A parallel reaction state monitoring method and system

Abstract

The application provides a parallel reaction state monitoring method and system, and relates to the technical field of parallel reaction monitoring. The method comprises the following steps: acquiring image data of each reaction unit in the parallel reaction device, and identifying a reaction area of each reaction unit from the image data; extracting a visual feature from the reaction area of each reaction unit; acquiring a preset priority parameter and a preset rule, wherein the preset rule comprises a trigger condition and a corresponding acquisition frequency adjustment behavior; evaluating an image change activity of each reaction unit according to a change of the visual feature; determining an image acquisition frequency of each reaction unit according to the image change activity, the preset priority parameter and the preset rule; and performing image acquisition on each reaction unit according to the image acquisition frequency, obtaining image data and storing the image data. The application improves the overall efficiency and economy of parallel reaction monitoring.

Description

Technical Field

[0001] This application relates to the field of parallel reaction monitoring technology, and more specifically, to a method and system for monitoring the state of a parallel reaction. Background Technology

[0002] In fields such as drug discovery and chemical synthesis, high-throughput parallel reaction systems are widely used to accelerate research and development, enabling efficient exploration of reaction conditions and compound screening. Initially, monitoring of these reactions relied primarily on endpoint detection or periodic sampling for offline analysis. While these methods provide accurate data, they are time-consuming and labor-intensive, and cannot provide real-time dynamic information about the reaction process, leading to insufficient understanding of reaction kinetics and mechanisms. Summary of the Invention

[0003] This application aims to address at least one of the technical problems existing in the prior art. To this end, this application proposes a parallel reaction state monitoring method and system, which aims to significantly improve the overall efficiency and economy of parallel reaction monitoring while ensuring monitoring effectiveness.

[0004] In a first aspect, the parallel reaction state monitoring method provided in the embodiments of this application includes: Image data of each reaction unit in the parallel reaction apparatus is acquired, and the reaction region of each reaction unit is identified from the image data; Visual features are extracted from the reaction region of each of the reaction units; Obtain preset priority parameters and preset rules, wherein the preset rules include trigger conditions and corresponding sampling frequency adjustment behaviors; The image change activity of each of the response units is evaluated based on the changes in the visual features. The image acquisition frequency of each reaction unit is determined based on the image change activity, the preset priority parameter, and the preset rule. Based on the image acquisition frequency, images are acquired for each reaction unit to obtain image data and store the image data.

[0005] According to some embodiments of this application, determining the image acquisition frequency of each reaction unit based on the image change activity, the preset priority parameter, and the preset rule includes: Identify whether there are frequency specification conflicts among multiple preset rules for the same reaction unit; When a frequency specification conflict exists, the frequency specified by the preset rule with the highest priority is selected as the initial decision frequency according to the preset priority sequence. The preliminary decision frequency is compared with the preset priority parameter to obtain the comparison result; The image acquisition frequency of the reaction unit is determined based on the image change activity and the comparison results.

[0006] According to some embodiments of this application, extracting visual features from the reaction region of each of the reaction units includes: Acquire visible light images and images within a specific wavelength range for each of the aforementioned reaction units; The visible light features of the visible light image and the specific wavelength range features of the image of the specific wavelength range are extracted from the region of each of the reaction units.

[0007] According to some embodiments of this application, after extracting visual features from the region of each of the reaction units, the method further includes: Compare the visible light characteristics with the characteristics of the specific wavelength range to identify and quantify physical layer dynamics; Calculate the physical layer activity based on the physical layer dynamics; The corrected image change activity is calculated by combining the image change activity with the physical layer activity.

[0008] According to some embodiments of this application, evaluating the image change activity of each response unit based on changes in the visual features includes: The visual features of each of the aforementioned reaction units are subjected to multi-frame cumulative averaging processing; Calculate the rate of change of the cumulative average visual features within a continuous time window; The rate of change is compared with a preset low threshold. When the rate of change is consistently higher than the low threshold, it is determined that there is a visual change, and the rate of change that is consistently higher than the low threshold is recorded as the target rate of change. The image change activity of each reaction unit is evaluated based on the target change rate.

[0009] According to some embodiments of this application, determining the image acquisition frequency of each reaction unit by fusing the image change activity, the preset priority parameter, and the preset rule includes: Receive the desired image acquisition frequency for each of the reaction units; The desired image acquisition frequency for each reaction unit is summarized to obtain the summarized frequency; The aggregation frequency is compared with a preset resource limit. When the aggregation frequency exceeds the preset resource limit, the desired image acquisition frequency and the preset rule are adjusted according to the preset priority parameters and image change activity of each reaction unit to determine the image acquisition frequency of each reaction unit.

[0010] According to some embodiments of this application, determining the image acquisition frequency of each of the reaction units includes: Based on the preset reaction stage information, adjust the weight and sensitivity threshold of each visual feature; Based on the adjusted weights and sensitivity thresholds, calculate the contribution of each visual feature to the preset rule triggering condition; When the preset rule triggering condition is met, the image acquisition frequency of each reaction unit is determined according to the contribution degree.

[0011] According to some embodiments of this application, evaluating the image change activity of each response unit based on changes in the visual features includes: Identify the visual change regions of each of the aforementioned reaction units; By separating visually changing regions of different properties from the visually changing regions, the target changing region is obtained; Extract the feature changes of the target change region; The feature changes are weighted and combined according to preset weights to obtain the target result; Based on the target results, evaluate the image change activity of each of the reaction units.

[0012] According to some embodiments of this application, after obtaining the preset priority parameters and preset rules, wherein the preset rules include triggering conditions and corresponding sampling frequency adjustment behaviors, the method further includes: Identify frequency specification conflicts between the preset priority parameter and the preset rule; When a frequency specification conflict exists, the acquisition frequency is adjusted according to a preset conflict resolution strategy.

[0013] Secondly, embodiments of this application provide a parallel reaction state monitoring system, including: The first acquisition module is used to acquire image data of each reaction unit in the parallel reaction device and identify the region of each reaction unit from the image data; An extraction module is used to extract visual features from the regions of each of the reaction units; The second acquisition module is used to acquire preset priority parameters and preset rules, wherein the preset rules include triggering conditions and corresponding acquisition frequency adjustment behaviors; An evaluation module is used to evaluate the image change activity of each of the response units based on the changes in the visual features; A determination module is used to fuse the image change activity, the preset priority parameter, and the preset rule to determine the image acquisition frequency of each reaction unit; The image acquisition module is used to acquire images of each reaction unit according to the image acquisition frequency, obtain image data, and store the image data.

[0014] The technical solution according to the embodiments of this application has at least the following beneficial effects: This application provides a parallel reaction state monitoring method. It acquires image data of each reaction unit in a parallel reaction apparatus and identifies the reaction area, thereby extracting visual features. Based on this, and combining preset priority parameters and preset rules including trigger conditions and acquisition frequency adjustment behavior, the image change activity of each reaction unit is evaluated according to changes in visual features. Finally, based on the image change activity, preset priority parameters, and preset rules, the image acquisition frequency of each reaction unit is dynamically determined, and images are acquired and stored accordingly.

