A monitoring lens data acquisition method and system

Abstract

The present application relates to the technical field of monitoring lens data collection, and particularly relates to a monitoring lens data collection method and system, the method comprising the following steps: setting a fixed reference object for optical performance monitoring; acquiring image information of the fixed reference object; evaluating the definition index of the image information; collecting environment state parameters related to optical performance degradation; judging whether the optical performance is degraded according to the definition index and the environment state parameters; identifying the optical performance degradation mode according to the degradation feature vector data packet; and transmitting the information of the optical performance degradation mode to a backend system. Through the technical scheme, the present application can realize real-time and objective judgment of the optical performance of the monitoring lens by setting the fixed reference object and acquiring the image information thereof, in combination with the definition index and the environment state parameters.

Description

Technical Field

[0001] This invention relates to the technical field of surveillance camera data acquisition, and specifically to a surveillance camera data acquisition method and system. Background Technology

[0002] In modern industrial production, data acquisition methods using surveillance cameras are crucial for ensuring smooth production processes, stable product quality, and a safe operating environment. These methods continuously acquire visual information by deploying industrial-grade cameras, providing data support for automated inspection, fault diagnosis, and process control. However, the complexity and harshness of industrial environments, especially the drastic fluctuations in ambient temperature, often pose hidden challenges to the long-term stable operation of high-precision optical equipment. Summary of the Invention

[0003] The purpose of this invention is to address the aforementioned shortcomings by proposing a method and system for acquiring surveillance camera data.

[0004] The present invention adopts the following technical solution:

[0005] A method for acquiring data from a surveillance camera, the method comprising the following steps:

[0006] Set up a fixed reference object for monitoring optical performance;

[0007] Acquire image information of a fixed reference object;

[0008] Evaluate the sharpness metrics of image information;

[0009] Collect environmental condition parameters related to optical performance degradation;

[0010] Determine whether the optical performance has deteriorated based on the sharpness index and environmental condition parameters;

[0011] Identify optical performance degradation patterns based on degradation feature vector data packets;

[0012] Information on optical performance degradation patterns is transmitted to the backend system.

[0013] Through this technical solution, this application can set a fixed reference object and obtain its image information, and combine it with the sharpness index and environmental state parameters to achieve real-time and objective judgment of the optical performance of the monitoring lens, and further identify the degradation mode. This effectively solves the problem of difficulty in capturing continuously changing and imperceptible local and non-uniform optical performance degradation in the prior art, and provides a basis for subsequent compensation or maintenance.

[0014] Furthermore, the steps for determining whether optical performance has deteriorated and identifying the mode of optical performance degradation include:

[0015] Divide the fixed reference point into multiple sub-regions;

[0016] Acquire image data from multiple sub-regions;

[0017] Calculate the multi-dimensional spatial focus features of the image data for each sub-region. The sharpness index is a multi-dimensional spatial focus feature, which includes the sub-region focus score, focus score gradient, focus score asymmetry index, and local sharpness of high-frequency edge features.

[0018] Collect multiple temperature values ​​and temperature change rates inside the camera, and calculate the cumulative operating time of the camera. The environmental state parameters include multiple temperature values, temperature change rates, and the cumulative operating time of the camera.

[0019] Based on multi-dimensional spatial focusing characteristics, multiple temperature values, temperature change rate, and camera cumulative operating time, determine whether the optical performance has deteriorated.

[0020] Encapsulate multi-dimensional spatial focusing features, multiple temperature values, temperature change rate, and camera cumulative running time to form a degraded feature vector data package;

[0021] Transmit the degraded feature vector data packet to the backend system;

[0022] The backend system identifies optical performance degradation patterns based on the degradation feature vector data packets;

[0023] The backend system performs image processing compensation or issues maintenance alerts based on the optical performance degradation mode.

[0024] Furthermore, the backend system identifies optical performance degradation modes based on the degradation feature vector data packet, including the following steps:

[0025] Receive degraded feature vector data packets;

[0026] Time series analysis is performed on multiple continuously received degraded feature vector data packets to calculate the change magnitude between each degraded feature vector data packet and the degraded feature vector data packet at the previous time step.

[0027] When the change amplitude is less than or equal to the preset stability threshold, the current degradation feature vector data packet is compared with the preset degradation mode feature vector to determine the matching optical performance degradation mode.

[0028] When the change exceeds the preset stability threshold, the feature vector accumulation window is activated to perform statistical analysis on the degraded feature vector data packets within the feature vector accumulation window and identify the optical performance degradation mode.

[0029] The confidence threshold for optical performance degradation pattern recognition is dynamically adjusted, and a mode switching suppression mechanism is introduced to enhance the stability of recognition.

[0030] Furthermore, the steps for dynamically adjusting the confidence threshold for optical performance degradation pattern recognition include:

[0031] After the backend system performs image processing compensation, the local sharpness of the compensated image is evaluated.

[0032] The improvement effect of the compensation is calculated based on the local sharpness of the compensated image and the local sharpness of the image before compensation.

[0033] When the improvement in compensation effect is greater than or equal to the preset effective compensation threshold, the confidence threshold for optical performance degradation pattern recognition is reduced.

[0034] When the improvement in compensation effect is less than the preset effective compensation threshold, the confidence threshold for optical performance degradation pattern recognition is increased.

[0035] Furthermore, when the change magnitude exceeds a preset stability threshold, a feature vector accumulation window is activated. Statistical analysis is performed on the degraded feature vector data packets within the feature vector accumulation window to identify optical performance degradation modes. The steps include:

[0036] Start the adaptively adjusted feature vector accumulation window;

[0037] Adjust the length of the feature vector accumulation window based on the internal changes of the degraded feature vector data packets within the feature vector accumulation window;

[0038] The weight of the degraded feature vector data packet is determined based on the similarity between the degraded feature vector data packet and the average feature vector within the feature vector accumulation window.

[0039] Based on the weights of the degraded feature vector data packets, a weighted statistical analysis is performed on the degraded feature vector data packets within the feature vector accumulation window to obtain the weighted statistics.

[0040] Matching is performed based on weighted statistics and preset optical performance degradation modes;

[0041] Calculate the confidence score of the match;

[0042] When the confidence scores of multiple matches are all higher than the preset matching threshold, analyze the evolution trend of degraded feature vector data packets within the feature vector accumulation window.

[0043] Identify optical performance degradation patterns.

[0044] Furthermore, the step of adjusting the length of the feature vector accumulation window based on the internal changes of the degraded feature vector data packets within the feature vector accumulation window includes:

[0045] Real-time monitoring of the instantaneous rate of change of degraded feature vector data packets within the feature vector accumulation window;

[0046] The instantaneous rate of change is compared with a preset dynamic rate of change threshold.

[0047] When the instantaneous rate of change equals the preset dynamic rate of change threshold, the length of the feature vector accumulation window remains unchanged;

[0048] When the instantaneous rate of change exceeds the preset dynamic rate of change threshold, the length of the feature vector accumulation window is shortened.

[0049] When the instantaneous rate of change is less than the preset dynamic rate of change threshold, the length of the feature vector accumulation window is extended;

[0050] The window length adjustment rules are preset according to the process characteristics, and the length adjustment of the feature vector accumulation window is triggered.

[0051] After adjusting the length of the cumulative window for the eigenvectors, the weighted statistics are recalculated.

[0052] Furthermore, the steps for introducing a mode-switching suppression mechanism include:

[0053] The difference between the degradation feature vector of the currently identified optical performance degradation mode and the degradation feature vector of the same optical performance degradation mode under historical stable conditions is monitored to obtain the deviation of the degradation feature vector.

[0054] When the difference exceeds the preset threshold for changes within the pattern, a short-term observation window is activated;

[0055] Analyze the changing trend of the degradation feature vector within the short-term observation window;

[0056] The suppression duration of the mode switching suppression mechanism is dynamically adjusted based on the changing trend and the deviation of the degradation feature vector.

[0057] Based on the changing trend and the deviation of the deteriorating feature vector, increase the confidence increment threshold for allowing mode switching;

[0058] Based on the feedback of the compensation effect after the backend system performs image processing compensation, the mode switching suppression mechanism is adjusted.

[0059] Furthermore, the method also includes an early warning mechanism step, the specific operation of which is as follows:

[0060] Real-time monitoring of the instantaneous rate of change of degraded feature vector data packets within a short observation window;

[0061] When the instantaneous rate of change of the degradation feature vector data packet within the short-term observation window shows a continuous acceleration or a jump near the set process node, and the trajectory of change is similar to the early evolution trajectory of the preset severe degradation mode, the early warning mechanism is triggered.

