Image resolution adaptive adjustment method and system for dual field thermal imaging scope

Abstract

The application discloses a dual-view field thermal imaging scope image resolution adaptive adjustment method and system, relates to the technical field of thermal imaging image processing, and comprises the following steps: acquiring a plurality of thermal imaging frames corresponding to a plurality of field state labels through a dual-view field thermal imaging scope, obtaining a first thermal imaging sequence through target correlation adaptive resolution adjustment, combing the field state labels to construct a field state change chain, performing resolution optimization on the first thermal imaging sequence to obtain a second thermal imaging sequence, performing frame rate constraint reconstruction on the second thermal imaging sequence according to a predetermined output frame rate to obtain a third thermal imaging sequence, and finally generating a fourth thermal imaging sequence through a dual-view field imaging loss identification model analysis and feedback adjustment, so that the resolution adaptive adjustment is realized. The application solves the technical problems that image processing in a dual-view field scene is not accurate in adaptation, transition is not coherent, and imaging quality is affected, achieves adaptive adaptation of image resolution in a dual-view field, and improves the technical effects of image coherence, definition and stability.

Description

Technical Field

[0001] This invention relates to the field of thermal imaging image processing technology, and in particular to a method and system for adaptive adjustment of image resolution of a dual-field-of-view thermal imaging sight. Background Technology

[0002] Dual-field-of-view thermal imaging sights are widely used in security, hunting, and other scenarios. Their image quality directly impacts target recognition and aiming accuracy, and adaptive image resolution is crucial for ensuring this. Therefore, the rationality of image data processing is paramount. Existing technologies often employ fixed resolution or simple switching image data processing methods, which are effective in single-field-of-view environments. However, when switching between dual fields of view, differences in field of view and changes in target scale exist. Traditional image data processing methods cannot achieve cross-field-of-view target association and progressive resolution adaptation, easily leading to problems such as image jitter, detail drift, and cumulative imaging loss. This results in inaccurate data processing and poor image coherence, making it difficult to meet the needs of high-precision aiming scenarios. Summary of the Invention

[0003] This application provides an image resolution adaptive adjustment method and system for a dual-field-of-view thermal imaging sight, which solves the technical problems of inaccurate adaptation, inconsistent transition, and affected imaging quality in image processing under dual-field-of-view scenarios.

[0004] The first aspect of this application provides an image resolution adaptive adjustment method for a dual-field-of-view thermal imaging sight. The method includes: monitoring a target area through the dual-field-of-view thermal imaging sight and acquiring multiple thermal imaging frames corresponding to multiple field-of-view status labels; performing target-associated adaptive resolution adjustment based on the multiple thermal imaging frames to acquire a first thermal imaging sequence; performing temporal analysis on the multiple field-of-view status labels to construct a field-of-view status change chain, and performing progressive resolution transition optimization on the first thermal imaging sequence based on the field-of-view status change chain to acquire a second thermal imaging sequence; performing frame rate constraint reconstruction on the second thermal imaging sequence based on a predetermined output frame rate to acquire a third thermal imaging sequence; and introducing a dual-field-of-view imaging loss recognition model to perform resolution feedback adjustment on the third thermal imaging sequence under different field-of-view imaging loss resolutions to acquire a fourth thermal imaging sequence.

[0005] A second aspect of this application provides an image resolution adaptive adjustment system for a dual-field-of-view thermal imaging sight. The system includes: a thermal imaging frame acquisition module, used to monitor a target area through the dual-field-of-view thermal imaging sight and acquire multiple thermal imaging frames corresponding to multiple field-of-view state tags; a first thermal imaging sequence acquisition module, used to perform target-related adaptive resolution adjustment based on the multiple thermal imaging frames to acquire a first thermal imaging sequence; a second thermal imaging sequence acquisition module, used to perform temporal analysis on the multiple field-of-view state tags, construct a field-of-view state change chain, and perform progressive resolution transition optimization on the first thermal imaging sequence based on the field-of-view state change chain to acquire a second thermal imaging sequence; a third thermal imaging sequence acquisition module, used to perform frame rate constraint reconstruction on the second thermal imaging sequence based on a predetermined output frame rate to acquire a third thermal imaging sequence; and a fourth thermal imaging sequence acquisition module, used to introduce a dual-field-of-view imaging loss recognition model to perform resolution feedback adjustment on the third thermal imaging sequence under different field-of-view imaging loss resolutions to acquire a fourth thermal imaging sequence.

[0006] One or more technical solutions provided in this application have at least the following technical effects or advantages: This application acquires thermal imaging frames corresponding to different field-of-view states through a dual-field-of-view thermal imaging sight. After processing such as target association analysis, field-of-view state change chain optimization, frame rate constraint reconstruction, dual-field-of-view imaging loss analysis and multi-level adjustment, an adapted thermal imaging sequence is obtained. The resolution is dynamically adjusted by combining field-of-view switching characteristics, frame rate parameters and imaging loss accumulation to meet the usage requirements of high-precision aiming scenarios. This achieves adaptive adaptation of image resolution under dual fields of view, improving the technical effect of image coherence, clarity and stability. Attached Figure Description

[0007] To more clearly illustrate the technical solutions in the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0008] Figure 1 This is a flowchart illustrating the image resolution adaptive adjustment method of the dual-field thermal imaging sight provided in this application embodiment.

[0009] Figure 2 This is a schematic diagram of the image resolution adaptive adjustment system of the dual-field thermal imaging sight provided in this application embodiment.

[0010] Explanation of reference numerals in the attached figures: Thermal imaging frame acquisition module 1, first thermal imaging sequence acquisition module 2, second thermal imaging sequence acquisition module 3, third thermal imaging sequence acquisition module 4, and fourth thermal imaging sequence acquisition module 5. Detailed Implementation

[0011] This application provides an image resolution adaptive adjustment method and system for a dual-field-of-view thermal imaging sight, which solves the technical problems of inaccurate adaptation, inconsistent transition, and affected imaging quality in image processing under dual-field-of-view scenarios.

[0012] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of this application, and not all of them. All other embodiments obtained by those skilled in the art based on the embodiments of this application without creative effort are within the scope of protection of this application.

[0013] It should be noted that the terms "first," "second," etc., in the specification and accompanying drawings of this application are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate so that the embodiments of this application described herein can be implemented in orders other than those illustrated or described herein. Furthermore, the terms "comprising" and "having," and any variations thereof, are intended to cover non-exclusive inclusion; for example, a process, method, system, product, or server that includes a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or modules not explicitly listed or inherent to such processes, methods, products, or devices.

[0014] Example 1, as Figure 1 As shown, an image resolution adaptive adjustment method for a dual-field-of-view thermal imaging sight includes: The target area is monitored by a dual-field-of-view thermal imaging sight, and multiple thermal imaging frames corresponding to multiple field-of-view status labels are acquired.

[0015] In this embodiment, the dual-field-of-view thermal imaging sight is an integrated thermal imaging device that combines wide-angle and telephoto lenses, can capture thermal radiation from the target area and generate thermal images, supports field-of-view switching to accommodate both wide-range search and long-distance precise observation, and is suitable for aiming at relevant scenarios.

[0016] Specifically, the dual-field-of-view thermal imaging sight is first activated, initializing the built-in wide-angle thermal imaging lens, telephoto thermal imaging lens, thermal imaging sensor, image processor, and field-of-view switching control module to ensure that each functional module meets the operating parameter requirements. The thermal imaging sensor continuously captures the thermal radiation signal of the target area, converting the captured thermal radiation signal into a continuous analog electrical signal. After receiving the analog electrical signal, the image processor converts it into a digital thermal imaging frame conforming to a preset format through analog-to-digital conversion technology, realizing the real-time acquisition of thermal imaging frames.

[0017] The field-of-view switching control module monitors the type of thermal imaging lens currently in operation in real time. When it detects that a wide-angle thermal imaging lens has started working, it generates a corresponding wide-angle thermal imaging field-of-view status label; when it detects that a telephoto thermal imaging lens has started working, it generates a corresponding telephoto thermal imaging field-of-view status label.

[0018] The image processor is equipped with a timing synchronization unit, which records the acquisition timestamp of each thermal imaging frame and the generation timestamp of each field-of-view status label. By comparing the timestamps, thermal imaging frames whose acquisition timestamps and generation timestamps are in the same time interval are associated and bound with the field-of-view status labels, so that each acquired thermal imaging frame clearly carries the corresponding "wide-angle thermal imaging" or "telephoto thermal imaging" field-of-view status label, ultimately completing the acquisition of multiple field-of-view status labels and corresponding multiple thermal imaging frames.

[0019] Based on the multiple thermal imaging frames, target association adaptive resolution adjustment is performed to obtain the first thermal imaging sequence.

