A method and system for detecting microbial colonies based on electron micrographs
By comparing and verifying microbial colony detection images, identifying and verifying initial colony areas, and dynamically updating the baseline image, the problems of detection error and false positives and false negatives in existing technologies are solved, achieving efficient and accurate colony counting.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- BEIJING JIMMY KANE TECHNOLOGY CO LTD
- Filing Date
- 2026-03-18
- Publication Date
- 2026-06-12
AI Technical Summary
Existing technologies for microbial colony detection suffer from long culture cycles, cumbersome operations, and susceptibility to environmental vibrations, temperature changes, and light source aging, leading to image acquisition errors. Furthermore, aseptic reference image contamination or differences in imaging conditions can result in false positive or false negative results.
By comparing real-time images of bacterial and sterile samples with a baseline image, the difference region is identified as the initial colony region. Non-colony regions are excluded through image quality screening and verification. The baseline image is dynamically updated to eliminate equipment state drift and environmental interference, and the number of monomeric colonies is accurately counted.
It improves the accuracy and reliability of colony identification, overcomes the effects of systematic errors and random noise, and achieves efficient and accurate colony counting, especially under high-density culture conditions.
Smart Images
Figure CN122199500A_ABST
Abstract
Description
Technical Field
[0001] This invention belongs to the field of microbial detection technology, specifically relating to a method and system for detecting microbial colonies based on electron microscopy images. Background Technology
[0002] Currently, the detection of microorganisms typically employs artificial colony culture and counting. However, this method has limitations in practical applications, such as long culture cycles and cumbersome operations. Furthermore, artificial colony culture relies heavily on the experience of professionals, thus presenting certain obstacles when facing modern industries with high-throughput and high-timeliness detection needs. In contrast, automated microbial colony detection technology based on image analysis significantly improves detection efficiency by acquiring sample images and analyzing them using algorithms.
[0003] In practical applications, microscopic imaging equipment is easily affected by environmental vibrations, temperature changes, and light source aging during long-term operation or different batch operations. This can lead to unexpected deviations in the brightness, contrast, sharpness, and geometry of the acquired images, resulting in initial errors that are difficult to eliminate in all subsequent analysis steps. Furthermore, in schemes that identify colonies by comparing images of bacterial and sterile samples, the quality of the sterile reference image itself is crucial. However, the reference sample may be contaminated with minute impurities, or its imaging conditions may differ slightly from those of the sample being tested, leading to inaccurate benchmarks and misjudgments in subsequent difference analyses, resulting in a large number of false positives or false negatives.
[0004] To address the aforementioned problems, this invention provides a method and system for detecting microbial colonies based on electron microscopy images. Summary of the Invention
[0005] The purpose of this invention is to provide a method and system for detecting microbial colonies based on electron microscopy images, so as to solve the problem that the existing technology cannot fully assess the status of monitoring equipment and the impact of its normal operation on microbial colonies.
[0006] To achieve the above objectives, the present invention adopts the following technical solution:
[0007] A method for detecting microbial colonies based on electron microscopy images, comprising:
[0008] The real-time comparison image of the bacterial sample is compared with the baseline image of the sterile sample to identify the difference area as the initial colony area; the initial colony area is verified to filter out non-bacterial areas, thereby obtaining the verified colony area.
[0009] Colony analysis is performed based on the validated colony areas. The colony analysis includes: counting the total number of colonies; determining the number of individual colonies; and classifying the bacterial samples based on the total number of colonies and the number of individual colonies.
[0010] This involves comparing real-time images of bacterial samples with baseline images of sterile samples to identify areas of difference as initial colony regions, including:
[0011] Image quality screening is performed on real-time comparison images to remove images that are out of focus, blurry, or have a color cast greater than a preset threshold, thereby obtaining normal images;
[0012] The normal image is compared with the reference image to extract the difference between the two.
[0013] Preferably, the sterile sample is a blank carrier, and the bacterial sample is formed by loading the target monitoring substance onto the blank carrier.
[0014] Preferably, verifying the initial colony area includes:
[0015] If a non-colony region is determined to exist in the initial colony region, the real-time comparison image used when generating the initial colony region is updated to the baseline image. After updating the baseline image, a new real-time image of the bacterial sample is acquired as the new comparison image, and the identification step of the difference region is performed again.