[0015] Additional aspects and advantages of this application will be set forth in part in the description which follows, and in part will be obvious from the description, or may be learned by practice of this application. Attached Figure Description

[0016] The accompanying drawings are used to provide a further understanding of the technical solutions of this application and constitute a part of the specification. They are used together with the embodiments of this application to explain the technical solutions of this application and do not constitute a limitation on the technical solutions of this application.

[0017] Figure 1 A flowchart illustrating a parallel reaction state monitoring method provided in one embodiment of this application; Figure 2 This is a schematic diagram of a parallel reaction state monitoring system provided in one embodiment of this application. Detailed Implementation

[0018] To make the objectives, technical methods, and advantages of this application clearer, the following detailed description is provided in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the scope of this application.

[0019] Based on the above, this application proposes a parallel reaction state monitoring method and system, aiming to improve the accuracy of parallel reaction state monitoring.

[0020] See Figure 1 , Figure 1 This is a flowchart illustrating a parallel reaction state monitoring method according to an embodiment of this application. The parallel reaction state monitoring method provided in this embodiment includes, but is not limited to, steps S110 to S150, which will be described in detail below.

[0021] S110. Acquire image data of each reaction unit in the parallel reaction apparatus, and identify the reaction region of each reaction unit from the image data; S120. Extract visual features from the reaction region of each reaction unit; S130. Obtain preset priority parameters and preset rules. The preset rules include trigger conditions and corresponding sampling frequency adjustment behaviors. S140. Evaluate the image change activity of each response unit based on changes in visual features; S150. Determine the image acquisition frequency of each reaction unit based on the image change activity, preset priority parameters, and preset rules. S160. Based on the image acquisition frequency, perform image acquisition on each reaction unit to obtain image data and store the image data.

[0022] To make the technical solution of this application easier and clearer to understand, some key terms involved will be explained first.

[0023] Parallel reaction apparatus refers to devices capable of simultaneously conducting multiple independent reactions, such as microplates or reactor arrays, where each reaction unit can be considered an independent reaction vessel. Image data refers to the visual information of the reaction unit acquired through optical devices (such as cameras), which can be still images or video frames. The reaction region refers to the specific spatial area in the image data corresponding to the actual chemical or physical reaction occurring. Visual features are quantifiable information extracted from the reaction region image, such as color, brightness, texture, shape, number of bubbles, and solid particle size; these features reflect the physicochemical changes during the reaction process. Image change activity is an indicator of the drastic changes in visual features within the reaction unit, reflecting the dynamics of the reaction. Preset priority parameters can be pre-defined importance levels for specific reaction units, specific reaction types, or specific experimental stages. Preset rules are a series of logical judgments containing trigger conditions and corresponding acquisition frequency adjustment behaviors, used to guide the dynamic adjustment of the image acquisition frequency. Image acquisition frequency refers to the number of times an image of the reaction unit is acquired per unit of time.

[0024] The core of the parallel reaction state monitoring method in this application lies in dynamically adjusting the image acquisition frequency based on the actual state of the reaction units. Specifically, this method first requires acquiring image data of each reaction unit in the parallel reaction apparatus. This can be achieved by mounting a high-resolution camera above the parallel reaction apparatus and configuring a corresponding lighting system. The camera periodically takes pictures of the entire reaction array to acquire image data. After acquiring the image data, it is necessary to identify the reaction region of each reaction unit from these image data. For example, image processing techniques, such as edge detection, region segmentation, or template matching, can be used to automatically identify the boundary of each independent reaction unit in the image and define its internal region as the reaction region.

[0025] Next, visual features are extracted from the reaction regions of each reaction unit. Extraction of visual features is crucial for evaluating the reaction state. For example, color features (such as RGB and HSV values), brightness features, texture features (such as gray-level co-occurrence matrix and local binary mode), and shape features (such as the outline, area, and perimeter of bubbles or solid particles) can be extracted from the image of the reaction region. The extraction of these features can be achieved through image processing algorithms, such as calculating the average color value of the reaction region to reflect color changes, or using morphological operations to identify and quantify bubbles or solid particles.

[0026] Simultaneously, the system needs to acquire preset priority parameters and preset rules. Preset priority parameters can be manually entered by the user based on the experimental purpose or the importance of the reaction; for example, setting the priority of a key reaction unit to "high" while setting other reaction units to "medium" or "low." Preset rules can be stored in the system's database beforehand; for example, a rule could be defined as: "If the color change rate of the reaction region exceeds X%, then increase the image acquisition frequency of that reaction unit to Y frames / second." These rules can be customized according to different reaction types, experimental stages, or monitoring needs.

[0027] Subsequently, the image change activity of each reaction unit is evaluated based on changes in visual features. This evaluation of image change activity forms the basis for dynamically adjusting the acquisition frequency. For example, the rate of change can be calculated by comparing the differences in visual features between consecutive frames. If the color, brightness, or number of bubbles in a reaction unit changes significantly within a short period, its image change activity is high, indicating that the reaction unit is in an active reaction phase. Conversely, if the changes in visual features are not significant, the activity is low.

[0028] After assessing the image change activity, the system determines the image acquisition frequency for each response unit by combining preset priority parameters and preset rules. This is a comprehensive decision-making process. For example, for a response unit with high image change activity, even if its priority parameter is not high, the system may increase its acquisition frequency according to preset rules to capture key changes. Conversely, for a response unit with low activity, even if its priority parameter is high, the system may appropriately reduce the acquisition frequency to save resources while meeting minimum monitoring requirements. When there are conflicting frequency assignments for the same response unit in the preset rules, the frequency specified by the preset rule with the highest priority can be selected as the initial decision frequency according to the preset priority sequence. Then, the initial decision frequency is compared with the preset priority parameters, and the final image acquisition frequency for that response unit is determined based on the image change activity and the comparison results.

[0029] Finally, based on the determined image acquisition frequency, images are acquired for each reaction unit, and image data is obtained and stored. For example, if the image acquisition frequency for a certain reaction unit is determined to be 5 frames per second, the image acquisition module will acquire images of that reaction unit at that frequency and store the acquired image data in the designated storage medium. For other reaction units, acquisition is performed according to their respective determined frequencies.

[0030] The parallel reaction state monitoring method proposed in this application dynamically adjusts the image acquisition frequency of each reaction unit in a parallel reaction apparatus by introducing image change activity, preset priority parameters, and preset rules. Compared with traditional fixed-frequency acquisition methods, this application can intelligently allocate image acquisition resources according to the actual progress and importance of the reaction. When a reaction unit is in an active state, the system can increase the acquisition frequency to ensure the capture of key reaction dynamic information; when a reaction unit is in an inactive state, the system can reduce the acquisition frequency to avoid collecting redundant data, thereby significantly reducing storage space occupation and the computational burden of subsequent data analysis. This adaptive monitoring strategy not only improves the efficiency and relevance of data acquisition but also provides more accurate and comprehensive data support for a deeper understanding and optimization of parallel reaction processes, greatly enhancing the monitoring capabilities and research efficiency of high-throughput experiments.