[0020] Optionally, target detection is first performed on multiple acquired thermal imaging frames to extract the positional and pixel scale features of targets in each frame. Then, cross-field-of-view target association analysis is performed on these features under different field-of-view state labels to establish cross-field-of-view target association relationships. Subsequently, based on these association relationships, the changing trends of target pixel scales in multiple thermal imaging frames are identified to obtain corresponding target resolution mismatch information. Finally, based on this mismatch information, adaptive resolution adjustment is performed on multiple thermal imaging frames to generate the first thermal imaging sequence. This step will be explained in detail later.

[0021] The multiple field-of-view state labels are time-series analyzed to construct a field-of-view state change chain. Based on the field-of-view state change chain, the first thermal imaging sequence is optimized for progressive resolution transition to obtain the second thermal imaging sequence.

[0022] In one embodiment of this application, firstly, each field of view status tag is bound to the acquisition timestamp of the corresponding thermal imaging frame when it is generated. The timestamp information of all field of view status tags is extracted, and the tags are sorted in chronological order. The type changes of adjacent tags in the tag sequence are recorded in turn, such as switching from a wide-angle thermal imaging tag to a telephoto thermal imaging tag, switching from a telephoto thermal imaging tag to a wide-angle thermal imaging tag, or keeping the same tag. The sorted tag sequence and the switching types of adjacent tags are integrated to form a complete field of view status change chain, clearly presenting the temporal evolution trajectory of the field of view status.

[0023] Next, based on the field of view state change chain, the distribution of field of view switching features is identified, and then the trends of target detail drift, contour drift and display jitter are identified. Based on this, a resolution progressive transition compensation strategy is formulated and the first thermal imaging sequence is adjusted to generate the second thermal imaging sequence. This step will be explained in detail in the following content.

[0024] The second thermal imaging sequence is reconstructed by frame rate constraint according to the predetermined output frame rate to obtain the third thermal imaging sequence.

[0025] Specifically, the output frame rate parameter of the second thermal imaging sequence is first obtained. Based on the relationship between this parameter and the predetermined output frame rate, the third thermal imaging sequence is obtained through inter-frame compensation reconstruction, inter-frame fusion reconstruction, or direct output. This step will be explained in detail later.

[0026] A dual-field-of-view imaging loss identification model is introduced to adjust the resolution of the third thermal imaging sequence under different field-of-view imaging loss analysis, thereby obtaining the fourth thermal imaging sequence.

[0027] Specifically, a dual-field-of-view imaging loss identification model is first introduced to analyze the imaging loss of the third thermal imaging sequence in different fields of view in order to construct an imaging loss cumulative effect map. Based on this, three resolution adjustment schemes are formulated for imaging, image processing, and display. After multi-resolution hierarchical collaborative adjustment, a fourth thermal imaging sequence is generated. This step will be explained in detail in the following content.

[0028] Furthermore, the method provided in this application embodiment includes: The dual-field-of-view thermal imaging sight includes a wide-angle thermal imaging lens and a telephoto thermal imaging lens.

[0029] Optionally, the wide-angle thermal imaging lens consists of an optical lens group adapted to the thermal radiation detection band. Its optical structure is specially designed to maintain the field of view within a preset wide range, matching the optical path characteristics of thermal imaging signal transmission. This lens collects thermal radiation energy within the target area through the optical lens group, converging thermal radiation signals emitted by various objects over a large area to the thermal imaging sensor. The sensor converts the received thermal radiation signals into electrical signals, which, after preliminary processing such as analog-to-digital conversion, form a thermal image with wide coverage. This enables rapid scanning of large target areas, efficiently capturing the thermal distribution of all objects within the area, ensuring no potential targets are missed, and providing extensive and comprehensive raw thermal imaging data for subsequent cross-field target correlation analysis.

[0030] The telephoto thermal imaging lens consists of a long-focal-length optical lens group adapted to the thermal radiation detection band. Its optical design focuses on the convergence and amplification of long-distance thermal radiation signals, meeting the optical path requirements for observing distant targets. This lens uses its long-focal-length optical structure to focus the thermal radiation signal emitted by distant targets, concentrating and amplifying the target's thermal radiation energy. The thermal imaging sensor receives the amplified thermal radiation signal and converts it into an electrical signal, which is then processed to form a thermal image with prominent target details. Its core function is to achieve accurate observation of distant targets, clearly presenting key features such as the target's outline and attitude, providing accurate thermal imaging data support for target pixel-scale feature extraction and resolution mismatch identification.

[0031] The wide-angle and telephoto thermal imaging lenses are linked through a field-of-view switching control module built into the dual-field-of-view thermal imaging sight. Based on the observation needs of the actual usage scenario, the control module switches the working states of both lenses. Either lens can be activated independently, or they can be activated alternately according to preset logic. During operation, each lens generates corresponding thermal images. Working in conjunction with the thermal imaging sensor and image processor, these images acquire thermal frames carrying "wide-angle thermal imaging" or "telephoto thermal imaging" field-of-view status tags. This provides fundamental data from different field-of-view dimensions for subsequent steps such as target association adaptive resolution adjustment, field-of-view status change chain construction, and imaging loss analysis. This ensures that the dual-field-of-view thermal imaging sight can achieve both wide-range target searching and precise observation of distant targets.

[0032] Furthermore, the method provided in this application embodiment includes: Target detection is performed on the multiple thermal imaging frames to obtain target position features and target pixel scale features for each frame; cross-field-of-view target association analysis is performed on the target position features and target pixel scale features for each frame under different field-of-view state labels to establish cross-field-of-view target association relationships; target pixel scale change trend identification is performed on the multiple thermal imaging frames based on the cross-field-of-view target association relationships to obtain multiple target resolution mismatch information; adaptive resolution adjustment is performed on the multiple thermal imaging frames based on the multiple target resolution mismatch information to generate the first thermal imaging sequence.

[0033] Specifically, a threshold segmentation method is used to perform target detection on multiple thermal imaging frames. The specific process is as follows: First, the grayscale histogram of each thermal imaging frame is calculated. The horizontal axis of the grayscale histogram represents the grayscale level of the pixels in the thermal imaging frame, which is usually in the range of 0-255, corresponding to different thermal radiation intensities carried by the pixels. The vertical axis represents the number of pixels corresponding to each grayscale level. The process of obtaining this histogram involves counting the distribution frequency of all pixels in the thermal imaging frame at each grayscale level to form a complete grayscale distribution statistical chart.

[0034] Next, the optimal segmentation threshold is determined using the maximum inter-class variance method. Specifically, the grayscale value range of the thermal imaging frame is first defined as 0-255, and each grayscale value in this range is used as a candidate threshold k. For each candidate threshold k, all pixels in the image are divided into two classes: the background class with grayscale values ​​less than or equal to k and the foreground class with grayscale values ​​greater than k, i.e., the target candidate class. The number of pixels and the mean grayscale value of each class are calculated, and then the variance between the two classes is solved. The above steps are repeated to traverse all candidate thresholds. Finally, the candidate grayscale value that maximizes the inter-class variance is selected as the optimal segmentation threshold. At this point, the grayscale difference between the background class and the foreground class is the most significant, which can separate the target and background regions to the greatest extent.

[0035] Then, regions in the thermal imaging frame with grayscale values ​​higher than the optimal segmentation threshold are identified as candidate target regions. After removing noise regions with areas smaller than a preset minimum pixel threshold, the effective target regions are obtained. For the effective target regions, rectangular bounding boxes are used to define their positional features. The pixel coordinates of the upper left and lower right corners of the bounding boxes are recorded, and the number of pixels in the horizontal and vertical directions of the rectangular bounding boxes are counted. These are used as the pixel scale features of the target in each frame.

[0036] The preset minimum pixel threshold is determined based on the application scenario of the dual-field-of-view thermal imaging sight, the pixel resolution of the thermal imaging sensor, and the lens focal length parameters. It is combined with the minimum pixel occupancy of typical minimum targets in the aiming scenario, such as small objects at a distance or detailed targets that need to be accurately identified, in the thermal imaging frame. Specifically, it is set by those skilled in the art through statistical analysis of the thermal imaging pixel distribution data of such typical minimum targets at different fields of view and distances, and after multiple experimental calibrations, a fixed pixel number threshold is set, or it is dynamically adjusted according to the target recognition requirements of the actual use scenario. The value must ensure that it can effectively eliminate isolated pixels and small invalid interference areas caused by background thermal noise, sensor interference, etc. in the thermal imaging frame, while avoiding the accidental elimination of real small target areas, thus ensuring the accuracy of target detection.

[0037] Next, the target location features (coordinate values) and pixel scale features (width and height in pixels) of each thermal imaging frame are combined into a feature vector. Target feature vectors are then extracted from all thermal imaging frames under both the wide-angle and telephoto field-of-view status labels. The Euclidean distance between each target feature vector in the wide-angle field of view and each target feature vector in the telephoto field of view is calculated. A distance threshold is set, determined statistically from a large number of dual-field-of-view target matching samples, typically 1.5 times the standard deviation of the feature vector dimension. When the Euclidean distance between target feature vectors in the two fields of view is less than this threshold, they are considered the same target, thus establishing a cross-field-of-view target association.

