0121] In yet another alternative embodiment, the ultraviolet anti-mis-touch module 3 comprises an infrared sensor, a relay and an anti-mis-touch control circuit; the infrared sensor is used to detect whether an object approaches the detection window, and when an object is detected to approach, it transmits a signal to the control circuit, and the control circuit controls the relay to cut off the power supply of the ultraviolet light source to prevent the ultraviolet light from being mis-irradiated onto a person or other object; the ultraviolet anti-mis-touch module 3 is connected to the control module through a GPIO interface, and the control module monitors the state of the infrared sensor in real time and performs corresponding control.

[0122] In yet another alternative embodiment, the initial model for interpreting positive colony fluorescence holes adopts the structure as Figure 4 shown, Figure 4 and presents an initial model for interpreting positive colony fluorescence holes, including: a Backbone module, a Neck module and a Prediction module;

[0123] The Backbone module includes a Stem unit, a FusedMBConv unit, an MBConv unit, an ECA unit, and an SPPAF unit; the Backbone module is a deep convolutional neural network, including: a series of convolutional layers, pooling layers and activation functions to complete feature extraction; the training speed is improved by reducing the overhead of parameters and FLOPs.

[0124] The Neck module includes a Concat unit, a CSP2_1 unit, an upsampling unit and a CBL unit; the Neck module is used to perform upsampling, feature splicing or feature fusion on the feature maps of different scales extracted by the Backbone module to generate a more powerful feature representation.

[0125] The Prediction module includes a number of convolutional units, which are used to generate the class labels of the targets through convolutional operations on the feature map after feature fusion.

[0126] It should be noted that the positive colony fluorescence hole interpretation model provided in this embodiment enhances the extraction of small target features, combines high-level semantic features with low-level spatial information, further improves the detection accuracy, realizes the interpretation and recognition of dense porous fluorescence imaging, and further improves the detection accuracy.

[0127] It should be noted that for the physical detection test using the technical solution provided in this embodiment, please refer to Figures 5 to 8 , Figures 5 to 8 which is a schematic diagram showing the results interpretation of the total number of colonies, total coliforms and Escherichia coli in the present invention.

[0128] Embodiment 2

[0129] Please refer to Figure 3 .Figure 3 Schematic diagram of the structure of a water quality bacteria detection box disclosed in an embodiment of the present invention.

[0130] As Figure 3 shown, a water quality bacteria detection box disclosed in an embodiment of the present invention includes: a box body 4, and a handheld water quality bacteria interpretation device 5, a reagent and consumable module for inspection, and an auxiliary equipment module arranged in the box body; the reagents and consumables are used for detecting conventional microbial indicators of water quality.

[0131] In another alternative embodiment, the reagent and consumable module for inspection includes a total bacterial count enzyme substrate detection reagent 6, a total bacterial count 84-well quantitative plate 14, a total coliform / Escherichia coli enzyme substrate detection reagent 8, a 100 mL detection bag 9, and a total bacterial count reagent water 10.

[0132] In another alternative embodiment, the auxiliary equipment module includes a portable alcohol lamp 15, a 10 mL centrifuge tube 7, a 5 mL pipettor 15, an electronic balance 11, a weighing cup 12, a compressed towel 13, a disinfected wet wipe, a charger 18 for the handheld water quality bacteria interpretation device, a charging cable 17, a marker pen 16, a signature pen; it also includes matches, a pipette, sterilized latex gloves, etc.

[0133] Finally, it should be noted that: what is disclosed by a handheld water quality bacteria interpretation device and a water quality bacteria detection box disclosed in an embodiment of the present invention is only a preferred embodiment of the present invention, which is only used to illustrate the technical solution of the present invention, rather than limiting it; although the present invention has been described in detail with reference to the foregoing embodiments, those of ordinary skill in the art should understand that they can still modify the technical solutions recorded in the foregoing embodiments, or perform equivalent replacements on some of the technical features; and these modifications or replacements do not make the essence of the corresponding technical solutions deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A handheld water quality bacteria interpretation device, characterized in that, It includes a control module, an imaging system, a fluorescence assistance module, an ultraviolet anti-misoperation module, a display module, and a power module that are electrically connected to the control module; the fluorescence assistance module is located below the imaging system; the ultraviolet anti-misoperation module is arranged below the fluorescence assistance module; The control module is embedded with a water quality bacteria interpretation system; the fluorescence assistance module is used to emit light of a specific wavelength to excite colonies to produce fluorescence; the imaging system is used to clearly image the colonies in a multi-well quantification plate by using a high-resolution lens and an image sensor; the water quality bacteria interpretation system is used to judge the positive colony fluorescence hole images obtained by the imaging system and obtain an interpretation result.

2. The handheld water quality bacteria interpretation device according to claim 1, wherein The water quality bacteria interpretation system judges the positive colony fluorescence hole images obtained by the imaging system and obtains an interpretation result, including: S1. Obtain a multi-well quantification plate image by using the fluorescence assistance module and the imaging system; the multi-well quantification plate image is an image with positive colony fluorescence holes. S2. Perform feature extraction processing on the positive colony fluorescence hole images in the multi-well quantification plate image to obtain a positive colony fluorescence hole feature map. S3. Use a preset positive colony fluorescence hole interpretation model to identify and count the positive colony fluorescence hole feature map to obtain the colony types and the number of colonies. S4. Call the MPN comparison table of the total number of colonies to compare the colony types and the number of colonies to obtain the detection result of the total number of colonies.

3. The handheld water quality bacteria interpretation device according to claim 2, characterized in that The performing feature extraction processing on the positive colony fluorescence hole images in the multi-well quantification plate image to obtain a positive colony fluorescence hole feature map includes: S21. Perform preprocessing on the positive colony fluorescence hole images in the multi-well quantification plate image to obtain a primary fluorescence hole image. S22. Perform detection processing on the primary fluorescence hole image to obtain a positive colony fluorescence hole feature map.