[0031] In one embodiment of this application, regarding the above step S150, the image acquisition frequency of each reaction unit is determined according to the image change activity, preset priority parameters and preset rules, including but not limited to steps S210 to S240. Each step will be described in turn below.

[0032] S210. Identify whether there is a conflict in the frequency specification of multiple preset rules for the same reaction unit; S220. When there is a frequency specification conflict, the frequency specified by the preset rule with the highest priority is selected as the initial decision frequency according to the preset priority sequence. S230. Compare the preliminary decision frequency with the preset priority parameters to obtain the comparison results; S240. Determine the image acquisition frequency of the reaction unit based on the image change activity and comparison results.

[0033] Specifically, when identifying frequency specification conflicts among multiple preset rules for the same reaction unit, the aim is to detect situations in parallel reaction devices where two or more preset rules are triggered simultaneously for a specific reaction unit, and the image acquisition frequencies suggested by these triggered preset rules are inconsistent. For example, one rule might suggest high-frequency acquisition, while another rule might suggest medium-frequency acquisition. When a frequency specification conflict exists, the frequency specified by the preset rule with the highest priority is selected as the initial decision frequency according to a preset priority sequence. The preset priority sequence can be predefined, for example, sorted according to rule importance, urgency, or user configuration. When a conflict is identified, the system will prioritize the image acquisition frequency specified by the preset rule with the highest priority, based on this sequence, as the initial basis for subsequent decisions. In practical applications, the initial decision frequency is compared with a preset priority parameter to obtain the comparison result. The preset priority parameter can be a baseline frequency, a safety upper or lower limit, or a weighting factor related to the importance of the reaction unit. By comparing the initial decision frequency with this parameter, the frequency can be further verified or adjusted to ensure it conforms to the overall monitoring strategy or resource constraints. The comparison results can indicate whether the initial decision frequency is within an acceptable range, or what degree of adjustment is needed. Ultimately, the image acquisition frequency of the reaction unit is determined based on the image change activity and the comparison results. The final determination of the image acquisition frequency involves comprehensively considering the current image change activity of the reaction unit with the aforementioned comparison results. For example, if the comparison results show that the initial decision frequency is too high or too low, it can be fine-tuned in conjunction with the image change activity; when the image change activity is high, even if the initial decision frequency is low, it may be increased, and vice versa, to achieve a dynamic balance.

[0034] Through the above technical solution, this application effectively solves the problem of inaccurate or inefficient image acquisition frequency determination caused by multiple rule conflicts in traditional methods. Specifically, by introducing conflict identification and priority arbitration mechanisms, the decision-making process for image acquisition frequency is ensured to be more reasonable and orderly in complex and ever-changing parallel reaction environments. Furthermore, by combining preset priority parameters and image change activity for comprehensive judgment, the final determined image acquisition frequency not only considers the importance of the rules but also takes into account the real-time dynamic changes of the reaction units and system resource constraints, thereby significantly improving the accuracy, real-time performance, and resource utilization efficiency of parallel reaction status monitoring.

[0035] In some preferred embodiments, a specific example is given below. Assume a parallel reaction apparatus contains multiple reaction units, one of which is performing a critical catalytic reaction. The system has two preset rules: Rule A (high priority) stipulates that when the reaction unit temperature exceeds a threshold, the image acquisition frequency should be adjusted to 10 frames per second; Rule B (low priority) stipulates that when the reaction unit's color change rate is below a certain threshold, the image acquisition frequency should be adjusted to 2 frames per second. Simultaneously, the preset priority parameter is set to a minimum acquisition frequency of no less than 3 frames per second. At a certain moment, the temperature of the reaction unit suddenly rises, triggering Rule A, which suggests a frequency of 10 frames per second. Simultaneously, its color change rate is also low, triggering Rule B, which suggests a frequency of 2 frames per second. At this point, the system identifies a frequency specification conflict for this reaction unit. According to the preset priority sequence, Rule A has a higher priority than Rule B, therefore, the 10 frames per second specified by Rule A is selected as the initial decision frequency. Subsequently, this initial decision frequency of 10 frames per second is compared with the preset priority parameter (minimum acquisition frequency of 3 frames per second). Since 10 frames per second is higher than 3 frames per second, the comparison result meets the requirements. Finally, considering the current image change activity of the reaction unit (e.g., image change activity is assessed as high due to temperature rise), the system ultimately determines the image acquisition frequency of the reaction unit to be 10 frames per second. In this way, even if rule conflicts exist, the system can make reasonable and efficient frequency decisions based on preset strategies and real-time status, ensuring effective monitoring of critical reaction states.

[0036] In one embodiment of this application, the step S120 described above, which extracts visual features from the reaction area of ​​each reaction unit, includes, but is not limited to, steps S310 to S320, which are described below.

[0037] S310. Acquire visible light images and images of a specific wavelength range for each reaction unit; S320, Extract visible light features of the visible light image and specific wavelength range features of the image from the region of each reaction unit.

[0038] Acquiring visible light images and images within specific wavelength ranges for each reaction unit refers to using different imaging devices or sensors to acquire multispectral or multimodal image data for each reaction unit in the parallel reaction apparatus. Visible light images typically reflect macroscopic information such as the color, shape, and brightness of the reaction unit, while images within specific wavelength ranges can reveal physical or chemical changes that are not easily detected under visible light, such as temperature distribution, absorption or emission spectra of specific substances, and fluorescence effects. For example, a visible light camera can be used to acquire conventional images of the reaction unit, while an infrared camera can be used to acquire temperature distribution images of the reaction unit, or an ultraviolet camera can be used to acquire fluorescence images of specific chemical substances.

[0039] Furthermore, extracting visible light features from the visible light image and specific wavelength range features from the image within a specific wavelength range of each reaction unit refers to processing the acquired image data of different wavelength ranges to obtain representative visual features for each. Visible light features may include, but are not limited to, color histograms, texture features (such as LBP and Gabor features), shape features, and edge information. Features within a specific wavelength range can be extracted based on their physical meaning; for example, for infrared images, temperature gradients and hotspot regions can be extracted; for ultraviolet or fluorescence images, intensity distribution and spectral peaks at specific wavelengths can be extracted. The extraction of these features aims to quantify the performance of the reaction unit under different spectra, providing multi-dimensional data support for subsequent reaction state assessment.

[0040] This application's solution captures the dynamic changes of reaction units from a more comprehensive perspective by acquiring visible light images and images within specific wavelength ranges, and extracting their corresponding visual features. Visible light images provide macroscopic, human-perceptible information about the reaction state, such as color changes, precipitation formation, and bubble generation. Images within specific wavelength ranges, such as infrared or ultraviolet images, provide deeper, non-visible physical or chemical information, such as reaction heat effects, the formation or consumption of specific products, and the presence of intermediate substances. By combining image features from these two or more different wavelength ranges, the limitations of a single image modality in information acquisition can be overcome, resulting in a richer and more accurate set of visual features, laying a solid foundation for subsequent assessment of image change activity.