[0038] Then, the pixel scale values ​​of the same target in different thermal imaging frames are sorted according to the frame acquisition time. A data point set is constructed with acquisition time as the x-axis and pixel scale value as the y-axis. Linear regression fitting is performed on this data point set to obtain the target pixel scale change trend line. The deviation value of each data point from the trend line is calculated, and a deviation threshold of 10% of the trend line slope is set. When the deviation value of a frame exceeds the threshold, it is determined that there is a resolution mismatch in that frame. The frame number, deviation direction, and deviation magnitude are recorded to form multiple target resolution mismatch information.

[0039] Finally, for thermal imaging frames with insufficient resolution (i.e., target pixel scale is too small, deviation value is negative and exceeds the deviation threshold), bilinear interpolation is used: with the target area as the center, the resolution enhancement ratio is determined according to the deviation magnitude in the target resolution mismatch information, and the gray value of the newly added pixel is generated by calculating the weighted average of the gray values ​​of the target area and surrounding pixels, thus achieving the image resolution magnification adjustment; for thermal imaging frames with resolution redundancy (i.e., target pixel scale is too large, deviation value is positive and exceeds the deviation threshold), mean downsampling is used: the resolution reduction ratio is determined according to the deviation magnitude, the image is divided into several equal-sized pixel blocks, the gray value of all pixels in each pixel block is calculated, and this mean is used as the gray value of the new pixel, thus achieving the image resolution reduction adjustment; for thermal imaging frames without resolution mismatch, their original resolution remains unchanged, and finally all adjusted thermal imaging frames are integrated to generate the first thermal imaging sequence.

[0040] By employing a series of steps—threshold segmentation to detect target features, Euclidean distance matching to establish cross-field-of-view correlation, linear fitting to identify scale change trends, and interpolation and downsampling to adaptively adjust resolution—precise adaptation between thermal imaging frame resolution and target features was achieved, generating the first thermal imaging sequence that meets the requirements for cross-field-of-view target observation.

[0041] Furthermore, the method provided in this application embodiment includes: Based on the field-of-view state change chain, field-of-view switching node feature recognition is performed to obtain the field-of-view switching feature distribution; based on the field-of-view switching feature distribution, target detail drift is identified in the first thermal imaging sequence to obtain the detail drift feature trend; based on the field-of-view switching feature distribution, target contour drift is identified in the first thermal imaging sequence to obtain the contour drift feature trend; based on the field-of-view switching feature distribution, display jitter is captured in the first thermal imaging sequence to obtain the display jitter feature trend; based on the detail drift feature trend, the contour drift feature trend, and the display jitter feature trend, resolution progressive transition compensation analysis is performed on the first thermal imaging sequence to obtain the resolution progressive transition compensation strategy; based on the resolution progressive transition compensation strategy, resolution progressive transition adjustment is performed on the first thermal imaging sequence to generate the second thermal imaging sequence.

[0042] Specifically, the process first traverses the field of view state change chain, identifying the moment when adjacent label types change as field of view switching nodes. A preset number of thermal imaging frames are extracted before and after each field of view switching node, typically the 5 frames before and 5 frames after the switching node. The key features of these frames are analyzed, including resolution parameters, pixel ratio of the target area, and average image grayscale. The difference in the change of each feature before and after the switching node is calculated. The feature change differences of all switching nodes are arranged in the node time sequence to form a field of view switching feature distribution, which intuitively reflects the feature abrupt changes of different switching nodes.

[0043] When identifying target detail drift based on the distribution of field-of-view switching features, the target region in the thermal imaging frame one frame before the field-of-view switching node is selected, and the detailed features of this region, such as edge texture and grayscale gradient, are extracted as a reference template. The corresponding target regions in each frame after the field-of-view switching node are matched with the reference template, and the matching similarity is calculated using the normalized correlation coefficient method, with a value range of 0-1. The closer to 1, the higher the matching degree. When the similarity is lower than the preset threshold of 0.8, it is determined that there is detail drift. The time sequence and similarity difference of the drifted frames are recorded, and the detail drift feature trend is formed by time-series fitting to reflect the change law of drift.

[0044] When identifying target contour drift based on the distribution of field-of-view switching features, the contours of the target in each frame before and after the field-of-view switching node are extracted using the Canny edge detection algorithm to obtain the set of pixel coordinates of the contours. Taking the target contour of the frame before the field-of-view switching node as the reference contour, the average Euclidean distance between the target contour and the reference contour in each subsequent frame is calculated. The corresponding pixels on the two contours are traversed, and the square root of the sum of the squares of the coordinate differences is calculated. The average distance of all pixels is taken. When the distance value is greater than a preset threshold, contour drift is determined to exist. The distance values ​​are recorded in frame time sequence to form the contour drift feature trend. The preset threshold is set according to the target scale, which is usually 5% of the average radius of the target contour.

[0045] When capturing display jitter features based on the field-of-view switching characteristic distribution, the grayscale difference between corresponding pixels in adjacent frames before and after the field-of-view switching node is calculated. The number of pixels in each frame whose grayscale difference exceeds a preset jitter threshold is counted. The preset jitter threshold is set according to the noise level of the thermal imaging sensor and is usually 5 grayscale levels. Then, the proportion of this number of pixels to the total number of pixels in the image is used as the jitter intensity value. The jitter intensity value of each frame is recorded sequentially. When the jitter intensity value exceeds the preset proportion threshold of 10% for 3 consecutive frames, display jitter is determined to exist. The temporal change curve of the jitter intensity value is fitted to form the display jitter feature trend.

[0046] Next, a weighted fusion approach is used to perform progressive resolution transition compensation analysis. Based on the image quality requirements in practical applications, weights are assigned to detail drift feature trends, contour drift feature trends, and display jitter feature trends. Typically, detail drift weight is 0.4, contour drift weight is 0.3, and display jitter weight is 0.3. The comprehensive drift jitter coefficient for each field-of-view switching node is calculated, which is the sum of the feature values ​​of each trend corresponding to the frame multiplied by the corresponding weight. The resolution adjustment range is determined based on the comprehensive drift jitter coefficient. The larger the coefficient, the larger the adjustment range. The maximum adjustment range does not exceed 30% of the original resolution. At the same time, the adjustment rate is set to be adjusted gradually frame by frame, with the adjustment range of each frame not exceeding 20% ​​of the total range. The adjustment range and rate are integrated to form a progressive resolution transition compensation strategy.

[0047] Finally, starting from the field-of-view switching node, the resolution of each frame after the field-of-view switching node is gradually adjusted according to the adjustment rate set by the progressive resolution transition compensation strategy. For example, when transitioning from wide-angle resolution to telephoto resolution, the resolution of each frame is increased by a set step size. The gray value of the newly added pixels is calculated by linear interpolation, that is, based on the gray weighted average of adjacent pixels, to ensure that the resolution adjustment is smooth and without abrupt changes. All thermal imaging frames that have undergone resolution adjustment are integrated in time sequence to generate the second thermal imaging sequence.

[0048] Through the above-mentioned sequential steps, a smooth transition of image resolution is achieved during the field of view switching process, effectively reducing the impact of detail drift, contour drift and display jitter, and improving the image continuity during dual field of view switching.

[0049] Furthermore, the method provided in this application embodiment includes: Obtain the output frame rate parameter of the second thermal imaging sequence; if the output frame rate parameter is less than the predetermined output frame rate, perform inter-frame compensation reconstruction on the second thermal imaging sequence to generate the third thermal imaging sequence; if the output frame rate parameter is greater than the predetermined output frame rate, perform inter-frame fusion reconstruction on the second thermal imaging sequence to generate the third thermal imaging sequence; if the output frame rate parameter is equal to the predetermined output frame rate, output the second thermal imaging sequence as the third thermal imaging sequence.

[0050] Specifically, firstly, the acquisition timestamp information of each thermal imaging frame in the second thermal imaging sequence is extracted. This timestamp is the raw time data bound to each frame at the time of generation, and the format is uniformly set to millisecond-level time encoding. All thermal imaging frames are arranged sequentially according to their timestamps from smallest to largest to ensure the temporal continuity and integrity of the frame sequence. A continuous statistical time period is selected for frame rate calculation, prioritizing a complete time period of 1 second. If the total duration of the second thermal imaging sequence is less than 1 second, the total duration of the sequence is used as the statistical time period. The total number of thermal imaging frames contained within this statistical time period is counted, and the difference between the start and end timestamps of this time period is calculated to obtain the actual statistical duration, converted to seconds.

[0051] Then, the initial output frame rate parameter of the second thermal imaging sequence is calculated using the frame rate calculation formula: Output frame rate parameter = Total number of frames in the statistical period ÷ Actual statistical duration. To ensure parameter accuracy, the above statistical process is repeated, and the initial frame rate parameter is calculated for three different consecutive statistical periods. The arithmetic mean of the three initial parameters is taken as the final output frame rate parameter of the second thermal imaging sequence. If, during the statistical process, the difference between the timestamp of a frame and the timestamp of an adjacent frame exceeds a preset reasonable range (based on the conventional frame interval setting of the thermal imaging sensor, typically three times the average frame interval), the frame is determined to be a timestamp aberration frame. After removing the aberration frame, the total number of frames and frame rate calculation are repeated to avoid abnormal data affecting the accuracy of the output frame rate parameter.