[0062] Furthermore, the steps for analyzing the changing trend of the degradation feature vector within a short-term observation window include:

[0063] Real-time collection of multiple degraded feature vector data packets within a short observation window;

[0064] Compare the differences between each pair of adjacent degraded feature vector data packets and calculate the difference value of the degraded feature vectors;

[0065] Based on the difference value, determine the magnitude of change between degraded feature vector data packets and identify the trend of degraded feature vector changes within a short observation window.

[0066] This application also discloses a surveillance camera data acquisition system applied to the above-mentioned surveillance camera data acquisition method, the system comprising:

[0067] The setup module allows you to set up a fixed reference object for optical performance monitoring.

[0068] The acquisition module acquires image information of a fixed reference object;

[0069] The evaluation module assesses the sharpness metrics of image information.

[0070] The acquisition module collects environmental condition parameters related to the degradation of optical performance;

[0071] The judgment module determines whether the optical performance has deteriorated based on the sharpness index and environmental condition parameters.

[0072] The identification module identifies optical performance degradation modes based on the degradation feature vector data packet.

[0073] The transmission module transmits information about the optical performance degradation mode to the backend system.

[0074] Through this technical solution, the monitoring lens data acquisition system provided in this application, through modular design, can effectively realize the above-mentioned monitoring lens data acquisition method, and provide hardware and software support for real-time monitoring, degradation judgment and pattern recognition of the optical performance of monitoring lenses in industrial environments. This solves the problem in the prior art that it is difficult to effectively capture continuously changing and imperceptible local and non-uniform optical performance degradation, and improves the intelligence level and reliability of the monitoring system.

[0075] This application, by combining fixed reference objects, multi-dimensional sharpness indicators, and environmental state parameters, can comprehensively and accurately perceive and identify optical performance degradation, especially in subtle degradation modes such as localized defocusing and aberrations, providing more refined judgments. This overcomes the shortcomings of existing technologies that rely on global indicators or static comparisons, avoiding the risk of backend analysis systems making judgments based on erroneous data due to minor degradation. This improves the accuracy, reliability, and intelligence level of monitoring data acquisition, providing a more solid data foundation for automated detection, fault diagnosis, and process control in industrial production.

[0076] To further understand the features and technical content of the present invention, please refer to the following detailed description and drawings of the present invention. However, the drawings provided are for reference and illustration only and are not intended to limit the present invention. Attached Figure Description

[0077] Figure 1 This is a flowchart of a method for acquiring data from a surveillance camera according to the present invention;

[0078] Figure 2 This is a schematic diagram of the structure of a surveillance camera data acquisition system according to the present invention. Detailed Implementation

[0079] The following specific embodiments illustrate the implementation of the present invention. Those skilled in the art can understand the advantages and effects of the present invention from the content disclosed in this specification. The present invention can be implemented or applied through other different specific embodiments, and various details in this specification can also be modified and changed based on different viewpoints and applications without departing from the spirit of the present invention. Furthermore, the accompanying drawings of the present invention are for simple illustrative purposes only and are not depictions of actual dimensions; this is stated in advance. The following embodiments will further describe the relevant technical content of the present invention in detail, but the disclosed content is not intended to limit the scope of protection of the present invention.

[0080] This embodiment provides a method and system for acquiring data from a surveillance camera, combined with... Figure 1 and Figure 2 As shown.

[0081] refer to Figure 1 A method for acquiring data from a surveillance camera, the method comprising the following steps:

[0082] Set up a fixed reference object for monitoring optical performance;

[0083] Acquire image information of a fixed reference object;

[0084] Evaluate the sharpness metrics of image information;

[0085] Collect environmental condition parameters related to optical performance degradation;

[0086] Determine whether the optical performance has deteriorated based on the sharpness index and environmental condition parameters;

[0087] Identify optical performance degradation patterns based on degradation feature vector data packets;

[0088] Information on optical performance degradation patterns is transmitted to the backend system.

[0089] In this context, a "fixed reference object" refers to an object or pattern with a known geometric shape and optical properties that is pre-set within the field of view of the monitoring lens. This reference object remains in position and state throughout the monitoring process, serving as a stable benchmark for evaluating the imaging quality of the monitoring lens. For example, the reference object could be a flat board printed with a high-contrast black-and-white checkerboard pattern, or a test chart containing stripes of a specific frequency.

[0090] "Image information" refers to visual data of a fixed reference object captured by a surveillance camera. This data usually exists in the form of digital images and contains information such as pixel brightness and color, which forms the basis for subsequent sharpness assessment.

[0091] A "sharpness index" is a parameter that quantifies the sharpness of an image. It reflects the sharpness of details and the clarity of edges in an image. For example, a sharpness index can be obtained by calculating the image's modulation transfer function (MTF) value, edge sharpness, contrast, or spatial frequency response.

[0092] "Environmental condition parameters" refer to external or internal environmental factors that may be related to the degradation of the optical performance of surveillance lenses. These parameters may include, but are not limited to, the camera's internal temperature, external ambient temperature, humidity, vibration frequency, and the camera's cumulative operating time. Collecting these parameters helps in analyzing the potential causes of optical performance degradation.

[0093] A "degradation feature vector data package" is a structured dataset containing key indicators describing the degradation state of the optical performance of a surveillance lens. This data package typically consists of sharpness indicators, environmental condition parameters, and other auxiliary information, used for subsequent degradation pattern identification.

[0094] "Optical performance degradation mode" refers to the specific manifestations or types of decline in the optical performance of a monitoring lens. For example, degradation modes may include, but are not limited to, focal length drift, localized defocusing, increased aberrations (such as coma, field curvature, and chromatic aberration), aggravated distortion, or lens fogging. Identifying these modes helps in taking targeted maintenance measures or performing image compensation.

[0095] The implementation environment of this application is typically an industrial production line, automated testing station, or security monitoring area, in which industrial-grade cameras are deployed and the long-term stable operation of the cameras needs to be monitored.

[0096] First, a fixed reference object needs to be set up for optical performance monitoring. This fixed reference object can be a test card with a high-contrast pattern, such as the ISO 12233 standard test chart, or a customized physical target containing specific geometric features (such as lines, circles, or grids). The reference object is placed at a fixed position within the field of view of the monitoring lens to ensure that its position and orientation remain consistent in the image when captured at different times. As a preferred implementation, the reference object can be integrated into a fixed workstation on the production line or installed on the wall of the monitored area.

[0097] Secondly, image information of a fixed reference object is acquired. This is typically achieved by having the surveillance camera automatically capture images of the fixed reference object at preset time intervals. For example, the surveillance system can be configured to trigger the camera to capture an image of the fixed reference object every hour or after a specific event (such as a drastic change in ambient temperature). The acquired image information can be raw RGB image data or an image stream that has undergone preliminary compression or encoding.

[0098] Next, the sharpness metrics of the image information are evaluated. After acquiring image information of a fixed reference object, these images need to be analyzed to quantify their sharpness. For example, a Fourier transform-based method can be used to calculate the energy distribution of the image at different spatial frequencies, thereby obtaining the MTF value as a sharpness metric. Sharpness can also be evaluated by analyzing the sharpness of edges in the image, such as calculating edge gradients or edge widths. Furthermore, methods based on wavelet transform or local contrast can also be used to obtain sharpness metrics.

[0099] Subsequently, environmental condition parameters related to optical performance degradation are collected. These parameters can be acquired using various sensors. For example, the internal temperature of the camera can be monitored in real time using a built-in temperature sensor; external ambient temperature and humidity can be obtained using an environmental sensor; and the camera's cumulative operating time can be statistically analyzed using system logs. The frequency of these parameter acquisitions can be consistent with the frequency of image information acquisition, or adjusted according to actual needs.

[0100] Based on the sharpness index and environmental condition parameters, determine whether optical performance has deteriorated. This step typically involves comparing the currently acquired sharpness index with a preset baseline or a historical sharpness index under normal conditions. Simultaneously, considering the changing trends of environmental condition parameters allows for a more accurate assessment of whether degradation has occurred. For example, if the sharpness index drops significantly, accompanied by an abnormal increase in the camera's internal temperature, it can be preliminarily determined that optical performance may have deteriorated. The judgment logic can be based on simple threshold comparisons or employ more complex machine learning models for classification.