[0041] In one embodiment of this application, after step S320 above, which extracts the visible light features of the visible light image and the specific wavelength range features of the image of a specific wavelength range from the region of each reaction unit, steps S410 to S430 are included, and each step is described below in sequence.

[0042] S410. Compare visible light characteristics with those of a specific wavelength range to identify and quantify physical layer dynamics; S420. Calculate the physical layer activity based on the physical layer dynamics; S430. By combining the image change activity and the physical layer activity, the corrected image change activity is calculated.

[0043] The comparison of visible light features with features within a specific wavelength range aims to reveal internal physical changes within the reaction unit by analyzing the differences or correlations between different spectral information. For example, features of a visible light image may reflect changes in color, transparency, or macroscopic morphology, while features of images within a specific wavelength range (such as infrared, ultraviolet, or fluorescence images) may reveal changes in temperature distribution, chemical component concentration, or microstructure. By comparing and analyzing these different types of features, physical layer dynamics that are difficult to detect with a single visual feature can be identified, such as phase transitions, precipitation, dissolution, or rearrangements of material structure caused by chemical reactions. Specific comparison methods may include, but are not limited to, pixel-level difference analysis, feature vector similarity calculation, multispectral image fusion algorithms, or machine learning-based pattern recognition.

[0044] Furthermore, based on the identified and quantified physical layer dynamics, physical layer activity can be calculated. Physical layer activity is a quantitative indicator of the intensity or rate of physical changes within a reaction unit. For example, if a violent phase transition or rapid precipitation process is identified, the corresponding physical layer activity will be calculated as a high value. This activity can be comprehensively evaluated based on factors such as the type, duration, and magnitude of change of the physical layer dynamics, for example, through weighted summation, statistical analysis, or a pre-defined model.

[0045] Therefore, the calculated physical layer activity is combined with the original image change activity to calculate the corrected image change activity. The original image change activity mainly reflects the macroscopic visual changes of the reaction unit in the visible light or specific wavelength range, while the physical layer activity provides deeper information on the physical process. By combining these two activities, a more comprehensive, accurate, and robust reaction state assessment index can be obtained. The combination method can employ weighted averaging, nonlinear fusion algorithms, or rule-based decision logic, where the weights can be adjusted according to the reaction type, monitoring target, or experimental experience.

[0046] The above technical solution significantly improves the accuracy and reliability of parallel reaction state monitoring. Especially when complex physical changes occur during the reaction process, and these changes are difficult to fully capture within a single spectral range, the solution provided in this application can offer more comprehensive information. This allows for a more accurate assessment of the true activity of each reaction unit, providing a more scientific basis for subsequent image acquisition frequency adjustments, thereby optimizing resource allocation and ensuring that critical reaction events are monitored in a timely and effective manner.

[0047] In some preferred embodiments, a specific example is given below. Assume a solution crystallization reaction is carried out in a parallel reaction apparatus, where crystals gradually precipitate from the solution within the reaction unit. Conventional image activity can be assessed primarily through changes in solution turbidity or macroscopic crystal growth. However, this information alone may not accurately reflect the onset time, rate, and microscopic changes in crystal morphology. The approach of this application can be specifically implemented as follows: First, visible light images and images within a specific wavelength range (e.g., polarized light images or near-infrared images) of the reaction unit are acquired. Visible light images are used to capture solution turbidity, color changes, and macroscopic crystal growth, while polarized light or near-infrared images may be more sensitive to changes in crystal microstructure, crystal form, or solute concentration in the solution. Next, by comparing visible light features (such as grayscale histograms and texture features) with features within the specific wavelength range (such as polarization degree and absorption peak intensity), physical layer dynamics can be identified and quantified. For example, a sudden increase in polarization degree may indicate the onset of crystal formation or a change in crystal form, while a change in near-infrared absorption peaks may indicate a decrease in solute concentration. These dynamics are quantified as physical layer activity, for example, assigned based on the rate or magnitude of polarization change. Finally, this physical layer activity is combined with image change activity based on the visible light image (e.g., turbidity change rate) to calculate the corrected image change activity. For example, when the visible light image shows little change but the polarized light image shows significant signs of crystal formation, the corrected image change activity increases due to the contribution of the physical layer activity, prompting the system to adjust the image acquisition frequency more promptly to capture key details of the crystallization process.

[0048] In one embodiment of this application, the above step S140, which evaluates the image change activity of each response unit based on changes in visual features, includes, but is not limited to, steps S510 to S540. Each step will be described in turn below.

[0049] S510. Perform multi-frame cumulative averaging on the visual features of each response unit. S520. Calculate the rate of change of visual features after cumulative averaging within a continuous time window; S530. The rate of change is compared with a preset low threshold. When the rate of change is continuously higher than the low threshold, it is determined that there is a visual change, and the rate of change that is continuously higher than the low threshold is recorded as the target rate of change. S540. Evaluate the image change activity of each reaction unit based on the target change rate.

[0050] Specifically, multi-frame cumulative averaging of the visual features of each response unit refers to statistically averaging the visual features of multiple frames of images acquired from the same response unit at different time points, such as arithmetic average or weighted average. The purpose is to smooth the visual feature data, eliminate instantaneous noise and occasional fluctuations, thereby obtaining a more stable visual feature representation that better reflects the true trend. The time window or number of frames for multi-frame cumulative averaging can be preset and adjusted according to the actual application scenario and response characteristics.

[0051] Furthermore, calculating the rate of change of the cumulatively averaged visual features within a continuous time window refers to quantifying the degree of change of visual features over time by comparing the cumulatively averaged visual features within the current time window with the cumulatively averaged visual features within one or more previous time windows. This rate of change can be calculated in various ways, such as calculating the difference in average feature values ​​within adjacent time windows, percentage changes, or more complex statistical indicators. The aim is to capture the dynamic changes in the state of the response unit, rather than merely static features.

[0052] Furthermore, the rate of change is compared with a preset low threshold. When the rate of change is consistently higher than the low threshold, a visual change is determined to exist, and the rate of change consistently higher than the low threshold is recorded as the target rate of change. The preset low threshold is used to distinguish between background noise, minor disturbances, and substantial visual changes. Only when the rate of change exceeds this low threshold, and this exceedance is sustained, is it considered a valid visual change. "Consistently higher than the low threshold" can be understood as the rate of change remaining above the low threshold for several consecutive time sampling points or time windows. This mechanism aims to avoid misjudging a high response due to brief, non-substantial fluctuations. The rate of change recorded as the target rate of change is a filtered and confirmed quantitative indicator representing a change in the true response state.

[0053] Finally, the image change activity of each reaction unit is evaluated based on the target change rate. This means that the evaluation of image change activity is no longer based on raw, potentially noisy, instantaneous visual features, but rather on a smoothed, quantized, and filtered target change rate. A higher target change rate generally indicates a higher image change activity of the reaction unit.