[0052] Next, when the output frame rate parameter is less than the predetermined output frame rate, the compensation frame rate parameter is calculated first, and then the target motion feature recognition and resolution evolution feature recognition of the neighboring frames are performed on the second thermal imaging sequence to obtain the target motion evolution trend and resolution evolution trend of the neighboring frames. Finally, based on the compensation frame rate parameter, the second thermal imaging sequence is interpolated and compensated in combination with the above two evolution trends to generate the third thermal imaging sequence. This step will be explained in detail in the following content.

[0053] When the output frame rate parameter is greater than the predetermined output frame rate, the redundant frame rate parameter is first calculated. The redundant frame rate parameter is the output frame rate parameter minus the predetermined output frame rate, thus determining the number of frames to be reduced to quantify the fusion requirements. Similar to the inter-frame compensation and reconstruction steps described above, the motion features of neighboring targets in the second thermal imaging sequence are identified to obtain the motion evolution trend of neighboring targets. Simultaneously, the resolution evolution features of neighboring frames in the second thermal imaging sequence are identified to obtain the resolution evolution trend of neighboring frames. Based on the redundant frame rate parameter, inter-frame fusion reconstruction is performed on the second thermal imaging sequence according to the motion evolution trend and resolution evolution trend of neighboring targets. The fusion group division rules and the number of output frames after fusion in each group are determined according to the redundant frame rate parameter. A weighted fusion method is used to integrate the target features and resolution information of adjacent frames within each group, preserving clear target details, maintaining motion continuity and resolution progressiveness, and eliminating redundant frame information. All fused effective frames are integrated in time sequence to generate the third thermal imaging sequence, ensuring that the output frame rate parameter of the third thermal imaging sequence equals the predetermined output frame rate.

[0054] When the output frame rate parameter is equal to the predetermined output frame rate, the consistency between the output frame rate parameter of the second thermal imaging sequence and the predetermined output frame rate is first checked. After confirming that the two values ​​are completely matched and there is no temporal fluctuation, the second thermal imaging sequence is directly output as the third thermal imaging sequence without additional adjustment processing, ensuring that the image quality and temporal continuity remain unchanged.

[0055] Furthermore, the method provided in this application embodiment includes: If the output frame rate parameter is less than the predetermined output frame rate, calculate the compensation frame rate parameter; perform neighboring frame target motion feature recognition on the second thermal imaging sequence to obtain the neighboring frame target motion evolution trend; perform neighboring frame resolution evolution feature recognition on the second thermal imaging sequence to obtain the neighboring frame resolution evolution trend; based on the compensation frame rate parameter, perform frame interpolation compensation on the second thermal imaging sequence according to the neighboring frame target motion evolution trend and the neighboring frame resolution evolution trend to obtain the third thermal imaging sequence.

[0056] Specifically, the first step is to obtain the predetermined output frame rate and clarify the application scenarios of the dual-field-of-view thermal imaging sight, including different scenarios such as security patrol, hunting aiming, and field observation. Referring to industry frame rate standards for thermal imaging observation in each scenario (e.g., 30fps is commonly used for security patrol, and 25fps is commonly used for high-precision aiming), and combining the maximum output frame rate of the sight's built-in thermal imaging sensor, the real-time processing capability of the image processor, and the refresh rate of the display module, multiple candidate predetermined output frame rates are determined. Through multiple sets of actual tests in different scenarios, the image smoothness, processor load, and power consumption data corresponding to each candidate frame rate are recorded. The frame rate that satisfies the scene observation requirements in terms of image smoothness and that keeps the processor load and power consumption within a reasonable range is selected as the final predetermined output frame rate. This frame rate is a fixed value adapted to the sight's core working scenarios.

[0057] When the output frame rate parameter is less than the predetermined output frame rate, the compensation frame rate parameter is calculated according to the formula: Compensation frame rate parameter = Predetermined output frame rate - Output frame rate parameter. This parameter specifies the number of frames that need to be added, providing a quantitative basis for subsequent frame interpolation compensation.

[0058] The Lucas-Kanade optical flow method was used to identify the motion features of targets in adjacent frames in the second thermal imaging sequence. For two adjacent frames in the second thermal imaging sequence, feature points within the target area of ​​the previous frame are extracted. Pixels with significant grayscale changes are selected as feature points using the Shi-Tomasi corner detection algorithm. Specifically, a 3×3 neighborhood window is set. For each pixel within the target area, the Sobel operator is used to calculate its grayscale gradient in the x and y directions. A covariance matrix is ​​constructed based on the squared gradients in the x and y directions and the product of the x and y gradients of all pixels within the neighborhood window. Two eigenvalues ​​of this covariance matrix are solved, and an eigenvalue threshold is set. This threshold is calibrated through actual measurements of the grayscale distribution characteristics of thermal imaging images and is typically set to 500-800. When both eigenvalues ​​are greater than the threshold, the pixel is determined to be a corner point with significant grayscale changes, thus selecting feature points. The corresponding positions of these feature points in the next frame are calculated, and the optical flow vector of each feature point is obtained. The direction of the optical flow vector is the target motion direction, and the ratio of the vector's magnitude to the time interval between the two frames is the target motion velocity. Then, all adjacent frame pairs are traversed, and the motion direction and speed of the target in each frame pair are counted. The frames are arranged in time sequence to form the motion evolution trend of the target in adjacent frames, clearly presenting the continuous change law of the target motion state.

[0059] When identifying the resolution evolution features of adjacent frames in the second thermal imaging sequence, the resolution parameters of each frame image are extracted, namely the number of pixels in the horizontal direction and the number of pixels in the vertical direction. The change in resolution between two adjacent frames is calculated, which is the resolution parameter of the later frame minus the resolution parameter of the previous frame, and the rate of change is calculated, which is the change divided by the resolution parameter of the previous frame. The resolution change and rate of change of each pair of adjacent frames are recorded sequentially according to the frame time sequence to form the resolution evolution trend of adjacent frames, which intuitively reflects the continuous evolution of resolution.

[0060] Next, when performing frame interpolation compensation based on the compensation frame rate parameter, the total number of frames to be inserted is first determined by multiplying the compensation frame rate parameter by the total duration of the second thermal imaging sequence. The result is rounded up to the nearest integer; if the product is a decimal, it is rounded up to ensure that the frame rate compensation meets the predetermined standard. The number of interpolated frames between each adjacent frame pair is allocated according to the time interval ratio. Specifically, the time interval of all adjacent frame pairs in the second thermal imaging sequence is extracted, i.e., the difference between the timestamp of the next frame and the timestamp of the previous frame. The sum of the time intervals of all adjacent frame pairs is calculated, and then the proportion of the time interval of each adjacent frame pair to the total time interval is calculated. This proportion is multiplied by the total number of frames to be inserted to obtain the theoretical number of interpolated frames for each adjacent frame pair. If the theoretical value is a decimal, it is rounded to the nearest integer. Finally, the number of interpolated frames for all frame pairs is summed. If there is a difference between this sum and the total number of frames, the difference is allocated to the adjacent frame pair with the largest time interval, thus determining the number of interpolated frames for each adjacent frame pair.

[0061] For each set of adjacent frames requiring interpolation, the process of calculating the target's position coordinates in the interpolated frame is as follows: First, obtain the time interval between the adjacent frames in the set. Determine the time step for each interpolated frame based on the allocated number of interpolated frames; the time step is the sum of the time interval between adjacent frames divided by the number of interpolated frames plus one. Extract the target's motion direction and velocity from the target's motion evolution trend in the adjacent frames. Calculate the target's displacement within each time step; the displacement is the velocity multiplied by the time step. Then, decompose the displacement into horizontal and vertical displacements according to the motion direction. Using the target's position coordinates from the previous frame as a reference, add the corresponding number of horizontal and vertical displacements to the target's position coordinates in each interpolated frame. The number of displacements increases sequentially according to the order of the interpolated frames. This method achieves continuous calculation of the target's position, ensuring consistency with the target's motion state in the preceding and following frames.

[0062] When determining the resolution parameters of the interpolation frame, it is necessary to first clarify the resolution change amount and rate of change: Extract the resolution of the previous and next frames in the group of adjacent frames. Each resolution includes the number of pixels in the horizontal and vertical directions. Calculate the horizontal and vertical resolution change amounts, where the change amount is the resolution of the next frame in the corresponding direction minus the resolution of the previous frame in the same direction. Then calculate the horizontal and vertical resolution change rates, where the change rate is the resolution change amount in the corresponding direction divided by the resolution of the previous frame in the same direction. If the resolution in a certain direction of the previous frame is zero, then the resolution of the corresponding direction of the next frame is directly taken as the reference for the resolution of that direction in the interpolation frame. Based on the number of interpolation frames, the horizontal and vertical resolutions of each interpolation frame gradually change along the resolution evolution trend of adjacent frames. That is, based on the resolution of the previous frame in the corresponding direction, the corresponding proportion of the resolution change amount is added sequentially according to the interpolation frame order. The proportion is the sum of the number of interpolation frames divided by the number of interpolation frames plus one, avoiding abrupt resolution changes.