[0016] Preferably, determining the presence of non-colony areas within the initial colony region includes:
[0017] In the baseline image, the test area corresponding to the initial colony area in space is extracted; when the brightness variation range of the test area is within the preset brightness range, and the similarity between the edge of the test area and the edge of the initial colony area is greater than the preset edge similarity threshold, the initial colony area is determined to be a non-colony area.
[0018] Preferably, determining the number of single-cell colonies includes:
[0019] For each colony image extracted from the verified colony area, determine whether it is a collection of interconnected colony units; if the colony image is a collection, calculate the number of colony units constituting the collection; if the colony image is not a collection, the number of its colony units is determined to be 1; sum up the number of colony units in all colony images to determine the number of individual colonies.
[0020] Preferably, classifying bacterial samples based on total bacterial count and number of single bacterial colonies includes:
[0021] The ratio of the number of individual bacterial colonies to the total number of colonies is calculated to determine the classification index. When the classification index is greater than the preset classification threshold, the bacterial sample is classified as a statistically acceptable sample. When the classification index is not greater than the preset classification threshold, the bacterial sample is classified as a sample requiring retesting.
[0022] Preferably, it further includes:
[0023] Multiple images of sterile samples are acquired at multiple time points; by comparing multiple images, it is determined whether there is a preset deviation in the image acquisition device, and a deviation reminder is output when a preset deviation is found.
[0024] This invention also discloses a microbial colony detection system based on electron microscopy images, comprising:
[0025] The image acquisition module is used to acquire images of sterile samples to establish a baseline image and to acquire real-time comparative images of bacterial samples.
[0026] The colony recognition module is used to compare real-time comparison images with reference images to identify the difference regions that serve as the initial colony areas.
[0027] The baseline correction module is used to verify the initial colony area to determine whether there is a non-colony area. When a non-colony area is determined to exist, the image acquisition module is controlled to update the corresponding real-time comparison image to the baseline image, thereby triggering the colony recognition module to re-execute the recognition operation.
[0028] The colony analysis module is used to perform colony analysis based on the verified colony areas output by the benchmark calibration module, in order to determine the total number of colonies and the number of single colonies, and to classify the bacterial samples.
[0029] The device self-test module is used to periodically control the image acquisition module to acquire multiple images of sterile samples and perform consistency analysis on the multiple images, so as to output a deviation reminder when it is determined that there is a preset deviation in the image acquisition device.
[0030] Preferably, the reference calibration module is configured as follows:
[0031] In the baseline image, extract the test area that spatially corresponds to the initial colony area;
[0032] When the brightness variation range of the test area is within the preset brightness range, and the similarity between the edge of the test area and the edge of the initial colony area is greater than the preset edge similarity threshold, the initial colony area is judged to be a non-colony area.
[0033] Preferably, the colony analysis module is configured as follows:
[0034] For each colony image extracted from the verified colony area, determine whether it is a collection of interconnected colony units; if the colony image is a collection, calculate the number of colony units constituting the collection; if the colony image is not a collection, the number of its colony units is determined to be 1.
[0035] The number of colony units in all colony images is summed to determine the number of individual colonies.
[0036] Beneficial effects
[0037] 1. This invention collects images of sterile samples at multiple time points for comparison to determine whether there is a preset deviation in the image acquisition device; at the same time, after identifying the initial colony area, the area is verified to determine whether there is a non-colony area. If a non-colony area exists, the real-time comparison image used when generating the area is updated to the reference image. Thus, through device self-checking and dynamic reference updating, interference introduced by device state drift or non-target particles in the environment is eliminated, improving the accuracy and reliability of colony identification, thereby overcoming the defects of existing technologies that are susceptible to systematic errors and random noise.
[0038] 2. After determining the total number of colonies, this invention determines whether each colony image is a collection formed by multiple interconnected colony units. If it is a collection, the number of colony units constituting the collection is calculated. Finally, the number of all colony units is summed to determine the number of individual colonies. This allows for precise subdivision and counting of densely growing or mutually adhering colonies, obtaining a more accurate number of individual colonies than the total number of colonies, thus solving the problem of inaccurate counting caused by colony fusion under high-density culture conditions. Attached Figure Description
[0039] Figure 1 This is a flowchart of the method of the present invention;
[0040] Figure 2 This is a system module diagram of the present invention. Detailed Implementation
[0041] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to specific embodiments. It should be understood that the specific embodiments described herein are merely for explaining the invention and are not intended to limit the scope of protection of the invention.