[0054] Through the above technical solution, this application overcomes the problem of inaccurate assessment of image change activity caused by instantaneous visual feature fluctuations or noise interference in traditional methods. Specifically, multi-frame cumulative averaging significantly improves the stability of visual features and reduces the false judgment rate; the calculation of the change rate and the introduction of a low threshold enable the system to distinguish substantial, continuous changes in reaction state from non-critical background noise or minor disturbances. Therefore, the assessed image change activity more realistically reflects the actual dynamics of the reaction unit, avoids unnecessary adjustments to the image acquisition frequency, improves the efficiency and reliability of the monitoring system, ensures timely and accurate response to critical reaction events, and thus optimizes the overall monitoring performance of the parallel reaction device.

[0055] In some preferred embodiments, it is assumed that the visual characteristics of a reaction unit (e.g., the average brightness value of the color in the reaction area) fluctuate slightly over a period of time. If the evaluation is based solely on the brightness changes of a single frame image, the "active" state may be frequently triggered due to slight changes in ambient light or sensor noise. However, according to the present application, the visual characteristics (e.g., brightness values) of the reaction unit are first subjected to multi-frame cumulative averaging, for example, averaging the brightness values ​​of the most recent 10 frames to smooth the data. Subsequently, the rate of change of the cumulative average brightness value over consecutive 5-second time windows is calculated. For example, if the average brightness value changes from 100 to 105 within 5 seconds, the rate of change is 5. This rate of change is compared with a preset low threshold (e.g., a rate of change greater than 3 is considered a valid change). If the rate of change is consistently higher than 3, a visual change is determined to exist, and this rate of change consistently higher than the low threshold is recorded as the target rate of change. For example, if the rate of change is higher than 3 for three consecutive 5-second windows, a continuous visual change is confirmed, and the image change activity of the reaction unit is evaluated based on these target rates of change. As a result, the system can effectively filter out transient noise and respond only to continuous and significant changes in the reaction state, thereby avoiding unnecessary adjustments to the image acquisition frequency and improving the accuracy and efficiency of monitoring.

[0056] In one embodiment of this application, the above step S150 determines the image acquisition frequency of each reaction unit based on the image change activity, preset priority parameters and preset rules, including but not limited to steps S610 to S630. Each step will be described in turn below.

[0057] S610, Receive the desired image acquisition frequency for each reaction unit; S620. Summarize the expected image acquisition frequencies for each reaction unit to obtain the summed frequencies; S630. Compare the summarization frequency with the preset resource limit. When the summarization frequency exceeds the preset resource limit, adjust the desired image acquisition frequency and preset rules according to the preset priority parameters and image change activity of each reaction unit to determine the image acquisition frequency of each reaction unit.

[0058] Specifically, the desired image acquisition frequency for each reaction unit refers to the fact that, in a parallel reaction device, each reaction unit can independently generate a desired image acquisition frequency based on its current reaction state, a preset monitoring strategy, or user needs. This desired frequency is personalized and reflects the intensity of the data acquisition demand of that reaction unit at a specific moment.

[0059] The desired image acquisition frequency of each reaction unit is summarized to obtain the aggregated frequency. This can be understood as aggregating the desired image acquisition frequencies generated by all reaction units to obtain the total image acquisition resource requirements of the entire parallel reaction device within a certain time period. This aggregated frequency is usually obtained by summation, taking the maximum value, or weighted averaging, with the aim of quantifying the overall resource load of the system.

[0060] In practical applications, comparing the aggregated frequency with preset resource limits means comparing the calculated aggregated frequency with the system's pre-set upper limit of resources available for image acquisition. These preset resource limits may include, but are not limited to, the processing power of the image processor, data storage bandwidth, network transmission bandwidth, or storage space. This comparison aims to determine whether the image acquisition requirements of all current reaction units are within the system's capacity.

[0061] When the aggregated frequency exceeds the preset resource limit, the desired image acquisition frequency and the preset rules are adjusted based on the preset priority parameters and image change activity of each reaction unit to determine the image acquisition frequency of each reaction unit. This means that when system resources are insufficient to meet the desired acquisition frequency of all reaction units, an intelligent resource allocation and adjustment mechanism needs to be activated. This mechanism comprehensively considers the preset priority parameters of each reaction unit (e.g., some reaction units may be marked as critical reactions requiring higher priority data protection) and image change activity (e.g., reaction units with drastic changes may require more frequent monitoring), and dynamically corrects the original desired image acquisition frequency and the preset rules that trigger the acquisition frequency adjustment. The adjustment process may include reducing the acquisition frequency of some low-priority or inactive reaction units, or temporarily disabling certain non-critical rules to ensure that high-priority or actively changing reaction units can obtain sufficient resources to determine the final image acquisition frequency.

[0062] This application's solution effectively addresses potential resource bottlenecks in parallel reaction monitoring by incorporating considerations of overall system resource constraints. First, by receiving and aggregating the desired image acquisition frequency for each reaction unit, the system can comprehensively grasp the current total resource requirements. Second, by comparing this aggregated frequency with preset resource limits, the system can promptly identify potential resource overload situations. Due to this comparison mechanism, when resource overload is detected, the system can intelligently adjust the desired acquisition frequency and preset rules based on the preset priority parameters and image activity levels of each reaction unit. This adjustment does not simply reduce all frequencies; instead, it strategically prioritizes the monitoring needs of critical or actively changing reaction units while appropriately lowering the frequencies of secondary or stable reaction units, thereby achieving optimal monitoring results with limited resources.

[0063] In one embodiment of this application, the determination of the image acquisition frequency of each reaction unit in step S150 includes, but is not limited to, steps S710 to S730, which will be described in turn below.

[0064] S710. Adjust the weight and sensitivity threshold of each visual feature according to the preset reaction stage information; S720. Based on the adjusted weights and sensitivity thresholds, calculate the contribution of each visual feature to the preset rule triggering conditions. S730. When the preset rule triggering conditions are met, the image acquisition frequency for each reaction unit is determined based on its contribution. Specifically, the preset reaction stage information refers to the division of the reaction process into several stages with different chemical or physical characteristics in a parallel reaction apparatus. For example, these stages may include the initiation stage, intermediate reaction stage, product formation stage, or termination stage. This information can be preset and stored using experimental data, theoretical models, or expert experience.

[0065] Adjusting the weight and sensitivity threshold of each visual feature refers to dynamically changing the importance and triggering conditions of different visual features in image change activity assessment and rule triggering based on the current stage of the reaction. For example, in the initial stage of the reaction, more attention might be paid to color change features, with a higher weight and a lower sensitivity threshold, to promptly capture the initiation of the reaction; while in the product formation stage, more attention might be paid to sedimentation or turbidity features, with their weights and sensitivity thresholds adjusted accordingly. Weights quantify the relative importance of a specific visual feature in the overall assessment, while the sensitivity threshold defines the degree of change required for that feature to be considered significant.