[0063] Finally, the grayscale value of each pixel in the interpolated frame is calculated using bilinear interpolation. The specific process is as follows: For any pixel position in the interpolated frame, the reference position of the pixel in the adjacent previous frame and the adjacent next frame are determined by combining the target motion offset. The two reference positions are processed separately. Four adjacent pixels around each reference position are taken, and the horizontal and vertical position ratios of the reference position relative to the top-left adjacent pixel are calculated. These ratios are the proportions of the distance between the reference position and the top-left pixel in the corresponding direction to the distance between adjacent pixels. The grayscale value of the reference position in the previous frame is calculated using bilinear interpolation logic, i.e., the grayscale values ​​of the four adjacent pixels are multiplied by their corresponding positional weights and then summed. Similarly, the grayscale value of the reference position in the next frame is calculated. Based on the temporal weights of interpolated frames between adjacent frames (the temporal weight is the sum of the number of interpolated frames divided by the number of interpolated frames plus one), the grayscale value of the reference position in the previous frame is multiplied by the complement of the temporal weight, and the grayscale value of the reference position in the next frame is multiplied by the temporal weight. The two are then added together to obtain the final grayscale value of the pixel. This weighted calculation ensures that the grayscale value of the interpolated frame is consistent with that of the preceding and following frames, thereby ensuring that the target position and resolution of the interpolated frame are consistent with those of the preceding and following frames. All generated interpolated frames are inserted into the corresponding positions of the second thermal imaging sequence according to the temporal order, and then integrated to generate the third thermal imaging sequence, making the output frame rate parameter of the third thermal imaging sequence equal to the predetermined output frame rate.

[0064] Furthermore, the method provided in this application embodiment includes: Based on the dual-field-of-view imaging loss recognition model, the imaging loss of the third thermal imaging sequence is analyzed for different fields of view, and an imaging loss cumulative effect map is constructed. Based on the imaging loss cumulative effect map, an imaging resolution adjustment decision is made for the third thermal imaging sequence to obtain an imaging resolution adjustment scheme. Based on the imaging loss cumulative effect map, an image processing resolution adjustment decision is made for the third thermal imaging sequence to obtain an image processing resolution adjustment scheme. Based on the imaging loss cumulative effect map, a display resolution adjustment decision is made for the third thermal imaging sequence to obtain a display resolution adjustment scheme. Based on the imaging resolution adjustment scheme, the image processing resolution adjustment scheme, and the display resolution adjustment scheme, the third thermal imaging sequence is subjected to multi-resolution hierarchical collaborative adjustment to generate the fourth thermal imaging sequence.

[0065] In one embodiment, firstly, a dual-field-of-view imaging loss recognition model, which includes a wide-angle thermal imaging loss recognition model and a telephoto thermal imaging loss recognition model, is activated. Wide-angle and telephoto imaging loss recognition is performed on the third thermal imaging sequence to obtain the corresponding loss distribution. After constructing the multi-field-of-view imaging loss temporal distribution through temporal alignment and correlation mapping, an imaging loss cumulative effect map is generated by analyzing the cross-field-of-view superposition effect. This step will be explained in detail in the following content.

[0066] When making imaging resolution adjustment decisions based on the cumulative effect of imaging loss map, a loss severity threshold matching method is adopted. First, the field of view type (wide-angle or telephoto), loss type, and quantified loss severity value for each frame are extracted from the cumulative effect of imaging loss map. A preset loss severity grading threshold is established, classifying the loss severity into three levels: mild, moderate, and severe. This threshold is calibrated using multiple sets of thermal imaging measurement data to ensure consistency with perceived image quality. For wide-angle fields of view, if the background thermal noise interference index is moderate or severe, the pixel sampling rate of the imaging resolution is increased; specifically, the sampling rate is increased by 20% for mild loss, 40% for moderate loss, and 60% for severe loss. If the field of view edge distortion index is moderate or severe, the pixel distribution density of the lens imaging is optimized, with the pixel density in the edge region increased by 30%-50% compared to the center region. If the imaging stability attenuation index exceeds the limit, the sampling rate is kept stable, and the inter-frame sampling synchronization is optimized. For telephoto fields of view, if the target structure distortion index is moderate or severe, increase the imaging pixel accuracy of the target area by 30%-70% depending on the degree of loss. If the target jitter amplification or imaging sharpness degradation index exceeds the standard, adjust the coordination parameters of imaging frame rate and pixel sampling to ensure imaging stability at high resolution. Organize the adjustment parameters of each frame in chronological order to form an imaging resolution adjustment scheme, specifying the specific settings of sampling rate, pixel distribution, accuracy parameters, etc. for each frame.

[0067] When making image processing resolution adjustment decisions based on the cumulative effect of imaging loss map, a loss type-algorithm parameter mapping method is adopted. First, the loss type combinations of each frame in the map are analyzed, and corresponding image processing algorithms and resolution parameters are matched for different loss types. When background thermal noise interference exists, a Gaussian noise reduction algorithm is activated, and the resolution adaptation value for noise reduction processing is set according to the degree of loss. For mild loss, the processing resolution of the algorithm is consistent with the imaging resolution; for moderate and severe loss, the processing resolution is increased by 20% and 40% respectively, enhancing the noise filtering effect. When there is distortion at the edge of the field of view, a perspective correction algorithm is used, adjusting the processing resolution dimension of the algorithm based on the degree of distortion to ensure the accuracy of edge region correction. The more severe the distortion, the higher the edge adaptation coefficient of the processing resolution. When there is target structure distortion or sharpness reduction, an edge enhancement algorithm is activated, and the resolution threshold for enhancement processing is set according to the degree of loss. For mild loss, the processing resolution is 1.2 times the imaging resolution; for moderate loss, it is 1.5 times; and for severe loss, it is 2.0 times, enhancing the details of the target structure. By integrating the resolution parameters and activation logic of various algorithms, an image processing resolution adjustment scheme is formed, clarifying the algorithm selection and resolution adaptation standards under different loss scenarios.

[0068] When making display resolution adjustment decisions based on the cumulative effect map of imaging loss, a dynamic adaptation method for display parameters is adopted. The overall loss level and temporal trend of each frame in the map are extracted, and combined with the hardware parameters of the dual-field-of-view thermal imaging sight display module, such as maximum display pixels and refresh rate, to set display resolution adjustment rules. When the overall loss level is mild, the display resolution remains consistent with the imaging resolution; for moderate loss, the display resolution is adapted to 1.2 times the imaging resolution, optimizing pixel rendering density; for severe loss, the display resolution is increased to 1.5 times the imaging resolution, and pixel interpolation optimization technology is enabled to ensure a smooth, jagged-free display. For jitter-type losses that occur consecutively in time, the adaptation relationship between the display refresh rate and resolution is adjusted synchronously. When jitter is severe, the refresh rate and display resolution are adjusted in a 1:1.3 ratio to improve image smoothness. Based on the adjustment requirements of each frame, a display resolution adjustment scheme is formed, specifying the number of display pixels, rendering density, refresh rate adaptation parameters, etc., for each frame.

[0069] Finally, a weighted allocation and integration approach is adopted for multi-resolution hierarchical coordinated adjustment. Preset adjustment weights for imaging resolution, image processing resolution, and display resolution are 40%, 30%, and 30%, respectively. These weights are determined based on the link influence relationship between imaging, processing, and display, ensuring optimal synergistic adjustment effects. For each frame of thermal imaging, the parameters of the three adjustment schemes are fused according to their weights. If there are conflicts in the adjustment parameters of different schemes, the imaging resolution adjustment scheme takes precedence, prioritizing the basic quality of the acquisition end. The fused adjustment parameters are applied frame by frame to the third thermal imaging sequence. First, the original imaging data is adjusted according to the imaging resolution adjustment scheme, then the algorithm is optimized using the image processing resolution adjustment scheme, and finally the output format is adapted according to the display resolution adjustment scheme, ensuring consistent resolution parameters at each stage. After traversing all frames and completing the adjustment, the results are integrated to form the fourth thermal imaging sequence, achieving coordinated adaptation of resolutions at each level.

[0070] By employing a coherent approach involving loss threshold matching, type-parameter mapping, dynamic display adaptation, and weight integration, resolution adjustment schemes at the imaging, image processing, and display levels were developed and applied collaboratively. This enabled precise resolution feedback adjustment based on imaging loss analysis, effectively improving the image quality and adaptability of the dual-field-of-view thermal imaging sight.