[0101] Furthermore, based on the degradation feature vector data package, optical performance degradation patterns are identified. Once optical performance degradation is determined, the specific degradation pattern needs to be further identified. The degradation feature vector data package contains current sharpness indicators, environmental state parameters, and other relevant information. For example, if the degradation feature vector data package shows normal center sharpness but significantly reduced edge sharpness, it may be identified as a pattern of increased field curvature or coma. If the overall image is blurry and accompanied by a sharp increase in the internal temperature of the camera, it may be identified as focus drift or lens fogging. The identification process can be based on a preset rule base or can be performed using a trained pattern recognition algorithm (such as support vector machine or neural network).

[0102] Finally, information on optical performance degradation modes is transmitted to the backend system. Identified degradation mode information, such as "focus drift," "partial defocus," or "lens fogging," along with relevant sharpness indicators and environmental condition parameters, is encapsulated into data packets and transmitted to the backend system via the network. The backend system can be a central control server, a data analysis platform, or a maintenance management system. This information will be used for subsequent image processing compensation, maintenance alarm triggering, or fault diagnosis.

[0103] This application further proposes steps for determining whether optical performance has deteriorated and identifying optical performance degradation modes, including:

[0104] Divide the fixed reference point into multiple sub-regions;

[0105] Acquire image data from multiple sub-regions;

[0106] Calculate the multi-dimensional spatial focus features of the image data for each sub-region. The sharpness index is a multi-dimensional spatial focus feature, which includes the sub-region focus score, focus score gradient, focus score asymmetry index, and local sharpness of high-frequency edge features.

[0107] Collect multiple temperature values ​​and temperature change rates inside the camera, and calculate the cumulative operating time of the camera. The environmental state parameters include multiple temperature values, temperature change rates, and the cumulative operating time of the camera.

[0108] Based on multi-dimensional spatial focusing characteristics, multiple temperature values, temperature change rate, and camera cumulative operating time, determine whether the optical performance has deteriorated.

[0109] Encapsulate multi-dimensional spatial focusing features, multiple temperature values, temperature change rate, and camera cumulative running time to form a degraded feature vector data package;

[0110] Transmit the degraded feature vector data packet to the backend system;

[0111] The backend system identifies optical performance degradation patterns based on the degradation feature vector data packets;

[0112] The backend system performs image processing compensation or issues maintenance alerts based on the optical performance degradation mode.

[0113] Specifically, the fixed reference object is divided into multiple sub-regions to achieve localized and refined monitoring of optical performance. By analyzing different sub-regions independently, local degradation phenomena that may occur in the lens at different angles of view or positions can be effectively captured, such as edge blurring or dust adhesion in specific areas, rather than relying solely on overall sharpness assessment.

[0114] Among these features, multi-dimensional spatial focus features are used to more comprehensively quantify the sharpness of an image. Specifically, the sub-region focus score can be understood as a quantitative value measuring the overall sharpness of a specific sub-region of the image; the focus score gradient characterizes the rate and direction of sharpness change within a sub-region, helping to identify problems such as gradual blurring or tilted defocus; the focus score asymmetry index is used to evaluate the uniformity of sharpness distribution within a sub-region; for example, when the lens has coma or astigmatism, this index will exhibit significant asymmetry; and the high-frequency edge feature local sharpness focuses on the degree of preservation of high-frequency details in the image, and is more sensitive to minor defocusing or aberrations. The combined use of these features can provide richer and more accurate degradation information than a single sharpness index.

[0115] Furthermore, the environmental parameters were further refined into multiple internal temperature values, temperature change rates, and cumulative camera operating time. The internal temperature values ​​reflect the thermal expansion and contraction of the lens components and the heat dissipation of internal electronic components, both of which can affect the focusing accuracy and aberration performance of the optical system. The temperature change rate indicates whether rapid temperature fluctuations occur inside the lens, which may lead to condensation on the lens surface or changes in internal stress, resulting in instantaneous or cumulative optical performance degradation. The cumulative camera operating time serves as a crucial indicator of the overall aging and wear of the lens, closely related to long-term degradation modes such as material fatigue and loosening of mechanical structures. The collection of these parameters aims to provide more targeted environmental background information for identifying optical performance degradation modes.

[0116] Therefore, multi-dimensional spatial focusing features, multiple temperature values, temperature change rates, and the camera's cumulative operating time are encapsulated to form a degradation feature vector data package. This data package integrates the degradation characteristics of the image itself and the potential environmental factors leading to degradation, providing comprehensive input for the backend system to accurately identify degradation patterns. Subsequently, this degradation feature vector data package is transmitted to the backend system. Upon receiving this data package, the backend system can identify the specific optical performance degradation pattern based on the rich information it contains, such as degradation caused by thermal defocusing, lens contamination, mechanical loosening, or electronic noise. Once the degradation pattern is identified, the backend system can perform corresponding image processing compensation based on the characteristics of the pattern, such as deblurring and color correction to improve image quality, or issue a maintenance alarm to prompt maintenance personnel to conduct on-site inspection or replacement, thereby achieving intelligent management and maintenance of the surveillance lens performance.

[0117] This application overcomes the problem that traditional single sharpness indicators cannot capture local degradation by introducing sub-region division of a fixed reference object and calculating multi-dimensional spatial focusing features of each sub-region. Specifically, features such as sub-region focus score, focus score gradient, focus score asymmetry index, and local sharpness of high-frequency edge features can quantify the image sharpness from different angles, such as identifying subtle degradation manifestations like edge blurring, non-uniform defocusing, or specific aberrations. Simultaneously, by collecting more specific environmental parameters such as multiple temperature values, temperature change rates, and cumulative camera operating time, this application can establish a correlation between optical performance degradation and specific environmental factors or operating states. For example, rapid temperature changes may cause lens condensation or thermal defocusing, while cumulative operating time is related to lens aging or mechanical wear. These detailed image features and environmental parameters are encapsulated into a degradation feature vector data package, enabling the backend system to perform pattern recognition based on more comprehensive information. Therefore, the backend system can not only determine whether degradation exists but also identify specific degradation patterns, such as defocusing due to thermal expansion, local blurring caused by dust adhesion, or lens aging due to long-term operation. This refined recognition capability makes subsequent image processing compensation or maintenance alerts more targeted and effective.

[0118] In some preferred embodiments, it is assumed that the optical performance of a surveillance camera deployed in an outdoor environment may deteriorate due to various factors after a period of operation. For example, when the temperature rises sharply during the day, the internal components of the lens may experience thermal expansion, causing focus shift in local areas. Conventional methods may only detect a slight decrease in overall sharpness, but it is difficult to pinpoint the problem. However, according to the solution of this application, a fixed reference object is divided into multiple sub-regions. When thermal expansion causes slight defocusing in a specific sub-region (e.g., the upper left corner of the image), the focus score of that sub-region may decrease, and the focus score gradient and asymmetry index may show specific change patterns. For example, the focus score asymmetry index may indicate the presence of coma or astigmatism in that region. At the same time, the temperature value and temperature change rate inside the camera are acquired in real time and show a significant increase synchronized with the optical performance degradation. These image features and environmental parameters are encapsulated into a degradation feature vector data packet and transmitted to the back-end system. By analyzing this degradation feature vector data packet, the back-end system can accurately identify the degradation pattern of "local thermal defocusing". Based on this identification result, the backend system can perform targeted image processing compensation, such as local deblurring of the affected sub-region, or issue a maintenance alarm that "there is a risk of thermal defocusing in the upper left corner of the lens," guiding maintenance personnel to check or adjust, thereby avoiding monitoring blind spots or information loss caused by blurring, and realizing early warning and precise intervention for potential degradation problems.

[0119] The backend system identifies optical performance degradation modes based on the degradation feature vector data packet, including the following steps:

[0120] Receive degraded feature vector data packets;

[0121] Time series analysis is performed on multiple continuously received degraded feature vector data packets to calculate the change magnitude between each degraded feature vector data packet and the degraded feature vector data packet at the previous time step.

[0122] When the change amplitude is less than or equal to the preset stability threshold, the current degradation feature vector data packet is compared with the preset degradation mode feature vector to determine the matching optical performance degradation mode.

[0123] When the change exceeds the preset stability threshold, the feature vector accumulation window is activated to perform statistical analysis on the degraded feature vector data packets within the feature vector accumulation window and identify the optical performance degradation mode.

[0124] The confidence threshold for optical performance degradation pattern recognition is dynamically adjusted, and a mode switching suppression mechanism is introduced to enhance the stability of recognition.