[0066] In practical applications, calculating the contribution of each visual feature to the triggering conditions of a preset rule refers to quantifying the degree to which changes in each visual feature affect the fulfillment of specific preset rule triggering conditions. For example, if the triggering condition of a preset rule is "color change exceeding X and temperature change exceeding Y," then both the color change feature and the temperature change feature contribute to the triggering of this rule. The contribution can be calculated based on the amount of change in the visual feature, its adjusted weight, and the comparison with a sensitivity threshold. For instance, when the amount of change in a visual feature exceeds its adjusted sensitivity threshold, its contribution will increase significantly.

[0067] Therefore, when the preset rule triggering conditions are met, the image acquisition frequency of each reaction unit is determined based on the contribution level. This means that the final image acquisition frequency depends not only on whether the rule is triggered, but also on the contribution level of each visual feature that caused the rule to be triggered, and is finely adjusted accordingly. For example, if the contribution of a certain key visual feature is very high, even if the contribution of other features is low, the acquisition frequency may be increased to a higher level to ensure detailed monitoring of this key change.

[0068] This application's solution introduces pre-defined reaction stage information and dynamically adjusts the weights and sensitivity thresholds of visual features accordingly. This allows the system to adaptively focus on the most critical visual changes at the current stage based on the actual progress of the reaction. It is precisely this stage-based adaptive adjustment that allows the contribution of each visual feature to the pre-defined rule triggering conditions to be calculated more accurately. By incorporating the contribution into the image acquisition frequency determination process, the system can more precisely identify and respond to visual changes that are significant to the current reaction stage, thereby avoiding over-acquisition in non-critical stages while ensuring timely and sufficient image data acquisition at critical stages.

[0069] In one embodiment of this application, the step S140 of evaluating the image change activity of each response unit based on changes in visual features includes: S810, Identify the visual change region of each response unit; S820. Separate visually changing regions of different natures within the visually changing region to obtain the target changing region; S830, Extract the feature changes in the target change region; S840. Based on the preset weights, the feature changes are weighted and combined to obtain the target result; S850. Based on the target results, evaluate the image change activity of each reaction unit.

[0070] Specifically, identifying visually changing regions for each response unit refers to detecting and locating sets or regions of pixels in an image that have undergone significant changes using image processing techniques such as background subtraction, inter-frame differencing, optical flow, or motion detection algorithms. These changes may include variations in brightness, color, texture, shape, or position. Visually changing regions can be understood as areas in an image whose pixel values ​​or local features differ significantly from the previous frame or background image.

[0071] Furthermore, separating visual change regions of different properties within the visual change regions to obtain target change regions refers to further analysis and classification of these regions after identification. Visual change regions of different properties can refer to visual changes caused by different physical or chemical phenomena, such as color changes, morphological changes, bubble formation, sediment formation, and liquid flow. The separation process can be achieved through image segmentation, feature clustering, pattern recognition, or machine learning algorithms, aiming to distinguish change regions closely related to the reaction process and possessing specific significance from other irrelevant or secondary changes, thereby obtaining target change regions. Target change regions are visual change regions that, after screening and classification, are considered to have high reference value for monitoring the reaction state.

[0072] Extracting the feature changes of the target change region refers to quantifying the degree of change of the identified and separated target change region over time. Feature changes may include, but are not limited to, changes in area, average brightness, color histogram, texture features (such as gray-level co-occurrence matrix), and shape descriptor. These feature changes aim to numerically characterize the dynamic properties of the target change region, providing a quantitative basis for subsequent activity assessment.

[0073] In practical applications, the target result is obtained by weighting the feature changes according to preset weights. This is because different types of visual changes or different feature changes may have different importance in monitoring reaction status, and this application introduces preset weights. These weights can be set based on domain expert knowledge, historical data analysis, or machine learning models to reflect the relative contribution of each feature change in assessing reaction activity. By multiplying the extracted feature changes with their corresponding pre-weights and summing the results, a comprehensive target result can be obtained, which can more comprehensively and accurately reflect the image change activity of the reaction unit.

[0074] Therefore, evaluating the image change activity of each reaction unit based on the target result means using the weighted combination of target results as the final evaluation criterion. The magnitude or trend of the target result directly determines the level of image change activity. For example, the larger the target result or the more drastic the change, the higher the image change activity, and vice versa. This evaluation method can more precisely capture and quantify subtle changes in the reaction process, thus providing a more reliable input for subsequent image acquisition frequency adjustment.

[0075] This application's solution, through refined identification, separation, and feature quantification of visual change regions, and the introduction of weighted combinations, effectively distinguishes between critical and non-critical visual changes during the reaction process, thus avoiding noise interference and misjudgments that may exist in traditional methods. It is precisely because different types of visual changes are differentiated and processed, and different weights are assigned according to their importance, that the final image change activity assessment result can more accurately reflect the true dynamics of the reaction, providing a more reliable and refined basis for subsequent image acquisition frequency adjustments.

[0076] In some preferred embodiments, a specific example is given below. Suppose that in a parallel reaction apparatus, a reaction unit is undergoing a liquid-phase reaction, during which two visual phenomena may occur: precipitation formation and solution color change.

[0077] First, the background subtraction algorithm is used to identify all areas in the image that have changed, i.e., areas of visual change.

[0078] Next, using image segmentation techniques, these visually changing regions are separated into "sedimentation regions" (manifested as changes in texture and brightness) and "color change regions" (manifested as changes in the intensity of specific color channels). These two separated regions are the target change regions.

[0079] Then, for the sedimentation area, the change in area and the change in average brightness are extracted as feature changes; for the color change area, the change in average intensity of a specific color channel is extracted as a feature change.

[0080] Subsequently, based on preset weights, for example, precipitation formation is more indicative of the reaction process, so its area change and brightness change are assigned higher weights (e.g., 0.6 and 0.2), while color change is assigned a lower weight (e.g., 0.2). These characteristic changes are weighted and combined to obtain a comprehensive target result.

[0081] Finally, the image activity of the reaction unit is evaluated based on the magnitude and trend of the target result. For example, if the target result consistently exceeds a certain threshold, it is considered to have high activity, requiring an increase in the image acquisition frequency; conversely, the acquisition frequency is reduced. In this way, the system can respond more intelligently to the actual changes in the reaction unit, ensuring that sufficiently dense image data is acquired at critical moments, while avoiding unnecessary frequent acquisitions during stable phases.

[0082] In one embodiment of this application, regarding the above step S130 of obtaining preset priority parameters and preset rules, the preset rules include triggering conditions and corresponding acquisition frequency adjustment behaviors, including but not limited to steps S910 to S920, each step will be described in turn below.

[0083] S910, Identify frequency specification conflicts between preset priority parameters and preset rules; S920. When a frequency specification conflict exists, the acquisition frequency is adjusted according to the preset conflict resolution strategy.