[0071] Furthermore, the method provided in this application embodiment includes: The dual-field-of-view imaging loss identification model is activated, which includes a wide-angle thermal imaging loss identification model and a telephoto thermal imaging loss identification model. Wide-angle thermal imaging loss identification is performed on the third thermal imaging sequence using the wide-angle thermal imaging loss identification model to obtain a first field-of-view imaging loss distribution. Telephoto thermal imaging loss identification is performed on the third thermal imaging sequence using the telephoto thermal imaging loss identification model to obtain a second field-of-view imaging loss distribution. Temporal alignment and correlation mapping are performed based on the first and second field-of-view imaging loss distributions to construct a multi-field-of-view imaging loss temporal distribution. Cross-field-of-view superposition effect analysis is performed on the multi-field-of-view imaging loss temporal distribution to generate the imaging loss cumulative effect map.

[0072] Optionally, a dual-field-of-view imaging loss recognition model is activated. This model includes a wide-angle thermal imaging loss recognition model and a telephoto thermal imaging loss recognition model. The wide-angle thermal imaging loss recognition model is built based on the random forest algorithm. The specific construction and training process is as follows: First, the model selection is determined to be random forest. This algorithm supports direct input of multi-dimensional quantized features, has strong anti-noise interference capabilities, and does not require complex feature transformation, which meets the needs of thermal imaging loss recognition scenarios. Next, three wide-angle thermal imaging loss recognition indicators are extracted and quantified. This step will be explained in detail later. Subsequently, sample preparation is carried out. 100-200 frames of wide-angle thermal imaging samples are collected under different environmental scenarios using a dual-field-of-view thermal imaging scope, covering four scenarios: no loss, slight loss, moderate loss, and severe loss. These samples are labeled by those skilled in the art according to "loss type + loss degree", such as background thermal noise - slight, field-of-view edge distortion - moderate, etc. Each sample's three quantified metrics were used as feature vectors, and the datasets were divided into training and testing sets in an 8:2 ratio. A random forest model was loaded using the open-source scikit-learn framework, with the following core parameters set: 80 decision trees, a maximum tree depth of 6 layers, and a minimum number of samples per leaf node (5). Other parameters remained at their default values. The feature vectors and labeled parameters were then input into the model for training. After training, the model's performance was validated using the test set. The model was considered complete and ready for use in wide-angle thermal imaging loss recognition when the loss type identification accuracy was no less than 90% and the loss degree quantization error was no more than 10%.

[0073] The telephoto thermal imaging loss recognition model employs the same random forest algorithm and core training process as the wide-angle thermal imaging loss recognition model, only replacing the input features with three telephoto thermal imaging loss recognition metrics. The specific quantification methods will be explained in detail later. Sample preparation, dataset partitioning, model parameter settings, and training and validation standards are all consistent with the wide-angle thermal imaging loss recognition model.

[0074] Next, the thermal imaging frames corresponding to all wide-angle fields of view in the third thermal imaging sequence are input frame by frame into the trained wide-angle thermal imaging loss recognition model. The model outputs the loss type of each frame, including background thermal noise interference, field-of-view edge distortion, imaging stability attenuation or composite loss, as well as the loss degree quantification value. These recognition results are integrated in the temporal order of the frames to form the first field-of-view imaging loss distribution. This distribution contains the temporal information, loss type and corresponding loss degree data of each frame of wide-angle image.

[0075] Then, using the same frame-by-frame input method as the wide-angle loss recognition, all thermal imaging frames corresponding to the telephoto field of view in the third thermal imaging sequence are input into the telephoto thermal imaging loss recognition model. The model outputs the loss type of each frame, including target structure distortion, target jitter amplification, image sharpness reduction or composite loss, as well as the loss degree quantification value. After being integrated in the temporal order of the frames, the second field of view imaging loss distribution is formed, which includes the temporal information, loss type and corresponding loss degree data of each telephoto image frame.

[0076] Subsequently, for the first and second field-of-view imaging loss distributions, timestamp information of all frames in both distributions was extracted. The frames of the two distributions were then mapped one-to-one using the timestamps. For time points with only single field-of-view loss data, the loss information for that field of view was retained, and the other field of view was marked as loss-free. Then, the Pearson correlation coefficient method was used to calculate the correlation between the wide-angle and telephoto loss levels at the same time point, quantifying their mutual influence. The time-aligned loss data and correlation data were integrated to construct a multi-field-of-view imaging loss temporal distribution. This distribution clearly presents the loss status and correlation characteristics of the wide-angle and telephoto fields of view at different time points.

[0077] Finally, the superposition of losses is analyzed segment by segment in chronological order: when there is only a single field-of-view loss at the same time point, the type and degree of the loss are directly retained; when there is a dual field-of-view loss at the same time point, the comprehensive degree of loss after superposition is calculated according to the influence weight of the loss type, which is specifically set by those skilled in the art based on the needs of thermal imaging observation, such as the target structure distortion weight being higher than the background thermal noise interference. At the same time, the duration and trend of loss superposition are recorded, and these analysis results are presented in the form of a graph to generate an imaging loss cumulative effect graph. The graph contains complete information in the time sequence dimension, loss type dimension, loss degree dimension, and cross-field superposition dimension.

[0078] By constructing and training a random forest model, dual-field-of-view exclusive loss recognition is achieved. Through temporal alignment, correlation mapping, and cross-field-of-view overlay analysis, an imaging loss cumulative effect map is accurately constructed, providing comprehensive and reliable loss data support for subsequent multi-level resolution adjustment.

[0079] Furthermore, the method provided in this application embodiment includes: The wide-angle thermal imaging loss recognition model includes background thermal noise interference, field-of-view edge distortion, and imaging stability attenuation indicators. The telephoto thermal imaging loss recognition model includes target structure distortion, target jitter amplification, and imaging sharpness attenuation indicators.

[0080] In one embodiment, the wide-angle thermal imaging loss identification indicators first include background thermal noise interference indicators, field-of-view edge distortion indicators, and imaging stability attenuation indicators. The specific acquisition and quantification process is as follows: When extracting the background thermal noise interference index, firstly, a non-target background region is selected from the wide-angle thermal imaging frame. This region must avoid the target subject and the edge of the field of view. Using an automatic selection tool, a continuous background region with an area no less than 10% of the total image area is selected, ensuring that there are no obvious structural features within the region. The grayscale values ​​of all pixels within this region are extracted, and the arithmetic mean of these grayscale values ​​is calculated. Then, the difference between each pixel's grayscale value and the mean is calculated. All differences are squared and summed. The sum is then divided by the total number of pixels in the region. The result is the quantified value of the background thermal noise interference index; the larger the value, the more severe the background thermal noise interference.

[0081] When extracting the field-of-view edge distortion index, the wide-angle thermal imaging lens is first calibrated to obtain the lens's ideal distortion-free parameters. Based on these parameters, a set of pixel coordinates for the ideal field-of-view edge is generated, where the ideal edge is a regular rectangular outline. The Canny edge detection algorithm is used to extract the field-of-view edge pixels of the actual wide-angle thermal imaging frame, obtaining the set of pixel coordinates for the actual edge. A one-to-one matching process is performed between the pixels of the ideal edge and the actual edge. Feature points on the edge are selected in the same order, and the Euclidean distance between each corresponding feature point is calculated. The sum of all Euclidean distances is divided by the total number of feature points, and the resulting average offset is the quantified value of the field-of-view edge distortion index. A larger average offset indicates a higher degree of field-of-view edge distortion.

[0082] When extracting the imaging stability attenuation index, ten consecutive wide-angle thermal imaging frames are selected. A fixed reference region is defined in the first frame; this region can be a stable structure or target subject in the background (if the target is stationary). The pixel coordinate range of this region is recorded and maintained in subsequent frames. The grayscale values ​​of all pixels in this reference region are extracted in each frame, and the mean grayscale value of the reference region in each frame is calculated, resulting in ten mean grayscale data points. The standard deviation of these ten mean grayscale values ​​is calculated; the standard deviation is the quantified value of the imaging stability attenuation index. A larger standard deviation indicates poorer imaging stability and a more severe degree of attenuation.

[0083] Next, the loss identification indicators for telephoto thermal imaging include target structure distortion indicators, target jitter amplification indicators, and image sharpness attenuation indicators. The specific extraction and quantification process is as follows: When extracting the target structure distortion index, the Canny edge detection algorithm is used to extract the edges of the target region in the telephoto thermal imaging frame, obtaining the actual contour pixel coordinate set of the target. The standard contour data of the target is then obtained. The standard contour can be determined by the contour extraction results of this type of target under clear imaging conditions, or by using a preset target template contour. The actual contour is aligned with the standard contour, and position calibration is performed using the geometric center of the contour as a reference. The ratio of the number of overlapping pixels between the two contours to the total number of pixels in the standard contour is calculated. Subtracting this ratio from 1 yields the quantized value of the target structure distortion index; a larger value indicates more severe target structure distortion.