[0042] Example 1
[0043] See Figure 1 This embodiment provides a method for detecting microbial colonies based on electron microscopy images, specifically including the following steps:
[0044] S1. Obtain the target monitoring substance and prepare a sample;
[0045] Obtain the target monitoring substance whose microbial content needs to be measured. Prepare two sterile carriers that are identical in physical and chemical properties.
[0046] One sterile carrier, without any loaded substance, serves as a blank carrier and is designated as the sterile sample. The purpose of this sterile sample is to provide a stable, non-biologically active background reference for the image acquisition device in subsequent steps, used for calibration and to eliminate visual interference from the environment and the carrier itself. The other carrier is loaded with the target monitoring substance to form a bacterial sample.
[0047] Sterile and bacterial samples are cultured under the same preset test conditions to ensure that their physical environments are completely identical, laying the foundation for subsequent differential imaging analysis. The preset test conditions refer to environmental parameters that are set in advance to ensure the consistency of experimental conditions, such as constant temperature, humidity and culture time.
[0048] S2. Acquire baseline and comparison images, and identify the initial colony area;
[0049] After the culture begins, images of sterile samples under preset test conditions are acquired using an image acquisition device and these images are used as the baseline images. At the same time, images of bacterial samples under the same preset test conditions are acquired in real time at preset time intervals and these real-time acquired images are used as comparison images.
[0050] Furthermore, the aforementioned baseline images accurately record the visual characteristics of the vector itself and the culture environment under sterile conditions.
[0051] Image quality screening is performed on each acquired comparison image to ensure the accuracy of image analysis. Specifically:
[0052] The comparison images are processed using preset image quality evaluation rules to remove out-of-focus images, blurred images, and images with color cast greater than a preset color cast threshold caused by instantaneous instability of the optical system. The processing may include the following steps: calculating the high-frequency component energy of the image to evaluate sharpness, and comparing the color histogram of the image with a reference histogram of the standard color space to quantify the degree of color cast.
[0053] The comparison images that pass through this screening process are identified as normal images to eliminate transient errors caused by imaging hardware and avoid misidentifying image artifacts as bacterial colonies.
[0054] By performing pixel-level subtraction between the normal image and the reference image, a difference image is obtained. The difference image is then binarized to identify regions where the pixel value difference is greater than a preset difference threshold. These regions theoretically correspond to newly formed structures due to microbial growth. Therefore, these identified regions are determined as initial colony regions for subsequent verification.
[0055] Furthermore, the initial colony region is an unvalidated potential colony candidate region.
[0056] S3. Monitor the stability of the image acquisition device;
[0057] To ensure the long-term reliability of the testing process, it is necessary to eliminate the impact of potential physical offsets or performance degradation of the image acquisition device itself. Multiple images of the sterile sample are continuously acquired at several consecutive time points. Since there should be no growth changes on the sterile sample, theoretically, these images acquired at different time points should remain highly consistent. By comparing multiple images, it is possible to determine whether there is a pre-set deviation in the image acquisition device. The specific judgment steps are as follows:
[0058] For each of the multiple images, edge information is extracted using a preset image processing rule. This rule is achieved by calculating the difference in gray values between adjacent pixels. When the difference exceeds a preset gradient threshold, the location is marked as an edge point. The set of all edge points constitutes a structured edge feature map, which is the edge information used to characterize the contour information of the image.
[0059] The similarity between any two edge feature maps is calculated. This similarity is determined by calculating the proportion of overlapping edge points in the two images to the total number of edge points. If the similarity calculated between consecutively acquired images is lower than the preset similarity threshold, it indicates that the position, focal length, or lighting conditions of the image acquisition device may have drifted, i.e., a preset deviation is determined. At this time, a deviation reminder is output to prompt the operator to perform calibration.