[0084] Specifically, frequency assignment conflict refers to a situation where one or more preset priority parameters and one or more preset rules act simultaneously on the same response unit, resulting in inconsistencies in the suggested or assigned image acquisition frequencies. For example, a preset priority parameter might indicate that the response unit has a low priority and recommend a lower acquisition frequency, while a preset rule, triggered by the detection of a specific response event, demands a higher acquisition frequency. The preset conflict resolution strategy can be understood as a series of predefined rules or algorithms used to determine the final image acquisition frequency to be adopted when a frequency assignment conflict is detected. In practical applications, this strategy can be based on a priority sequence, such as specifying that the priority of a preset rule is higher than that of a preset priority parameter, or vice versa; it can also be based on mathematical methods such as weighted averaging, taking the maximum or minimum value, to comprehensively consider and obtain a compromise or optimal acquisition frequency. Its purpose is to ensure that, under any circumstances, each response unit can be assigned a clear and reasonable image acquisition frequency.

[0085] This application's solution effectively resolves the frequency specification conflict that may arise from preset priority parameters and preset rules by introducing a conflict identification and resolution mechanism. Specifically, when determining the image acquisition frequency, the system first identifies whether there are inconsistencies in frequency specification caused by different sources (preset priority parameters and preset rules). Once such a conflict is identified, the system will no longer simply superimpose or randomly select frequencies, but will instead judge and adjust according to a preset conflict resolution strategy. For example, through a preset priority sequence, it can ensure that critical reaction event triggering rules take precedence over regular priority parameters, thereby guaranteeing the timely capture of important reaction changes; or, through comprehensive consideration, it can avoid excessive resource consumption or information omissions due to extreme requirements of a single factor. It is precisely because of this intelligent conflict handling capability that the process of determining the image acquisition frequency becomes more stable and reliable.

[0086] In some preferred embodiments, a specific example is given below. Assume a parallel reaction device contains multiple reaction units. For a particular reaction unit, its preset priority parameter might mark it as "routine monitoring," with a default acquisition frequency of 1 frame per minute. However, during the reaction, a preset rule is triggered, the trigger condition of which is "a drastic color change occurs within the reaction unit," and the corresponding acquisition frequency adjustment behavior is "increase the acquisition frequency to 5 frames per second." At this time, the system recognizes a frequency specification conflict between the preset priority parameter (1 frame per minute) and the preset rule (5 frames per second). To resolve this conflict, the system processes it according to a preset conflict resolution strategy. For example, if the preset conflict resolution strategy stipulates that "the priority of the preset rule is higher than the preset priority parameter," then the image acquisition frequency of this reaction unit will be adjusted to 5 frames per second. As another example, if the strategy is "select a higher frequency in case of conflict," then 5 frames per second will also be selected. In this way, even if there are multiple conflicting frequency specifications, the system can determine a final and reasonable image acquisition frequency based on a clear strategy. This ensures that image data can be acquired in a timely and high-frequency manner when there are drastic changes in the response, so as to carry out detailed analysis and avoid system decision-making confusion caused by conflicts.

[0087] See Figure 2 , Figure 2 This is a schematic diagram of a parallel reaction state monitoring system provided in one embodiment of this application. The parallel reaction state monitoring system 1000 includes: The first acquisition module 1010 is used to acquire image data of each reaction unit in the parallel reaction device and identify the region of each reaction unit from the image data; Extraction module 1020 is used to extract visual features from the regions of each reaction unit; The second acquisition module 1030 is used to acquire preset priority parameters and preset rules. The preset rules include triggering conditions and corresponding acquisition frequency adjustment behaviors. Evaluation module 1040 is used to evaluate the image change activity of each response unit based on changes in visual features; The determination module 1050 is used to fuse image change activity, preset priority parameters and preset rules to determine the image acquisition frequency of each reaction unit; The image acquisition module 1060 is used to acquire images of each reaction unit according to the image acquisition frequency, obtain image data, and store the image data.

[0088] The first acquisition module is used to acquire image data of each reaction unit in the parallel reaction apparatus and identify the region of each reaction unit from the image data. Specifically, this module can be configured with a high-resolution camera and a corresponding illumination system to periodically take pictures of the entire reaction array to acquire image data. After acquiring the image data, the module further uses image processing techniques, such as edge detection, region segmentation, or template matching methods, to automatically identify the boundary of each individual reaction unit in the image and define its internal region as the reaction region.

[0089] The extraction module is used to extract visual features from the regions of each reaction unit. Visual feature extraction is crucial for evaluating the reaction state. This module can extract color features (such as RGB values, HSV values), brightness features, texture features (such as gray-level co-occurrence matrix, local binary mode), and shape features (such as the outline, area, and perimeter of bubbles or solid particles) from the image of the reaction region. The extraction of these features can be achieved through image processing algorithms, for example, by calculating the average color value of the reaction region to reflect color changes, or by using morphological operations to identify and quantify bubbles or solid particles.

[0090] The second acquisition module is used to acquire preset priority parameters and preset rules. These preset rules include trigger conditions and corresponding acquisition frequency adjustment behaviors. This module can receive preset priority parameters manually input by the user according to the experimental purpose or the importance of the reaction. For example, setting the priority of a key reaction unit to "high" and other reaction units to "medium" or "low". Simultaneously, this module can read pre-stored preset rules from the system database. For example, a rule can be defined as: "If the color change rate of the reaction area exceeds X%, then increase the image acquisition frequency of that reaction unit to Y frames / second." These rules can be customized according to different reaction types, experimental stages, or monitoring needs.

[0091] The evaluation module assesses the image change activity of each reaction unit based on changes in the visual features. This assessment of image change activity forms the basis for dynamically adjusting the acquisition frequency. The module calculates the rate of change by comparing the differences in visual features between consecutive frames. If the color, brightness, or number of bubbles in a reaction unit changes significantly within a short period, its image change activity is high, indicating that the reaction unit is in an active reaction phase. Conversely, if the visual feature changes are not significant, the activity is low.

[0092] The determination module integrates the image change activity level, the preset priority parameters, and the preset rules to determine the image acquisition frequency for each response unit. This is a comprehensive decision-making process. For example, for a response unit with high image change activity, even if its priority parameter is not high, the module may increase its acquisition frequency according to the preset rules to capture key changes. Conversely, for a response unit with low activity, even if its priority parameter is high, the module may appropriately reduce the acquisition frequency to save resources while meeting the minimum monitoring requirements. When there is a conflict in the frequency specification for the same response unit among the preset rules, the module can select the frequency specified by the preset rule with the highest priority as the preliminary decision frequency according to the preset priority sequence. Then, the module compares the preliminary decision frequency with the preset priority parameters and, based on the image change activity level and the comparison result, finally determines the image acquisition frequency for that response unit.

[0093] The image acquisition module is used to acquire images of each of the reaction units according to the image acquisition frequency, obtain image data, and store the image data. For example, if the image acquisition frequency of a certain reaction unit is determined to be 5 frames per second, the module will acquire images of that reaction unit at that frequency and store the acquired image data in a designated storage medium. For other reaction units, acquisition is performed according to their respective determined frequencies.