[0084] When extracting the target jitter amplification index, target detection is performed on multiple consecutive long-range thermal imaging frames to determine the set of target contour pixel coordinates in each frame. The centroid coordinates of the target contour in each frame are calculated. The centroid x-coordinate is obtained by calculating the arithmetic mean of the x-coordinates of all pixels in the contour, and the centroid y-coordinate is obtained by calculating the arithmetic mean of the y-coordinates of all pixels in the contour. The Euclidean distance between the target centroid coordinates of two adjacent frames is calculated to obtain the target displacement of each adjacent frame pair. The average displacement obtained by summing the displacements of all adjacent frame pairs and dividing by the number of frame pairs is the quantized value of the target jitter amplification index. The larger the average displacement, the more obvious the target jitter amplification.

[0085] When extracting the image sharpness attenuation index, the entire target area is first selected within the telephoto thermal imaging frame, ensuring coverage of all critical structures. The Sobel operator is then used to process this target area, calculating the grayscale gradient values ​​of each pixel in the x and y directions. Convolution operations using gradient operators are then performed to obtain the gradient matrices in the x and y directions. For each pixel, the sum of the squares of the x and y gradient values ​​is taken, and the square root is used to obtain the gradient magnitude for that pixel. The arithmetic mean of the gradient magnitudes of all pixels within the target area is calculated; this mean is the quantized value of the image sharpness attenuation index. A smaller mean value indicates lower image sharpness and a more severe degree of attenuation.

[0086] In summary, the image resolution adaptive adjustment method for the dual-field thermal imaging sight provided in this application has the following technical effects: This application acquires thermal imaging frames through a dual-field-of-view thermal imaging sight, performs target-associative adaptive resolution adjustment, progressive resolution transition optimization, and frame rate constraint reconstruction, and then analyzes the imaging loss using a dual-field-of-view imaging loss recognition model. A multi-level resolution adjustment scheme is formulated and coordinated to obtain a high-quality fourth thermal imaging sequence, thereby improving imaging quality and adaptability. This achieves the technical effect of adaptive adaptation of image resolution under dual-field-of-view, improving image coherence, clarity, and stability.

[0087] Example 2, as Figure 2As shown, based on the same inventive concept as in Embodiment 1 above, this application provides an image resolution adaptive adjustment system for a dual-field-of-view thermal imaging sight, the system comprising: Thermal imaging frame acquisition module 1 is used to monitor the target area through a dual-field-of-view thermal imaging sight and acquire multiple thermal imaging frames corresponding to multiple field-of-view status labels.

[0088] The first thermal imaging sequence acquisition module 2 is used to perform target association adaptive resolution adjustment based on the plurality of thermal imaging frames to acquire a first thermal imaging sequence.

[0089] The second thermal imaging sequence acquisition module 3 is used to perform time-series sorting of the multiple field-of-view state labels, construct a field-of-view state change chain, and perform resolution progressive transition optimization on the first thermal imaging sequence according to the field-of-view state change chain to acquire the second thermal imaging sequence.

[0090] The third thermal imaging sequence acquisition module 4 is used to perform frame rate constraint reconstruction on the second thermal imaging sequence according to a predetermined output frame rate to acquire the third thermal imaging sequence.

[0091] The fourth thermal imaging sequence acquisition module 5 is used to introduce a dual-field-of-view imaging loss recognition model to perform resolution feedback adjustment on the third thermal imaging sequence under different field-of-view imaging loss analysis, and to acquire the fourth thermal imaging sequence.

[0092] Furthermore, the first thermal imaging sequence acquisition module 2 is used to perform the following steps: Target detection is performed on the multiple thermal imaging frames to obtain target position features and target pixel scale features for each frame; cross-field-of-view target association analysis is performed on the target position features and target pixel scale features for each frame under different field-of-view state labels to establish cross-field-of-view target association relationships; target pixel scale change trend identification is performed on the multiple thermal imaging frames based on the cross-field-of-view target association relationships to obtain multiple target resolution mismatch information; adaptive resolution adjustment is performed on the multiple thermal imaging frames based on the multiple target resolution mismatch information to generate the first thermal imaging sequence.

[0093] Furthermore, the second thermal imaging sequence acquisition module 3 is used to perform the following steps: Based on the field-of-view state change chain, field-of-view switching node feature recognition is performed to obtain the field-of-view switching feature distribution; based on the field-of-view switching feature distribution, target detail drift is identified in the first thermal imaging sequence to obtain the detail drift feature trend; based on the field-of-view switching feature distribution, target contour drift is identified in the first thermal imaging sequence to obtain the contour drift feature trend; based on the field-of-view switching feature distribution, display jitter is captured in the first thermal imaging sequence to obtain the display jitter feature trend; based on the detail drift feature trend, the contour drift feature trend, and the display jitter feature trend, resolution progressive transition compensation analysis is performed on the first thermal imaging sequence to obtain the resolution progressive transition compensation strategy; based on the resolution progressive transition compensation strategy, resolution progressive transition adjustment is performed on the first thermal imaging sequence to generate the second thermal imaging sequence.

[0094] Furthermore, the third thermal imaging sequence acquisition module 4 is used to perform the following steps: Obtain the output frame rate parameter of the second thermal imaging sequence; if the output frame rate parameter is less than the predetermined output frame rate, perform inter-frame compensation reconstruction on the second thermal imaging sequence to generate the third thermal imaging sequence; if the output frame rate parameter is greater than the predetermined output frame rate, perform inter-frame fusion reconstruction on the second thermal imaging sequence to generate the third thermal imaging sequence; if the output frame rate parameter is equal to the predetermined output frame rate, output the second thermal imaging sequence as the third thermal imaging sequence.

[0095] Furthermore, the third thermal imaging sequence acquisition module 4 is used to perform the following steps: If the output frame rate parameter is less than the predetermined output frame rate, calculate the compensation frame rate parameter; perform neighboring frame target motion feature recognition on the second thermal imaging sequence to obtain the neighboring frame target motion evolution trend; perform neighboring frame resolution evolution feature recognition on the second thermal imaging sequence to obtain the neighboring frame resolution evolution trend; based on the compensation frame rate parameter, perform frame interpolation compensation on the second thermal imaging sequence according to the neighboring frame target motion evolution trend and the neighboring frame resolution evolution trend to obtain the third thermal imaging sequence.

[0096] Furthermore, the fourth thermal imaging sequence acquisition module 5 is used to perform the following steps: Based on the dual-field-of-view imaging loss recognition model, the imaging loss of the third thermal imaging sequence is analyzed for different fields of view, and an imaging loss cumulative effect map is constructed. Based on the imaging loss cumulative effect map, an imaging resolution adjustment decision is made for the third thermal imaging sequence to obtain an imaging resolution adjustment scheme. Based on the imaging loss cumulative effect map, an image processing resolution adjustment decision is made for the third thermal imaging sequence to obtain an image processing resolution adjustment scheme. Based on the imaging loss cumulative effect map, a display resolution adjustment decision is made for the third thermal imaging sequence to obtain a display resolution adjustment scheme. Based on the imaging resolution adjustment scheme, the image processing resolution adjustment scheme, and the display resolution adjustment scheme, the third thermal imaging sequence is subjected to multi-resolution hierarchical collaborative adjustment to generate the fourth thermal imaging sequence.

[0097] Furthermore, the fourth thermal imaging sequence acquisition module 5 is used to perform the following steps: The dual-field-of-view imaging loss identification model is activated, which includes a wide-angle thermal imaging loss identification model and a telephoto thermal imaging loss identification model. Wide-angle thermal imaging loss identification is performed on the third thermal imaging sequence using the wide-angle thermal imaging loss identification model to obtain a first field-of-view imaging loss distribution. Telephoto thermal imaging loss identification is performed on the third thermal imaging sequence using the telephoto thermal imaging loss identification model to obtain a second field-of-view imaging loss distribution. Temporal alignment and correlation mapping are performed based on the first and second field-of-view imaging loss distributions to construct a multi-field-of-view imaging loss temporal distribution. Cross-field-of-view superposition effect analysis is performed on the multi-field-of-view imaging loss temporal distribution to generate the imaging loss cumulative effect map.

[0098] Furthermore, the thermal imaging frame acquisition module 1 is used to perform the following steps: The dual-field-of-view thermal imaging sight includes a wide-angle thermal imaging lens and a telephoto thermal imaging lens.

[0099] Furthermore, the fourth thermal imaging sequence acquisition module 5 is used to perform the following steps: The wide-angle thermal imaging loss recognition model includes background thermal noise interference, field-of-view edge distortion, and imaging stability attenuation indicators. The telephoto thermal imaging loss recognition model includes target structure distortion, target jitter amplification, and imaging sharpness attenuation indicators.

[0100] The image resolution adaptive adjustment system of the dual-field thermal imaging sight provided in this embodiment of the invention can execute the image resolution adaptive adjustment method of the dual-field thermal imaging sight provided in any embodiment of the invention, and has the corresponding functional modules and beneficial effects of the method.