[0060] S4. Verify the initial colony area and exclude non-colony areas;
[0061] The identified initial colony areas may contain real colonies, or they may contain non-biological artifacts that appear on the carrier over time, such as the inherent texture of the carrier material itself, air bubbles or sediment generated during the culture process. Therefore, each initial colony area needs to be verified to determine whether it is a non-colony area. The verification process is as follows:
[0062] Extract the region in the baseline image that corresponds perfectly to the initial colony region in spatial coordinates, and determine it as the test area;
[0063] The analysis of the above-mentioned test area specifically includes: calculating the standard deviation of the brightness values of all pixels in the test area to determine the range of brightness variation in the test area, which is used to quantify the complexity of texture or structure in the test area; simultaneously extracting the edges of the test area and calculating their similarity to the edges of the initial colony area. If the range of brightness variation is within a preset brightness range and the similarity is greater than a preset edge similarity threshold, then the initial colony area is judged to be a pre-existing structure, i.e., the initial colony area is judged to be a non-colony area; if a non-colony area is judged to exist, then a benchmark update strategy is executed, i.e., the comparison image used when generating the initial colony area is updated to a new benchmark image, so that the system incorporates the misjudged artifact structure into the current background, thereby automatically ignoring it in subsequent image comparisons;
[0064] After updating the baseline image, a real-time image of the bacterial sample is reacquired as a new comparison image, and the difference identification described above is performed again, so that the system can dynamically adapt to the gradual change of the background to improve the accuracy of detection.
[0065] Furthermore, if the brightness variation range is within a preset brightness range, it indicates that the region already possessed significant texture or structure before cultivation. If the similarity is greater than a preset edge similarity threshold, it indicates that the currently identified structure highly overlaps with the original background structure.
[0066] S5. Count the total number of colonies;
[0067] Within all initial colony regions that have passed the verification process and been confirmed as real colonies, the regions are processed using preset connected component analysis rules. The interconnected regions on the pixels are identified as independent image blocks, and each image block is an independent colony image. The total number of colony images is calculated and determined as the total number of colonies.
[0068] S6. Distinguish between single colonies and aggregates, and count the number of single colonies;
[0069] Under high-density culture conditions, multiple independent colonies may come into contact with each other and connect together due to growth, forming a visual aggregate. Therefore, it is necessary to judge each colony image to determine whether it is composed of a single colony or an aggregate formed by multiple interconnected colony units. The judgment steps are as follows:
[0070] Extract the boundary contours of each colony image. For any two spatially adjacent colony images, analyze the pixel connectivity of the gap region between their boundaries. If there is a continuous path connecting the two colony images in the gap region, consisting of pixels with a brightness higher than the average background brightness, then the two adjacent colony images are determined to be interconnected and merged into a set. This process is carried out iteratively until all colony images with pixel-level connectivity are assigned to the corresponding set.
[0071] If a colony image is not connected to any other colony image, then it is an independent colony unit. For colony images that are identified as aggregates, the number of colony units constituting them is calculated. The specific calculation process is as follows:
[0072] After smoothing the image of the aggregate, regions with local peak brightness values are identified within it. Each local peak region and its adjacent high-brightness regions are defined as independent colony units. By counting the number of local peak regions, the number of colony units constituting the aggregate can be determined. If the colony image is not an aggregate, its colony unit count is determined to be 1. The colony unit counts of all colony images are summed to determine the number of individual colonies, and the center position information of each colony unit in the comparison image is output.
[0073] S7. Classify the samples;
[0074] Based on the statistical results of the preceding steps, the samples are finally evaluated and classified. The ratio of the number of individual colonies to the total number of colonies is calculated, and this ratio is determined as the classification index. This index objectively reflects the dispersion or fusion of colonies. Based on the classification index, the samples are classified according to classification rules, the specific rules of which are as follows:
[0075] If the classification index is greater than the preset classification threshold, it indicates that the colony distribution is clear and independent, and easy to count accurately. At this time, the sample is classified as a statistically significant sample.
[0076] If the classification index is less than or equal to the preset classification threshold, it indicates that the sample may have excessively dense colonies or large-area fusion, resulting in inaccurate monomer counts. In this case, the sample is classified as a sample that needs to be retested, and a corresponding prompt is triggered.
[0077] Example 2
[0078] See Figure 2 This embodiment provides a microbial colony detection system based on electron microscopy images, including the following modules:
[0079] The image acquisition module is configured to acquire electron microscopic images for microbial colony detection.
[0080] In practical applications, this module images sterile samples to acquire and establish one or more baseline images; when a target monitoring substance is loaded onto a blank carrier to form a bacterial sample, this module performs real-time, continuous or periodic imaging of the bacterial sample to obtain a series of real-time comparative images. All acquired images are transmitted to subsequent modules for processing.
[0081] The colony identification module is configured to initially identify potential colony regions from real-time comparison images. This module receives real-time comparison images and a reference image provided by the image acquisition module.