[0125] Receiving the degradation feature vector data packet refers to the backend system receiving a data packet from the monitoring lens data acquisition device that encapsulates multi-dimensional spatial focusing features, multiple temperature values, temperature change rate, and the camera's cumulative operating time. This data packet contains comprehensive information about the current optical performance of the monitoring lens and the environmental condition.

[0126] Furthermore, time-series analysis is performed on multiple continuously received degradation feature vector data packets to calculate the magnitude of change between each degradation feature vector data packet and the previous time packet. The aim is to assess the stability or drastic change in optical performance by analyzing the trend of data changes over time. The magnitude of change can be understood as the distance or difference between the current data packet and the previous data packet in the feature space, and can be calculated using metrics such as Euclidean distance, cosine similarity, or Mahalanobis distance.

[0127] In practical applications, when the change amplitude is less than or equal to a preset stability threshold, it indicates that the optical performance is in a relatively stable state. At this time, the current degradation feature vector data package can be directly compared with the feature vectors of various preset and known degradation modes. For example, by calculating similarity or distance, the best-matching optical performance degradation mode can be determined. The preset degradation mode feature vectors are representative features of typical degradation modes that are pre-trained and stored based on historical data or expert experience.

[0128] Furthermore, when the magnitude of the change exceeds a preset stability threshold, it indicates that the optical performance may be undergoing significant changes or is in an unstable state. In this case, to avoid misjudgment, the system activates a feature vector accumulation window. This window collects multiple degradation feature vector data packets over a period of time and performs statistical analysis on these packets, such as calculating the mean, median, variance, or performing cluster analysis, to obtain more stable and representative degradation trend information, thereby identifying optical performance degradation patterns.

[0129] To enhance the stability of recognition, this application also proposes dynamically adjusting the confidence threshold for optical performance degradation pattern recognition and introducing a mode switching suppression mechanism. The dynamically adjusted confidence threshold aims to flexibly adjust the stringency of pattern recognition based on actual operating conditions and recognition performance. The mode switching suppression mechanism prevents frequent switching of identified degradation patterns within a short period, thereby improving the system's robustness to transient noise or short-term fluctuations and ensuring the continuity and reliability of recognition results.

[0130] This application's solution effectively distinguishes between stable and dynamically changing states of optical performance by introducing time series analysis and variation amplitude judgment. When optical performance is stable, pattern matching is performed directly, improving the real-time performance and efficiency of recognition. When optical performance changes drastically, statistical analysis is performed by activating a feature vector accumulation window, which effectively smooths out instantaneous fluctuations and obtains more reliable degradation trend information, thereby avoiding misjudgments caused by data anomalies at a single moment. In addition, dynamically adjusting the confidence threshold and introducing a mode switching suppression mechanism further improves the accuracy and stability of degradation pattern recognition, enabling the system to output recognition results more robustly when facing complex and ever-changing working environments, reducing unnecessary mode switching and false alarms.

[0131] In some preferred embodiments, it is assumed that the optical performance of a surveillance camera begins to degrade slightly during long-term operation. The backend system continuously receives degradation feature vector data packets. Initially, the changes in these data packets are small, less than a preset stability threshold. The system directly compares the current data packet with a preset "slightly blurred" degradation mode feature vector and identifies it as a "slightly blurred" mode with high confidence. At this point, the backend system may perform corresponding image processing compensation.

[0132] However, over time, if condensation or dust accumulation occurs inside the lens, causing a sharp decline in optical performance, the variation in the continuously received degradation feature vector data packets will suddenly increase, exceeding a preset stability threshold. The system will immediately activate a feature vector accumulation window, for example, set to collect the five most recent data packets. Within these five data packets, the system will perform statistical analysis, such as calculating the average feature vector of these data packets and matching it with preset degradation mode feature vectors such as "severe condensation" or "dust accumulation."

[0133] During this process, if the image processing compensation performed by the backend system results in a low improvement, the system will raise the confidence threshold for optical performance degradation pattern recognition, requiring a higher matching degree to confirm the pattern. Simultaneously, a mode switching suppression mechanism will take effect. For example, if the recognition result frequently jumps between "slight blur" and "severe condensation" within a short period (e.g., within 30 seconds), the system will suppress mode switching until more stable evidence confirms a mode shift has indeed occurred. In this way, even under conditions of rapid optical performance deterioration or complex and variable environments, the system can stably and accurately identify the true degradation pattern and promptly trigger corresponding maintenance alerts or higher-level compensation measures.

[0134] This application further proposes a step for dynamically adjusting the confidence threshold for optical performance degradation pattern recognition, including:

[0135] After the backend system performs image processing compensation, the local sharpness of the compensated image is evaluated.

[0136] The improvement effect of the compensation is calculated based on the local sharpness of the compensated image and the local sharpness of the image before compensation.

[0137] When the improvement in compensation effect is greater than or equal to the preset effective compensation threshold, the confidence threshold for optical performance degradation pattern recognition is reduced.

[0138] When the improvement in compensation effect is less than the preset effective compensation threshold, the confidence threshold for optical performance degradation pattern recognition is increased.

[0139] Specifically, after the backend system performs image processing compensation, a local sharpness evaluation is needed for the compensated image. This local sharpness can be calculated using various existing techniques, such as quantifying it by analyzing high-frequency components, edge sharpness, or contrast. The purpose of evaluating the local sharpness of the compensated image is to obtain the actual impact of the compensation measures on image quality. Subsequently, the local sharpness of the compensated image is compared with that of the image before compensation to calculate the magnitude of the improvement. The magnitude of the improvement can be defined as the difference or ratio between the two values, used to measure the actual effect of the image processing compensation.

[0140] Among these parameters, the preset effective compensation threshold is crucial, and its setting should be based on the actual application scenario and the expected compensation effect. When the calculated improvement in compensation effect is greater than or equal to the preset effective compensation threshold, it indicates that the current image processing compensation is effective and can significantly improve image quality. In this case, the confidence threshold for optical performance degradation pattern recognition can be lowered. Lowering the confidence threshold means that the system's "confidence" in the recognition results is reduced, allowing the system to more flexibly and quickly identify degradation patterns in subsequent recognitions, especially when compensation measures have already been effective, avoiding overly conservative approaches.

[0141] Conversely, when the improvement effect of the compensation is lower than the preset effective compensation threshold, it indicates that the current image processing compensation effect is poor and has failed to effectively improve image quality. This may mean that the currently identified degradation pattern is inaccurate, or that the compensation strategy is not suitable for the current degradation situation. In this case, it is necessary to increase the confidence threshold for optical performance degradation pattern recognition. Increasing the confidence threshold aims to make the system more cautious when identifying degradation patterns, requiring higher matching degrees or stronger evidence to avoid misjudgments or inaccurate pattern recognition, thereby providing a more reliable basis for subsequent more effective compensation or maintenance.

[0142] In some preferred embodiments, assuming that the image from the monitoring lens becomes slightly blurry after operating for a period of time, the system analyzes the degradation feature vector data packet to initially identify it as a "slight defocus" degradation mode, and the backend system performs corresponding image processing compensation, such as adjusting the digital focus parameters. After compensation, the system immediately acquires the compensated image and evaluates its local sharpness. For example, if the local sharpness of the image before compensation is 0.75, the local sharpness of the image after compensation is improved to 0.85. At this time, the improvement in compensation effect is 0.10. If the preset effective compensation threshold is 0.05, since 0.10 is greater than 0.05, it indicates that the compensation effect is significant. In this case, the system will lower the confidence threshold for optical performance degradation mode recognition, for example, from 0.90 to 0.80. This means that in subsequent degradation mode recognition, the system will more easily identify "slight defocus" or other related degradation modes, thereby speeding up the response.

[0143] On the other hand, if the local sharpness of the compensated image only improves to 0.78, the improvement is only 0.03. Since 0.03 is lower than the preset effective compensation threshold of 0.05, this indicates poor compensation. In this case, the system will increase the confidence threshold for optical performance degradation pattern recognition, for example, from 0.90 to 0.95. This will prompt the system to be more stringent in identifying degradation patterns, potentially requiring more evidence or a stronger match to confirm the current degradation pattern, thus avoiding erroneous judgments based on insufficient compensation and guiding the system to re-evaluate the degradation pattern or adjust the compensation strategy.