4. The handheld water quality bacteria interpretation device according to claim 3, wherein The performing preprocessing on the positive colony fluorescence hole images in the multi-well quantification plate image to obtain a primary fluorescence hole image includes: S211. Perform non-uniform illumination correction on the positive colony fluorescence hole images to obtain a first fluorescence hole image. S212. Eliminate the background fluorescence of the first fluorescence hole image to obtain a second fluorescence hole image. S213. Use a band-pass filter to remove the reflected light of the UV wavelength in the second fluorescence hole image and enhance the signal in the fluorescence emission wavelength range to obtain a third fluorescence hole image. S214. Perform noise reduction processing on the third fluorescence hole image by using Gaussian blur and median filtering to obtain a fourth fluorescence hole image. S215. Perform image enhancement processing on the fourth fluorescence hole image to obtain a primary fluorescence hole image.

5. The hand-held water quality bacteria interpretation device according to claim 3, wherein The performing detection processing on the primary fluorescence hole image to obtain a positive colony fluorescence hole feature map includes: S221. Perform color space conversion processing on the positive colony fluorescence hole images in the multi-well quantification plate image to obtain a first detection image. S222. Perform threshold segmentation on the first detection image to obtain a second detection image. S223. Perform connected region analysis on the second detection image to obtain a third detection image. S224. Perform morphological processing on the adherent colonies in the third detection image to obtain a fourth detection image; S225. Perform screening processing on the adherent colonies in the fourth detection image to obtain a fifth detection image; S226. Perform edge elimination processing on the fifth detection image to obtain a fluorescence cavity feature map of positive colonies.

6. The handheld water quality bacteria interpretation device according to claim 2, wherein, The method for constructing the preset fluorescence cavity interpretation model for positive colonies includes: S31. Construct an initial fluorescence cavity interpretation model for positive colonies; S32. Obtain a sample set for fluorescence cavity interpretation of positive colonies; S33. Perform recognition processing on the sample set for fluorescence cavity interpretation of positive colonies to obtain an identified sample set for fluorescence cavity interpretation of positive colonies; each sample colony in the identified sample set for fluorescence cavity interpretation of positive colonies corresponds to identification information; S34. Use the identified sample set for fluorescence cavity interpretation of positive colonies to train the initial fluorescence cavity interpretation model for positive colonies to obtain the preset fluorescence cavity interpretation model for positive colonies.

7. The handheld water quality bacteria interpretation device according to claim 6, characterized in that, The performing recognition processing on the sample set for fluorescence cavity interpretation of positive colonies to obtain an identified sample set for fluorescence cavity interpretation of positive colonies includes: S321. Perform feature extraction processing on the samples in the sample set for fluorescence cavity interpretation of positive colonies to obtain a first identified sample set for fluorescence cavity interpretation of positive colonies; S322. Perform color classification on the samples in the first identified sample set for fluorescence cavity interpretation of positive colonies to obtain a second identified sample set for fluorescence cavity interpretation of positive colonies; S323. Perform marking on the samples in the second identified sample set for fluorescence cavity interpretation of positive colonies, and associate the identification information with the sample to obtain an identified sample set for fluorescence cavity interpretation of positive colonies.

8. The handheld water quality bacteria interpretation device according to claim 6, characterized in that, The using the identified sample set for fluorescence cavity interpretation of positive colonies to train the initial fluorescence cavity interpretation model for positive colonies to obtain the preset fluorescence cavity interpretation model for positive colonies includes: S341. Divide the identified sample set for fluorescence cavity interpretation of positive colonies into a training set and a test set; S342. Select any sample from the training set, and use the initial fluorescence cavity interpretation model for positive colonies to identify whether the any sample contains multiple colony targets to obtain a first judgment result; If the first judgment result is yes, then execute step S343, otherwise, execute step S344; S343. Classify the types of colonies contained in the any sample, and execute step S344 for each type of colony; S344. Use the initial fluorescence cavity interpretation model for positive colonies to identify the varieties and numbers of all types of colonies in the any sample to obtain the colony type and number information corresponding to the any sample; S345. Compare the colony type and number information corresponding to the any sample with the identification information associated with the sample to obtain the recognition accuracy rate of the initial fluorescence cavity interpretation model for positive colonies; S346. Loop and execute steps S342 - S345 to complete the recognition and statistics of all samples in the identified sample set for fluorescence cavity interpretation of positive colonies to obtain recognition result statistical information; S347. Judge whether the recognition result statistical information is lower than a preset first recognition success rate to obtain a second judgment result; When the second judgment result is yes, perform parameter tuning on the initial model for judging the fluorescence holes of positive colonies, and execute step S342; otherwise, execute step S348; S348. Use the test set to test the initial model for judging the fluorescence holes of positive colonies to obtain statistical information on the test recognition results; when the statistical information on the test recognition results is lower than the preset second recognition success rate, perform parameter tuning on the initial model for judging the fluorescence holes of positive colonies, and execute step S342; otherwise, use the trained initial model for judging the fluorescence holes of positive colonies as the preset model for judging the fluorescence holes of positive colonies.

9. The handheld water quality bacteria interpretation device according to claim 1, wherein The fluorescence auxiliary module is a 365nm ultraviolet excitation module.

10. A water quality bacteria inspection box, characterized in that, It includes a box body, and the handheld water quality bacteria judgment device described in any one of claims 1-9 is provided in the box body; the box body further includes: a reagent consumable module for inspection and an auxiliary equipment module; the handheld water quality bacteria judgment device is built in the box body.

Images