[0082] In practical applications, image quality screening is performed on the received real-time comparison images to remove unqualified images caused by factors such as defocus, blur, or color cast. That is, images with defocus, blur, or color cast greater than the preset color cast threshold are filtered out, and only normal images with qualified quality are retained for subsequent analysis.
[0083] The normal image is compared with the reference image at the pixel level or feature level to accurately extract the difference regions between them. These difference regions are considered to be newly emerging targets due to microbial growth and are therefore identified as initial colony regions, which are then passed to the reference correction module. The comparison is preferably performed using algorithms such as image subtraction or background modeling.
[0084] The baseline calibration module verifies the authenticity of the initial colony areas and dynamically updates the baseline image to eliminate artifacts introduced by slow changes in the sample background. This module receives the initial colony areas output by the colony recognition module. For each initial colony area, this module performs a verification operation, specifically:
[0085] Extract the test area that spatially corresponds perfectly to the current initial colony area from the benchmark image, and analyze the image features of the test area. When a preset condition is met—that is, when the brightness variation range of the test area is within a preset brightness range, and the similarity between the edge contour of the test area and the edge contour of the initial colony area is greater than a preset edge similarity threshold—the initial colony area is determined to be a previously existing but unrecorded background feature, i.e., it is determined to be a non-colony area. When a non-colony area is determined to exist, this module executes a benchmark update strategy:
[0086] The image acquisition module controls the updating of the real-time comparison image used when generating the initial colony area to a new baseline image. Since this area has been confirmed as background, the current image more accurately reflects the true background condition.
[0087] After the baseline image is updated, the control image acquisition module triggers the colony recognition module, which uses the new baseline image to re-perform the recognition operation on the original real-time comparison image. If the initial colony area passes the verification, it is confirmed as a verified colony area and transmitted to the colony analysis module.
[0088] The colony analysis module performs qualitative and quantitative colony analysis based on the validated colony regions. This module receives the validated colony regions output from the benchmark calibration module, and its analysis process includes the following steps:
[0089] The total number of colonies is calculated by counting each independent, disconnected, validated colony region as an independent colony, thus obtaining the total number of colonies.
[0090] To determine the number of individual colonies, and to more accurately assess the growth units of microorganisms, this module performs morphological analysis on the colony images within each verified colony region to determine whether the colony image is an aggregate formed by multiple colony units adhering to or overlapping each other. If the colony image is determined to be an aggregate by image segmentation techniques such as concave point detection and watershed algorithms, the number of colony units constituting the aggregate is further calculated. If the colony image is not an aggregate, its number of colony units is determined to be 1. The number of colony units in all colony images is summed to obtain the number of individual colonies.
[0091] The module classifies samples containing bacteria by assessing their growth status based on statistical results, calculating the ratio of individual colony counts to the total number of colonies, and using this ratio as a classification index. This index is then compared to a preset classification threshold. When the index is greater than the threshold, it indicates that the colony distribution is relatively dispersed and easy to count accurately; therefore, the sample is classified as a statistically valid sample. Conversely, when the index is less than or equal to the threshold, it may mean that the colonies are too dense and adhered too tightly, making accurate counting difficult. In this case, the sample is classified as requiring retesting, and a prompt is sent to the user.
[0092] The device self-test module is used to monitor the stability and consistency of the image acquisition device to ensure the reliability of the test results.
[0093] In practical applications, the image acquisition module can be controlled to acquire multiple images of the same sterile sample at multiple different time points, either periodically or when the system is idle, to obtain multiple images; consistency analysis is then performed on these multiple images, for example by calculating the cross-correlation coefficient, mean square error, or structural similarity index between the images;
[0094] If the analysis results show that there are changes in the images that exceed the preset deviation, such as brightness drift, field of view shift, or decreased sharpness, it is determined that there is a preset deviation in the image acquisition device. At this time, a deviation reminder is output to notify the operator to check or calibrate the equipment.