[0144] This application further proposes a step to identify optical performance degradation modes by activating a feature vector accumulation window when the change magnitude exceeds a preset stability threshold, and performing statistical analysis on the degraded feature vector data packets within the feature vector accumulation window:

[0145] Start the adaptively adjusted feature vector accumulation window;

[0146] Adjust the length of the feature vector accumulation window based on the internal changes of the degraded feature vector data packets within the feature vector accumulation window;

[0147] The weight of the degraded feature vector data packet is determined based on the similarity between the degraded feature vector data packet and the average feature vector within the feature vector accumulation window.

[0148] Based on the weights of the degraded feature vector data packets, a weighted statistical analysis is performed on the degraded feature vector data packets within the feature vector accumulation window to obtain the weighted statistics.

[0149] Matching is performed based on weighted statistics and preset optical performance degradation modes;

[0150] Calculate the confidence score of the match;

[0151] When the confidence scores of multiple matches are all higher than the preset matching threshold, analyze the evolution trend of degraded feature vector data packets within the feature vector accumulation window.

[0152] Identify optical performance degradation patterns.

[0153] Specifically, activating the adaptively adjusted feature vector accumulation window means that when the change in the degradation feature vector data packet exceeds a preset stability threshold, the system no longer simply activates a fixed-length accumulation window, but instead activates a window whose length can be dynamically adjusted according to the actual situation. This adaptively adjusted feature vector accumulation window can better adapt to the needs of different degradation rates and modes.

[0154] The feature vector accumulation window's length is adjusted based on the internal changes of the degraded feature vector data packets within it. This can be understood as the system continuously monitoring changes in the degraded feature vector data packets within the window, such as their instantaneous rate of change, volatility, or trend. When internal changes are drastic, the window length can be shortened to respond more quickly to current changes; when internal changes are gradual, the window length can be extended to accumulate more data and improve statistical stability.

[0155] In practical applications, the weight of a degraded feature vector data packet is determined based on its similarity to the average feature vector within the feature vector accumulation window. Specifically, for each degraded feature vector data packet within the window, the similarity between it and the average feature vector of all data packets within the current window is calculated (e.g., using Euclidean distance, cosine similarity, etc.). Higher similarity indicates that the data packet better represents the overall degradation state of the current window and is therefore assigned a higher weight; conversely, data packets with low similarity may be considered outliers or noise, and their weight is reduced. The aim is to highlight data packets that contribute more to the current degradation pattern recognition while reducing the impact of outliers.

[0156] Furthermore, based on the weights of the degradation feature vector data packets, a weighted statistical analysis is performed on the degradation feature vector data packets within the feature vector accumulation window to derive a weighted statistic. This means that when performing statistical analysis (such as calculating the mean, variance, trend, etc.), instead of simply applying equal weights to all data packets, a weighted calculation is performed based on their predetermined weights. Therefore, the resulting weighted statistic can more accurately reflect the true state and evolution trend of the current degradation mode.

[0157] Building upon this, matching based on weighted statistics with preset optical performance degradation patterns involves comparing the weighted statistics obtained through weighted statistical analysis with the feature vectors or models of various known optical performance degradation patterns stored in advance. This matching process can employ machine learning algorithms, pattern recognition algorithms, or rule-based matching methods.

[0158] Meanwhile, calculating the confidence level of the match refers to quantifying the degree of matching between the current weighted statistic and a preset degradation pattern. The higher the confidence level, the stronger the reliability of the match.

[0159] Furthermore, when the confidence levels of multiple matches are all higher than a preset matching threshold, the evolution trend of degraded feature vector data packets within the feature vector accumulation window is analyzed. This means that when the system finds that the current data has a high degree of matching with multiple degradation patterns, in order to avoid misjudgment, it will further analyze the trend of degradation feature vector data packets within the window over time. For example, its growth / decline rate, fluctuation cycle, inflection point, etc., can be analyzed to distinguish degradation patterns that are similar in features but have different evolution paths.

[0160] Finally, through the above comprehensive analysis, the optical performance degradation mode was identified.

[0161] In some preferred embodiments, it is assumed that the optical performance of a surveillance camera begins to degrade after long-term operation. The system first determines that the optical performance has degraded based on sharpness indicators and environmental condition parameters, and generates degradation feature vector data packets. When the backend system receives these data packets and performs time-series analysis, it finds that the change amplitude of the degradation feature vector data packets suddenly increases, exceeding a preset stability threshold, indicating that a new or accelerated degradation mode may exist.

[0162] At this point, the proposed solution initiates an adaptively adjusted feature vector accumulation window. For example, the initial window length might be set to 10 data packets. The system monitors the internal changes of these 10 data packets in real time. If the rate of change in the first 5 data packets is found to be very high, indicating that degradation is rapidly progressing, the system may shorten the window length to 5 data packets to more quickly focus on the latest degradation state. Conversely, if the rate of change within the window is low, the system may extend the window length to 15 data packets to collect more data and improve statistical stability.

[0163] Simultaneously, for each degraded feature vector data packet within the window, the system calculates its similarity to the average feature vector of all data packets within the current window. For example, if a data packet has a small Euclidean distance from the average feature vector, its similarity is high, and it will be assigned a weight of 0.8; if a data packet has a large distance, it may be assigned a weight of 0.2. Subsequently, the system uses these weights to perform weighted statistical analysis on the degraded feature vector data packets within the window, calculating weighted statistics such as the weighted average and weighted variance.

[0164] Next, this weighted statistic is used to match the feature vectors of preset optical performance degradation modes such as "lens fogging," "lens scratches," or "sensor contamination." Suppose the matching results show that the confidence level for the "lens fogging" mode is 0.9, and the confidence level for the "sensor contamination" mode is 0.85, both exceeding the preset matching threshold of 0.8. In this case, the system does not immediately make a judgment but further analyzes the evolution trend of the degradation feature vector data packets within the feature vector accumulation window. For example, if it is found that the degradation feature vectors exhibit a continuous, non-linear upward trend in a short period of time, and this trend is more consistent with the typical evolution trajectory of the "lens fogging" mode than with the slow accumulation characteristics of the "sensor contamination" mode, then the system will ultimately identify the optical performance degradation mode as "lens fogging." In this way, even when multiple degradation modes have similar characteristics or the degradation process is dynamically complex, the scheme of this application can achieve more accurate and robust degradation mode identification.

[0165] This application further proposes a step of adjusting the length of the feature vector accumulation window based on the internal changes of the degraded feature vector data packet within the feature vector accumulation window, including:

[0166] Real-time monitoring of the instantaneous rate of change of degraded feature vector data packets within the feature vector accumulation window;

[0167] The instantaneous rate of change is compared with a preset dynamic rate of change threshold.

[0168] When the instantaneous rate of change equals the preset dynamic rate of change threshold, the length of the feature vector accumulation window remains unchanged;

[0169] When the instantaneous rate of change exceeds the preset dynamic rate of change threshold, the length of the feature vector accumulation window is shortened.

[0170] When the instantaneous rate of change is less than the preset dynamic rate of change threshold, the length of the feature vector accumulation window is extended;

[0171] The window length adjustment rules are preset according to the process characteristics, and the length adjustment of the feature vector accumulation window is triggered.

[0172] After adjusting the length of the cumulative window for the eigenvectors, the weighted statistics are recalculated.

[0173] Specifically, real-time monitoring of the instantaneous change rate of degraded feature vector data packets within the feature vector accumulation window refers to the system continuously tracking and calculating the rate or trend of numerical change of degraded feature vector data packets within a short time interval within the feature vector accumulation window. This can be done by comparing the differences in degraded feature vector data packets at adjacent time points and combining the time interval for calculation, such as using methods like differencing, moving average, or Kalman filtering to estimate the instantaneous change rate. Its purpose is to capture the immediate dynamics of the degradation process. The comparison between the instantaneous change rate and a preset dynamic change rate threshold aims to determine the activity level of the degradation process. The preset dynamic change rate threshold can be determined based on historical data, expert experience, or system debugging results, and is used to distinguish between stable, accelerated, or decelerated states of the degradation process.