[0094] The parallel reaction state monitoring system proposed in this application dynamically adjusts the image acquisition frequency for each reaction unit in a parallel reaction apparatus by introducing a first acquisition module, an extraction module, a second acquisition module, an evaluation module, a determination module, and an image acquisition module. Compared with traditional fixed-frequency acquisition methods, this system can intelligently allocate image acquisition resources according to the actual progress and importance of the reaction. When a reaction unit is in an active state, the system can increase the acquisition frequency to ensure the capture of key reaction dynamic information; when a reaction unit is in an inactive state, the system can reduce the acquisition frequency to avoid collecting redundant data, thereby significantly reducing storage space occupation and the computational burden of subsequent data analysis. This adaptive monitoring strategy not only improves the efficiency and relevance of data acquisition but also provides more accurate and comprehensive data support for a deeper understanding and optimization of parallel reaction processes, greatly enhancing the monitoring capabilities and research efficiency of high-throughput experiments.

Claims

1. A method for monitoring the state of a parallel reaction, characterized in that, Applied to parallel reaction devices, the method includes the following steps: Image data of each reaction unit in the parallel reaction device is acquired, and the reaction region of each reaction unit is identified from the image data, wherein the reaction region is the spatial range in which a chemical or physical reaction occurs; Visual features are extracted from the reaction region of each of the reaction units; Obtain preset priority parameters and preset rules, wherein the preset rules include trigger conditions and corresponding sampling frequency adjustment behaviors; The image change activity of each of the response units is evaluated based on the changes in the visual features. The image acquisition frequency of each reaction unit is determined based on the image change activity, the preset priority parameter, and the preset rule. Based on the image acquisition frequency, image acquisition is performed on each of the reaction units to obtain image data and store the image data; The step of determining the image acquisition frequency of each reaction unit based on the image change activity, the preset priority parameter, and the preset rule includes: Receive the desired image acquisition frequency for each of the reaction units; The desired image acquisition frequency for each reaction unit is summarized to obtain the summarized frequency; The aggregation frequency is compared with a preset resource limit. When the aggregation frequency exceeds the preset resource limit, the desired image acquisition frequency and the preset rule are adjusted according to the preset priority parameters and image change activity of each reaction unit to determine the image acquisition frequency of each reaction unit. Extracting visual features from the reaction region of each of the reaction units includes: Visible light images and images within a specific wavelength range are acquired for each of the reaction units; the specific wavelength range is either infrared or ultraviolet light. The visible light features of the visible light image and the specific wavelength range features of the image of the specific wavelength range are extracted from the reaction region of each of the reaction units; After extracting the visible light features of the visible light image and the specific wavelength range features of the image from the reaction region of each of the reaction units, the method further includes: Compare the visible light characteristics with the characteristics of the specific wavelength range to identify and quantify physical layer dynamics; Calculate the physical layer activity based on the physical layer dynamics; The corrected image change activity is calculated by combining the image change activity with the physical layer activity.

2. The method according to claim 1, characterized in that, The step of determining the image acquisition frequency of each reaction unit based on the image change activity, the preset priority parameter, and the preset rule includes: Identify whether there are frequency specification conflicts among multiple preset rules for the same reaction unit; When a frequency specification conflict exists, the frequency specified by the preset rule with the highest priority is selected as the initial decision frequency according to the preset priority sequence. The preliminary decision frequency is compared with the preset priority parameter to obtain the comparison result; The image acquisition frequency of the reaction unit is determined based on the image change activity and the comparison results.

3. The method according to claim 1, characterized in that, The step of evaluating the image change activity of each response unit based on the changes in the visual features includes: The visual features of each of the aforementioned reaction units are subjected to multi-frame cumulative averaging processing; Calculate the rate of change of visual features after cumulative averaging within a continuous time window; The rate of change is compared with a preset low threshold. When the rate of change is consistently higher than the low threshold, it is determined that there is a visual change, and the rate of change that is consistently higher than the low threshold is recorded as the target rate of change. The image change activity of each reaction unit is evaluated based on the target change rate.

4. The method according to claim 1, characterized in that, Determining the image acquisition frequency for each of the reaction units includes: Based on the preset reaction stage information, adjust the weight and sensitivity threshold of each visual feature; Based on the adjusted weights and sensitivity thresholds, calculate the contribution of each visual feature to the preset rule triggering condition; When the preset rule triggering condition is met, the image acquisition frequency of each reaction unit is determined according to the contribution degree.

5. The method according to claim 1, characterized in that, The step of evaluating the image change activity of each response unit based on the changes in the visual features includes: Identify the visual change regions of each of the aforementioned reaction units; By separating visually changing regions of different properties from the visually changing regions, the target changing region is obtained; Extract the feature changes of the target change region; The feature changes are weighted and combined according to preset weights to obtain the target result; Based on the target results, evaluate the image change activity of each of the reaction units.

6. The method according to claim 1, characterized in that, After obtaining the preset priority parameters and preset rules, where the preset rules include trigger conditions and corresponding sampling frequency adjustment behaviors, the method further includes: Identify frequency specification conflicts between the preset priority parameter and the preset rule; When a frequency specification conflict exists, the acquisition frequency is adjusted according to a preset conflict resolution strategy.

7. A parallel reaction state monitoring system, characterized in that, Applications in parallel reaction devices include: The first acquisition module is used to acquire image data of each reaction unit in the parallel reaction device and identify the reaction area of ​​each reaction unit from the image data, wherein the reaction area is the spatial range in which a chemical reaction or a physical reaction occurs. An extraction module is used to extract visual features from the reaction region of each of the reaction units; The second acquisition module is used to acquire preset priority parameters and preset rules, wherein the preset rules include triggering conditions and corresponding acquisition frequency adjustment behaviors; An evaluation module is used to evaluate the image change activity of each of the response units based on the changes in the visual features; The determination module is used to determine the image acquisition frequency of each reaction unit based on the image change activity, the preset priority parameter, and the preset rule; The step of determining the image acquisition frequency of each reaction unit based on the image change activity, the preset priority parameter, and the preset rule includes: Receive the desired image acquisition frequency for each of the reaction units; The desired image acquisition frequency for each reaction unit is summarized to obtain the summarized frequency; The aggregation frequency is compared with a preset resource limit. When the aggregation frequency exceeds the preset resource limit, the desired image acquisition frequency and the preset rule are adjusted according to the preset priority parameters and image change activity of each reaction unit to determine the image acquisition frequency of each reaction unit. An image acquisition module is used to acquire images of each of the reaction units according to the image acquisition frequency, obtain image data, and store the image data; Extracting visual features from the reaction region of each of the reaction units includes: Visible light images and images within a specific wavelength range are acquired for each of the reaction units; the specific wavelength range is either infrared or ultraviolet light. The visible light features of the visible light image and the specific wavelength range features of the image of the specific wavelength range are extracted from the reaction region of each of the reaction units; After extracting the visible light features of the visible light image and the specific wavelength range features of the image from the reaction region of each of the reaction units, the method further includes: Compare the visible light characteristics with the characteristics of the specific wavelength range to identify and quantify physical layer dynamics; Calculate the physical layer activity based on the physical layer dynamics; The corrected image change activity is calculated by combining the image change activity with the physical layer activity.

Images