Claims
1. A method for detecting microbial colonies based on electron microscopy images, characterized in that, include: The real-time comparison image of the bacterial sample is compared with the baseline image of the sterile sample to identify the difference area as the initial colony area. The initial colony area is validated to filter out non-colony areas, thereby obtaining the validated colony area; Colony analysis is performed based on the validated colony areas. The colony analysis includes: counting the total number of colonies; determining the number of individual colonies; and classifying the bacterial samples based on the total number of colonies and the number of individual colonies. This involves comparing real-time images of bacterial samples with baseline images of sterile samples to identify areas of difference as initial colony regions, including: Image quality screening is performed on real-time comparison images to remove images that are out of focus, blurry, or have a color cast greater than a preset threshold, thereby obtaining normal images; The normal image is compared with the reference image to extract the difference between the two.
2. The method for detecting microbial colonies based on electron microscopy images according to claim 1, characterized in that, Sterile samples are blank carriers, while bacterial samples are formed by loading the target monitoring substance onto a blank carrier.
3. The method for detecting microbial colonies based on electron microscopy images according to claim 1, characterized in that, Verification of the initial colony area includes: If a non-colony region is determined to exist in the initial colony region, the real-time comparison image used when generating the initial colony region is updated to the baseline image. After updating the baseline image, a new real-time image of the bacterial sample is acquired as the new comparison image, and the identification step of the difference region is performed again.
4. The method for detecting microbial colonies based on electron microscopy images according to claim 3, characterized in that, Determining the presence of non-colony regions within the initial colony region includes: In the baseline image, the test area corresponding to the initial colony area in space is extracted; when the brightness variation range of the test area is within the preset brightness range, and the similarity between the edge of the test area and the edge of the initial colony area is greater than the preset edge similarity threshold, the initial colony area is determined to be a non-colony area.
5. The method for detecting microbial colonies based on electron microscopy images according to claim 1, characterized in that, Determining the number of single-cell colonies includes: For each colony image extracted from the verified colony area, determine whether it is a collection of interconnected colony units; if the colony image is a collection, calculate the number of colony units constituting the collection; if the colony image is not a collection, the number of its colony units is determined to be 1; sum up the number of colony units in all colony images to determine the number of individual colonies.
6. The method for detecting microbial colonies based on electron microscopy images according to claim 1, characterized in that, Based on the total number of colonies and the number of single colonies, bacterial samples are classified as follows: The ratio of the number of individual bacterial colonies to the total number of colonies is calculated to determine the classification index. When the classification index is greater than the preset classification threshold, the bacterial sample is classified as a statistically acceptable sample. When the classification index is not greater than the preset classification threshold, the bacterial sample is classified as a sample requiring retesting.
7. The method for detecting microbial colonies based on electron microscopy images according to claim 1, characterized in that, Also includes: Multiple images of sterile samples were acquired at multiple time points; By comparing multiple images, it can be determined whether the image acquisition device has a preset deviation, and a deviation reminder will be output when a preset deviation is found.
8. A microbial colony detection system based on electron microscopy images, characterized in that, include: The image acquisition module is used to acquire images of sterile samples to establish a baseline image and to acquire real-time comparative images of bacterial samples. The colony recognition module is used to compare real-time comparison images with reference images to identify the difference regions that serve as the initial colony areas. The baseline correction module is used to verify the initial colony area to determine whether there is a non-colony area. When a non-colony area is determined to exist, the image acquisition module is controlled to update the corresponding real-time comparison image to the baseline image, thereby triggering the colony recognition module to re-execute the recognition operation. The colony analysis module is used to perform colony analysis based on the verified colony areas output by the benchmark calibration module, in order to determine the total number of colonies and the number of single colonies, and to classify the bacterial samples. The device self-test module is used to periodically control the image acquisition module to acquire multiple images of sterile samples and perform consistency analysis on the multiple images, so as to output a deviation reminder when it is determined that there is a preset deviation in the image acquisition device.
9. A microbial colony detection system based on electron microscopy images according to claim 8, characterized in that, The reference calibration module is configured as follows: In the baseline image, extract the test area that spatially corresponds to the initial colony area; When the brightness variation range of the test area is within the preset brightness range, and the similarity between the edge of the test area and the edge of the initial colony area is greater than the preset edge similarity threshold, the initial colony area is judged to be a non-colony area.
10. A microbial colony detection system based on electron microscopy images according to claim 8, characterized in that, The colony analysis module is configured as follows: For each colony image extracted from the verified colony area, determine whether it is a collection of interconnected colony units; if the colony image is a collection, calculate the number of colony units constituting the collection; if the colony image is not a collection, the number of its colony units is determined to be 1. The number of colony units in all colony images is summed to determine the number of individual colonies.