[0174] In practical applications, when the instantaneous rate of change equals the preset dynamic rate of change threshold, it indicates that the degradation process is in a relatively stable state. At this time, the length of the feature vector accumulation window remains unchanged to maintain a stable basis for statistical analysis. When the instantaneous rate of change is greater than the preset dynamic rate of change threshold, it means that the degradation process is accelerating or undergoing drastic changes. In this case, shortening the length of the feature vector accumulation window allows the system to respond more quickly to new degradation trends, avoids interference from old data in current judgments, and improves the sensitivity of identification. Conversely, when the instantaneous rate of change is less than the preset dynamic rate of change threshold, it indicates that the degradation process is becoming more gradual or decelerating. In this case, extending the length of the feature vector accumulation window can incorporate more historical data, smooth out short-term fluctuations, and improve the robustness of statistical analysis and the accuracy of identification. Furthermore, setting window length adjustment rules based on process characteristics and triggering the adjustment of the feature vector accumulation window length means that the evolution of degradation patterns may differ in different monitoring scenarios or lens usage processes. Therefore, one or more sets of window length adjustment rules can be preset according to specific process characteristics. For example, a more aggressive shortening strategy can be used in critical processes or high-risk stages, while a more conservative extending strategy can be used in stable operation stages. These rules are triggered when specific conditions are met to enable more intelligent window management. Recalculating the weighted statistics after adjusting the length of the feature vector accumulation window ensures that subsequent degradation pattern recognition is based on the latest dataset that best reflects the current degradation state. Changes in window length affect the data samples within the window, therefore, re-performing the weighted statistical analysis is necessary to obtain accurate weighted statistics, which in turn supports subsequent pattern matching and confidence calculation.

[0175] This application's solution effectively addresses the inadequacy of traditional fixed or simple window length adjustments in handling complex dynamic degradation processes by introducing real-time monitoring of the instantaneous change rate of degradation feature vector data packets and dynamically adjusting the length of the feature vector accumulation window based on this data. When the degradation process accelerates, shortening the window length allows the system to quickly focus on the latest degradation data, avoiding dilution by outdated historical data, thereby improving the ability to capture sudden or rapidly evolving degradation patterns. Conversely, when the degradation process stabilizes, extending the window length allows for the inclusion of data over a longer period, smoothing out short-term noise and enhancing the stability and accuracy of statistical analysis. This adaptive window length adjustment mechanism enables the system to flexibly adjust the "field of vision" of data analysis according to the actual dynamic characteristics of the degradation process, ensuring that degradation patterns can be identified with the optimal dataset at any stage. Furthermore, pre-setting adjustment rules based on process characteristics further enhances the practicality and relevance of the solution, making it better suited to the needs of different application scenarios.

[0176] In some preferred embodiments, assuming a surveillance camera operates outdoors for extended periods, its optical performance may be affected by various environmental factors such as temperature, humidity, and dust accumulation, leading to a non-linear dynamic degradation process. For example, after a sandstorm, dust may rapidly accumulate on the lens surface, causing a sharp decline in clarity, at which point the instantaneous change rate of the degradation feature vector data packets will significantly increase. According to the solution of this application, when the system detects that the instantaneous change rate exceeds a preset dynamic change rate threshold, the length of the feature vector accumulation window is automatically shortened, for example, from the original 10 data packets to 3 data packets. In this way, the system can quickly focus on the latest data after the sandstorm, rapidly identify the "dust accumulation" degradation mode, and promptly trigger image processing compensation or maintenance alarms.

[0177] On the other hand, during stable daily operation, the optical performance of a lens may deteriorate slowly and gradually, for example, due to coating aging. In this case, the instantaneous rate of change of the degradation feature vector data packets will be relatively small. When the system detects that the instantaneous rate of change is less than a preset dynamic rate of change threshold, the length of the feature vector accumulation window is extended, for example, from 10 data packets to 20 data packets. By incorporating data over a longer period, the system can smooth out short-term measurement noise and more stably identify the slowly evolving degradation pattern of "coating aging," avoiding misjudgments due to short-term fluctuations. Furthermore, depending on the process characteristics, such as the initial break-in phase after lens installation, a shorter window length adjustment rule can be preset to quickly capture potential initial defects; while during long-term operation, a longer window length adjustment rule can be preset to improve the accuracy of identifying slow degradation trends. After adjusting the length of the feature vector accumulation window, the system immediately recalculates the weighted statistics to ensure that subsequent pattern matching and confidence calculations are always based on the dataset that best reflects the current degradation state, thereby achieving accurate and real-time identification of the degradation patterns of the monitored lens's optical performance.

[0178] This application further proposes steps for introducing a mode switching suppression mechanism, including:

[0179] The difference between the degradation feature vector of the currently identified optical performance degradation mode and the degradation feature vector of the same optical performance degradation mode under historical stable conditions is monitored to obtain the deviation of the degradation feature vector.

[0180] When the difference exceeds the preset threshold for changes within the pattern, a short-term observation window is activated;

[0181] Analyze the changing trend of the degradation feature vector within the short-term observation window;

[0182] The suppression duration of the mode switching suppression mechanism is dynamically adjusted based on the changing trend and the deviation of the degradation feature vector.

[0183] Based on the changing trend and the deviation of the deteriorating feature vector, increase the confidence increment threshold for allowing mode switching;

[0184] Based on the feedback of the compensation effect after the backend system performs image processing compensation, the mode switching suppression mechanism is adjusted.

[0185] Specifically, the deviation of the degradation feature vector refers to the quantitative difference between the degradation feature vector of the currently identified optical performance degradation mode and the degradation feature vector of the same mode in its historical stable state. This difference can be obtained by calculating the Euclidean distance, cosine similarity, or Mahalanobis distance between the two vectors, and is used to measure the degree of deviation between the current degradation state and the typical state of the mode.

[0186] The preset threshold for changes within a pattern is an empirical value or a threshold determined through training. It is used to define whether the changes in the current degradation feature vector have exceeded the normal fluctuation range of the pattern. When the deviation exceeds this threshold, it indicates that the current degradation state may be evolving into other patterns or is in an unstable state. In this case, a short-term observation window needs to be activated for further analysis.

[0187] The short-term observation window is a preset time period during which the system continuously collects degradation feature vector data packets and monitors and analyzes them in real time to capture instantaneous changes in degradation trends.

[0188] Analyzing the changing trend of degradation feature vectors within a short-term observation window specifically involves comparing consecutive degradation feature vector data packets within the window, calculating their difference values, and determining the magnitude of change between degradation feature vector data packets based on these difference values, thereby identifying whether the degradation is continuously aggravating, tending to stabilize, or exhibiting abnormal fluctuations.

[0189] The suppression duration of the dynamic mode switching suppression mechanism refers to flexibly setting the length of time during which mode switching is prevented or delayed based on the changing trend and deviation of the degradation feature vector. For example, when the degradation trend is not obvious or the deviation is small, the suppression duration can be shorter; when the degradation trend is severe or the deviation is large, the suppression duration can be extended to avoid the system frequently switching modes when uncertainty is high.

[0190] Increasing the confidence increment threshold for allowing mode switching means that during the activation of the mode switching suppression mechanism, the new mode recognition result must reach a higher confidence increment to be confirmed as a valid mode switch. This helps filter out false positives caused by short-term fluctuations or noise, ensuring that switching only occurs when the new deterioration mode has a high degree of certainty.

[0191] The adjustment of the mode switching suppression mechanism is triggered by combining the feedback of the compensation effect after the backend system performs image processing compensation. This means using the actual effect of the image compensation by the backend system as feedback information to optimize the parameters of the mode switching suppression mechanism. For example, if the compensation effect is poor, it may mean that the currently identified degradation mode is inaccurate or the degree of degradation exceeds expectations. In this case, the suppression duration or confidence increment threshold can be adjusted to enable the system to more accurately identify the true degradation mode.

[0192] This application's solution effectively addresses the stability issues that may exist in pattern recognition by introducing a mode-switching suppression mechanism. Specifically, by continuously monitoring the difference between the degradation feature vector of the current deteriorated mode and the feature vector of the same mode in historical stable states, the system can quantify the deviation of the current degradation state. When the deviation exceeds a preset threshold, the system does not immediately switch modes but instead initiates a short-term observation window to meticulously analyze the changing trend of the degradation feature vector. This mechanism avoids misjudgments caused by instantaneous fluctuations or noise. Furthermore, based on the analyzed changing trend and deviation, the system can dynamically adjust the suppression duration of mode switching and the confidence increment threshold required to allow mode switching. For example, when the degradation trend is unclear or fluctuates significantly, the suppression duration is extended, and the confidence requirement for switching to the new mode is higher, thus providing the system with a longer observation period and a higher confirmation threshold. In addition, combined with the actual effect feedback after the backend system performs image processing compensation, this mechanism can self-optimize, ensuring the effectiveness and adaptability of the mode-switching suppression strategy. Therefore, the solution proposed in this application can ensure that the identification of optical performance degradation modes is more stable and accurate in complex and ever-changing real-world application scenarios, avoiding unnecessary frequent mode switching.

[0193] In some preferred embodiments, it is assumed that after long-term operation, the optical performance degradation mode of a surveillance lens is identified as "slight fogging". The system continuously monitors the degradation feature vectors associated with the "slight fogging" mode, such as local sharpness and high-frequency edge features. If, at some moment, due to minor fluctuations in ambient temperature or changes in lighting conditions, a slight instantaneous change occurs in the degradation feature vector, it causes a certain deviation from the historical stable feature vector of the "slight fogging" mode.

[0194] If the deviation exceeds a preset threshold for changes within the mode, the system will not immediately switch to "moderate fogging" or another mode. Instead, it will initiate a short observation window. During this window, the system will intensively collect degradation feature vector data packets and analyze their changing trends. If the analysis shows that the change is a transient, non-persistent fluctuation, and the deviation does not continue to increase, the mode switching suppression mechanism will extend the suppression duration and increase the confidence increment threshold for allowing mode switching. This means that unless subsequent degradation feature vectors consistently and significantly point to the "moderate fogging" mode and reach a higher confidence level, the system will maintain the current "slight fogging" mode recognition.

[0195] Furthermore, if the backend system performs image processing compensation on the "slightly fogged" mode, and the compensation feedback shows that image clarity has been effectively improved, this further verifies the accuracy of the current pattern recognition. In this case, the mode switching suppression mechanism may be fine-tuned based on the compensation feedback; for example, if the accuracy of the current pattern recognition is confirmed, the sensitivity to short-term fluctuations may be appropriately relaxed.

[0196] Conversely, if the degraded feature vector within the short-term observation window continuously and significantly evolves towards the "moderate fogging" mode, and the deviation continues to increase, while the compensation effect feedback shows poor compensation for "slight fogging," then the mode switching suppression mechanism, based on this information, will allow the system to switch to the "moderate fogging" mode after meeting a higher confidence increment threshold, and trigger corresponding image processing compensation or maintenance alarms. In this way, the proposed solution effectively avoids mode misjudgment and frequent switching caused by transient noise or uncertainty, ensuring the stability and accuracy of pattern recognition.

[0197] This application further proposes a method for acquiring data from a surveillance camera, which also includes an early warning mechanism step, as detailed below:

[0198] Real-time monitoring of the instantaneous rate of change of degraded feature vector data packets within a short observation window;

[0199] When the instantaneous rate of change of the degradation feature vector data packet within the short-term observation window shows a continuous acceleration or a jump near the set process node, and the trajectory of change is similar to the early evolution trajectory of the preset severe degradation mode, the early warning mechanism is triggered.

[0200] Specifically, the early warning mechanism aims to identify and alert to potential severe degradation of optical performance at an early stage. "Real-time monitoring of the instantaneous change rate of degradation feature vector data packets within a short observation window" refers to the system continuously calculating the rate at which the degradation feature vector data packets change over time within a short observation window. This instantaneous change rate directly reflects the speed of optical performance degradation, such as a rapid decline in indicators like sub-region focus scores or local sharpness of high-frequency edge features within the degradation feature vector data packets, or an abnormal increase in temperature values ​​or the rate of temperature change. "Set process nodes" can be understood as specific time points, operational stages, or environmental conditions, determined based on experience or historical data analysis, during the operation of the monitored lens, that are prone to drastic changes in optical performance, such as equipment startup, sudden load changes, or abrupt changes in ambient temperature. "Continuous acceleration or abrupt increase" describes the non-linear growth trend of the instantaneous change rate of the degradation feature vector data packets, where "continuous acceleration" indicates a continuously accelerating degradation rate, and "abrupt increase" refers to a large and sudden increase in the degradation rate within a short period. "The early evolution trajectory of the preset severe degradation mode" refers to the specific change pattern or trend of the degradation feature vector data packet before severe optical performance degradation (such as lens breakage, severe contamination, internal optical component displacement, etc.) occurs, summarized by analyzing historical failure data. When the above conditions are met simultaneously, that is, when the instantaneous rate of change shows an abnormal acceleration or jump near the critical process node, and its change trajectory is highly consistent with the known early signs of severe degradation, the system will "trigger the early warning mechanism" and issue a high-level alarm message.

[0201] Specifically, by monitoring the instantaneous rate of change of degradation feature vector data packets within a short observation window in real time, the system can capture dynamic changes during the optical performance degradation process, especially those nonlinear accelerations or abrupt upward trends that foreshadow serious problems. When such abnormal changes occur near a set process node, and their trajectory is similar to the early evolution trajectory of a preset severe degradation mode, the system can quickly identify potential serious risks. Thus, the early warning mechanism can issue an alert before the optical performance reaches a critical degradation level, or even before the mode switching suppression mechanism has fully identified a new severe degradation mode, buying valuable time for the backend system to take emergency intervention measures (such as immediate shutdown for inspection, component replacement, or adjustment of operating parameters).

[0202] The steps for analyzing the changing trend of deterioration feature vectors within a short observation window include:

[0203] Real-time collection of multiple degraded feature vector data packets within a short observation window;

[0204] Compare the differences between each pair of adjacent degraded feature vector data packets and calculate the difference value of the degraded feature vectors;

[0205] Based on the difference value, determine the magnitude of change between degraded feature vector data packets and identify the trend of degraded feature vector changes within a short observation window.

[0206] Specifically, once a short-term observation window is activated, the system continuously and in real-time collects multiple degradation feature vector data packets generated within that window. These data packets contain detailed information about the current optical performance of the monitoring lens, such as multi-dimensional spatial focusing characteristics, multiple temperature values, temperature change rate, and cumulative camera operating time.

[0207] Subsequently, to understand the dynamics of the degradation process, these collected degradation feature vector data packets need to be compared and analyzed. Specifically, any pair of adjacent degradation feature vector data packets within a short observation window can be selected, and the difference between them can be calculated. This difference value can reflect the degree of change in optical performance degradation characteristics between two consecutive time points. For example, this difference can be quantified by calculating the Euclidean distance, cosine similarity, or Manhattan distance between the vectors.

[0208] Furthermore, based on the calculated differences in the degradation feature vectors, the magnitude of change between the degradation feature vector data packets can be determined. For example, a large difference indicates that the degradation features have changed significantly in a short period of time; a small difference indicates that the degradation features are relatively stable. Through continuous observation and analysis of these magnitudes of change, the overall trend of degradation feature vector changes within a short observation window can be identified, such as accelerated degradation, slowed degradation, fluctuating degradation, or a tendency to stabilize.

[0209] refer to Figure 2 This application proposes a surveillance camera data acquisition system, applied to a surveillance camera data acquisition method, the system comprising:

[0210] The setup module allows you to set up a fixed reference object for optical performance monitoring.

[0211] The acquisition module acquires image information of a fixed reference object;

[0212] The evaluation module assesses the sharpness metrics of image information.

[0213] The acquisition module collects environmental condition parameters related to the degradation of optical performance;

[0214] The judgment module determines whether the optical performance has deteriorated based on the sharpness index and environmental condition parameters.

[0215] The identification module identifies optical performance degradation modes based on the degradation feature vector data packet.

[0216] The transmission module transmits information about the optical performance degradation mode to the backend system.

[0217] The setup module is configured to set a fixed reference object for optical performance monitoring. This fixed reference object can be a physical object with a known geometry and texture, placed within the field of view of the monitoring lens to facilitate periodic or real-time image acquisition by the system.

[0218] The acquisition module is configured to acquire image information of a fixed reference object. This module typically captures visual data of the fixed reference object using a camera or other image sensor and converts it into a digital image format suitable for subsequent processing.

[0219] The evaluation module is configured to evaluate the sharpness index of image information. This module receives image information provided by the acquisition module and uses image processing algorithms (e.g., methods based on edge gradient, contrast, or frequency domain analysis) to calculate the image sharpness index to quantify the optical performance of the surveillance lens.

[0220] The acquisition module is configured to collect environmental condition parameters related to optical performance degradation. These parameters may include, but are not limited to, the camera's internal temperature, humidity, and operating time, which have a potential impact on the optical performance of the surveillance lens.

[0221] The judgment module is configured to determine whether optical performance has deteriorated based on sharpness indicators and environmental condition parameters. This module comprehensively analyzes and evaluates the sharpness indicators provided by the assessment module and the environmental condition parameters provided by the acquisition module, and uses preset rules or models to determine whether the optical performance of the monitoring lens has deviated from its normal operating condition.

[0222] The identification module is configured to identify optical performance degradation patterns based on degradation feature vector data packets. Once the judgment module confirms optical performance degradation, the identification module receives a degradation feature vector data packet containing sharpness indicators and environmental state parameters, and compares it with known degradation pattern feature vectors to identify the specific degradation type (e.g., blur, out of focus, chromatic aberration, etc.).

[0223] The transmission module is configured to transmit information about optical performance degradation patterns to the backend system. This module is responsible for sending the identified optical performance degradation patterns and their related data to the backend processing system via network or other communication methods for further analysis, alerting, or compensation processing.

[0224] The content disclosed above is only a preferred and feasible embodiment of the present invention, and is not intended to limit the scope of protection of the present invention. Therefore, all equivalent technical changes made based on the content of the present invention specification and drawings are included within the scope of protection of the present invention. Furthermore, the elements therein can be updated as technology develops.

Claims

1. A surveillance lens data acquisition method, characterized by, The method includes the following steps: Set up a fixed reference object for monitoring optical performance; Acquire image information of a fixed reference object; Evaluate the sharpness metrics of image information; Collect environmental condition parameters related to optical performance degradation; Determine whether the optical performance has deteriorated based on the sharpness index and environmental condition parameters; Identify optical performance degradation patterns based on degradation feature vector data packets; Information on optical performance degradation patterns is transmitted to the backend system.

2. The method of claim 1, wherein the monitoring the lens data collection comprises: The steps to determine whether optical performance has deteriorated and to identify the mode of optical performance degradation include: Divide the fixed reference object into multiple sub-regions; Acquire image data from multiple sub-regions; Calculate the multi-dimensional spatial focus features of the image data for each sub-region. The sharpness index is a multi-dimensional spatial focus feature, which includes the sub-region focus score, focus score gradient, focus score asymmetry index, and local sharpness of high-frequency edge features. Collect multiple temperature values ​​and temperature change rates inside the camera, and calculate the cumulative operating time of the camera. The environmental state parameters include multiple temperature values, temperature change rates, and the cumulative operating time of the camera. Based on multi-dimensional spatial focusing characteristics, multiple temperature values, temperature change rate, and camera cumulative operating time, determine whether the optical performance has deteriorated; Encapsulate multi-dimensional spatial focusing features, multiple temperature values, temperature change rate, and camera cumulative running time to form a degraded feature vector data package; Transmit the degraded feature vector data packet to the backend system; The backend system identifies optical performance degradation patterns based on the degradation feature vector data packets; The backend system performs image processing compensation or issues maintenance alerts based on the optical performance degradation mode.

3. The method for acquiring surveillance camera data as described in claim 2, characterized in that, The backend system identifies optical performance degradation modes based on the degradation feature vector data packet, including the following steps: Receive degraded feature vector data packets; Time series analysis is performed on multiple continuously received degraded feature vector data packets to calculate the change magnitude between each degraded feature vector data packet and the degraded feature vector data packet at the previous time step. When the change amplitude is less than or equal to the preset stability threshold, the current degradation feature vector data packet is compared with the preset degradation mode feature vector to determine the matching optical performance degradation mode. When the change exceeds the preset stability threshold, the feature vector accumulation window is activated to perform statistical analysis on the degraded feature vector data packets within the feature vector accumulation window and identify the optical performance degradation mode. The confidence threshold for optical performance degradation pattern recognition is dynamically adjusted, and a mode switching suppression mechanism is introduced to enhance the stability of recognition.

4. The method for acquiring surveillance camera data as described in claim 3, characterized in that, The steps for dynamically adjusting the confidence threshold for optical performance degradation pattern recognition include: After the backend system performs image processing compensation, the local sharpness of the compensated image is evaluated. The improvement effect of the compensation is calculated based on the local sharpness of the compensated image and the local sharpness of the image before compensation. When the improvement in compensation effect is greater than or equal to the preset effective compensation threshold, the confidence threshold for optical performance degradation pattern recognition is reduced. When the improvement in compensation effect is less than the preset effective compensation threshold, the confidence threshold for optical performance degradation pattern recognition is increased.

5. The method for acquiring surveillance camera data as described in claim 3, characterized in that, When the change exceeds a preset stability threshold, the feature vector accumulation window is activated. Statistical analysis is performed on the degraded feature vector data packets within the feature vector accumulation window to identify the optical performance degradation mode. The steps include: Start the adaptively adjusted feature vector accumulation window; Adjust the length of the feature vector accumulation window based on the internal changes of the degraded feature vector data packets within the feature vector accumulation window; The weight of the degraded feature vector data packet is determined based on the similarity between the degraded feature vector data packet and the average feature vector within the feature vector accumulation window. Based on the weights of the degraded feature vector data packets, a weighted statistical analysis is performed on the degraded feature vector data packets within the feature vector accumulation window to obtain the weighted statistics. Matching is performed based on weighted statistics and preset optical performance degradation modes; Calculate the confidence score of the match; When the confidence scores of multiple matches are all higher than the preset matching threshold, analyze the evolution trend of degraded feature vector data packets within the feature vector accumulation window. Identify optical performance degradation patterns.

6. The method for acquiring data from a surveillance camera as described in claim 5, characterized in that, The steps for adjusting the length of the feature vector accumulation window based on the internal changes of the degraded feature vector data packets within the feature vector accumulation window include: Real-time monitoring of the instantaneous rate of change of degraded feature vector data packets within the feature vector accumulation window; The instantaneous rate of change is compared with a preset dynamic rate of change threshold. When the instantaneous rate of change equals the preset dynamic rate of change threshold, the length of the feature vector accumulation window remains unchanged; When the instantaneous rate of change exceeds the preset dynamic rate of change threshold, the length of the feature vector accumulation window is shortened. When the instantaneous rate of change is less than the preset dynamic rate of change threshold, the length of the feature vector accumulation window is extended; The window length adjustment rules are preset according to the process characteristics, and the length adjustment of the feature vector accumulation window is triggered. After adjusting the length of the cumulative window for the eigenvectors, the weighted statistics are recalculated.

7. The method for acquiring surveillance camera data as described in claim 3, characterized in that, The steps for introducing a mode switching suppression mechanism include: The difference between the degradation feature vector of the currently identified optical performance degradation mode and the degradation feature vector of the same optical performance degradation mode under historical stable conditions is monitored to obtain the deviation of the degradation feature vector. When the difference exceeds the preset threshold for changes within the pattern, a short-term observation window is activated; Analyze the changing trend of the degradation feature vector within the short-term observation window; The suppression duration of the mode switching suppression mechanism is dynamically adjusted based on the changing trend and the deviation of the degradation feature vector. Based on the changing trend and the deviation of the deteriorating feature vector, increase the confidence increment threshold for allowing mode switching; Based on the feedback of the compensation effect after the backend system performs image processing compensation, the mode switching suppression mechanism is adjusted.

8. The method for acquiring data from a surveillance camera as described in claim 7, characterized in that, The method also includes an early warning mechanism step, the specific operation of which is as follows: Real-time monitoring of the instantaneous rate of change of degraded feature vector data packets within a short observation window; When the instantaneous rate of change of the degradation feature vector data packet within the short-term observation window shows a continuous acceleration or a jump near the set process node, and the trajectory of change is similar to the early evolution trajectory of the preset severe degradation mode, the early warning mechanism is triggered.

9. A method for acquiring data from a surveillance camera as described in claim 7, characterized in that, The steps for analyzing the changing trend of deterioration feature vectors within a short observation window include: Real-time collection of multiple degraded feature vector data packets within a short observation window; Compare the differences between each pair of adjacent degraded feature vector data packets and calculate the difference value of the degraded feature vectors; Based on the difference value, determine the magnitude of change between degraded feature vector data packets and identify the trend of degraded feature vector changes within a short observation window.

10. A surveillance camera data acquisition system, applied to the surveillance camera data acquisition method as described in claim 1, characterized in that, The system includes: The setup module allows you to set up a fixed reference object for optical performance monitoring. The acquisition module acquires image information of a fixed reference object; The evaluation module assesses the sharpness metrics of image information. The acquisition module collects environmental condition parameters related to the degradation of optical performance; The judgment module determines whether the optical performance has deteriorated based on the sharpness index and environmental condition parameters. The identification module identifies optical performance degradation modes based on the degradation feature vector data packet. The transmission module transmits information about the optical performance degradation mode to the backend system.

Images