[0101] Although this application makes various references to certain modules in the system according to the embodiments of this application, any number of different modules can be used and run on user terminals and / or servers. The various units and modules included are only divided according to functional logic, but are not limited to the above division, as long as the corresponding functions can be achieved; in addition, the specific names of each functional unit are only for easy distinction between each other and are not used to limit the scope of protection of this invention.

[0102] The specific embodiments described above do not constitute a limitation on the scope of protection of this application. Those skilled in the art should understand that various modifications, combinations, and substitutions can be made according to design requirements and other factors. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of this application should be included within the scope of protection of this application. In some cases, the actions or steps described in this application can be performed in a different order than that shown in the embodiments and still achieve the desired results. Furthermore, the processes depicted in the accompanying drawings do not necessarily require a specific or sequential order to achieve the desired results. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.

Claims

1. An image resolution adaptive adjustment method for a dual-field-of-view thermal imaging sight, characterized in that, The method includes: The target area is monitored by a dual-field-of-view thermal imaging sight, and multiple thermal imaging frames corresponding to multiple field-of-view status labels are acquired. Based on the multiple thermal imaging frames, target association adaptive resolution adjustment is performed to obtain a first thermal imaging sequence; The multiple field-of-view state labels are sequentially analyzed to construct a field-of-view state change chain. Based on the field-of-view state change chain, the first thermal imaging sequence is optimized for progressive resolution transition to obtain a second thermal imaging sequence. The second thermal imaging sequence is reconstructed by frame rate constraint according to the predetermined output frame rate to obtain the third thermal imaging sequence. A dual-field-of-view imaging loss identification model is introduced to adjust the resolution of the third thermal imaging sequence under different field-of-view imaging loss analysis, thereby obtaining the fourth thermal imaging sequence.

2. The image resolution adaptive adjustment method of the dual-field-of-view thermal imaging sight as described in claim 1, characterized in that, Based on the multiple thermal imaging frames, target association adaptive resolution adjustment is performed to obtain a first thermal imaging sequence, including: Target detection is performed on the multiple thermal imaging frames to obtain the target position features and target pixel scale features of each frame; Cross-view target association analysis is performed on the target position features and target pixel scale features of each frame under different view state labels to establish cross-view target association relationships; Based on the cross-field-of-view target association relationship, target pixel scale change trend identification is performed on the multiple thermal imaging frames to obtain multiple target resolution mismatch information; Based on the multiple target resolution mismatch information, the resolution of the multiple thermal imaging frames is adaptively adjusted to generate the first thermal imaging sequence.

3. The image resolution adaptive adjustment method of the dual-field-of-view thermal imaging sight as described in claim 1, characterized in that, Based on the field-of-view state change chain, the first thermal imaging sequence is optimized for progressive resolution transition to obtain a second thermal imaging sequence, including: Based on the field of view state change chain, the field of view switching node feature is identified to obtain the field of view switching feature distribution; Based on the field-of-view switching feature distribution, target detail drift is identified in the first thermal imaging sequence to obtain the detail drift feature trend. Based on the field-of-view switching feature distribution, target contour drift is identified in the first thermal imaging sequence to obtain contour drift feature trends; Based on the field-of-view switching feature distribution, the first thermal imaging sequence is subjected to display jitter feature capture to obtain the display jitter feature trend; Based on the detail drift feature trend, the contour drift feature trend, and the display jitter feature trend, the first thermal imaging sequence is analyzed for progressive resolution transition compensation to obtain a progressive resolution transition compensation strategy. The first thermal imaging sequence is adjusted for progressive resolution transition according to the resolution progressive transition compensation strategy to generate the second thermal imaging sequence.

4. The image resolution adaptive adjustment method of the dual-field-of-view thermal imaging sight as described in claim 1, characterized in that, The second thermal imaging sequence is reconstructed using a frame rate constraint based on a predetermined output frame rate to obtain a third thermal imaging sequence, including: Obtain the output frame rate parameter of the second thermal imaging sequence; If the output frame rate parameter is less than the predetermined output frame rate, the second thermal imaging sequence is reconstructed by inter-frame compensation to generate the third thermal imaging sequence. If the output frame rate parameter is greater than the predetermined output frame rate, the second thermal imaging sequence is subjected to inter-frame fusion reconstruction to generate the third thermal imaging sequence. If the output frame rate parameter is equal to the predetermined output frame rate, the second thermal imaging sequence is output as the third thermal imaging sequence.

5. The image resolution adaptive adjustment method of the dual-field-of-view thermal imaging sight as described in claim 4, characterized in that, If the output frame rate parameter is less than the predetermined output frame rate, inter-frame compensation reconstruction is performed on the second thermal imaging sequence to generate the third thermal imaging sequence, including: If the output frame rate parameter is less than the predetermined output frame rate, calculate the compensation frame rate parameter; The motion features of neighboring targets in the second thermal imaging sequence are identified to obtain the motion evolution trend of neighboring targets. The neighboring frame resolution evolution feature is identified in the second thermal imaging sequence to obtain the neighboring frame resolution evolution trend; Based on the compensation frame rate parameter, the second thermal imaging sequence is interpolated and compensated according to the motion evolution trend of the neighboring frame target and the resolution evolution trend of the neighboring frame to obtain the third thermal imaging sequence.

6. The image resolution adaptive adjustment method of the dual-field-of-view thermal imaging sight as described in claim 1, characterized in that, A dual-field-of-view imaging loss identification model is introduced to perform resolution feedback adjustment on the third thermal imaging sequence under different field-of-view imaging loss analyses, thereby obtaining a fourth thermal imaging sequence, including: Based on the dual-field-of-view imaging loss identification model, the imaging loss of the third thermal imaging sequence is analyzed for different fields of view, and an imaging loss cumulative effect map is constructed. Based on the imaging loss cumulative effect map, an imaging resolution adjustment decision is made for the third thermal imaging sequence to obtain an imaging resolution adjustment scheme. Based on the imaging loss cumulative effect map, an image processing resolution adjustment decision is made for the third thermal imaging sequence to obtain an image processing resolution adjustment scheme. Based on the imaging loss cumulative effect map, a display resolution adjustment decision is made for the third thermal imaging sequence to obtain a display resolution adjustment scheme. The third thermal imaging sequence is adjusted in a multi-resolution hierarchical manner according to the imaging resolution adjustment scheme, the image processing resolution adjustment scheme, and the display resolution adjustment scheme to generate the fourth thermal imaging sequence.

7. The image resolution adaptive adjustment method of the dual-field-of-view thermal imaging sight as described in claim 6, characterized in that, Based on the dual-field-of-view imaging loss identification model, the imaging loss of the third thermal imaging sequence is analyzed for different fields of view, and an imaging loss cumulative effect map is constructed, including: Activate the dual-field-of-view imaging loss recognition model, which includes a wide-angle thermal imaging loss recognition model and a telephoto thermal imaging loss recognition model. Based on the wide-angle thermal imaging loss identification model, wide-angle thermal imaging loss identification is performed on the third thermal imaging sequence to obtain the first field-of-view imaging loss distribution. Based on the long-focus thermal imaging loss identification model, the long-focus thermal imaging loss identification is performed on the third thermal imaging sequence to obtain the second field of view imaging loss distribution. Based on the first field-of-view imaging loss distribution and the second field-of-view imaging loss distribution, temporal alignment and correlation mapping are performed to construct a multi-field-of-view imaging loss temporal distribution. The cross-field superposition effect is analyzed on the temporal distribution of the multi-field imaging loss to generate the cumulative effect map of the imaging loss.

8. The image resolution adaptive adjustment method of the dual-field-of-view thermal imaging sight as described in claim 1, characterized in that, The dual-field-of-view thermal imaging sight includes a wide-angle thermal imaging lens and a telephoto thermal imaging lens.

9. The image resolution adaptive adjustment method of the dual-field-of-view thermal imaging sight as described in claim 7, characterized in that, The wide-angle thermal imaging loss identification model includes the following wide-angle thermal imaging loss identification indices: background thermal noise interference index, field of view edge distortion index, and imaging stability attenuation index. The long-focus thermal imaging loss recognition model includes target structure distortion index, target jitter amplification index, and imaging sharpness attenuation index.

10. An image resolution adaptive adjustment system for a dual-field-of-view thermal imaging sight, characterized in that, The system is used to implement the image resolution adaptive adjustment method of the dual-field-of-view thermal imaging sight according to any one of claims 1-9, the system comprising: The thermal imaging frame acquisition module is used to monitor the target area through a dual-field-of-view thermal imaging sight and acquire multiple thermal imaging frames corresponding to multiple field-of-view status labels. The first thermal imaging sequence acquisition module is used to perform target association adaptive resolution adjustment based on the plurality of thermal imaging frames to acquire the first thermal imaging sequence. The second thermal imaging sequence acquisition module is used to perform time-series sorting of the multiple field-of-view state labels, construct a field-of-view state change chain, and perform resolution progressive transition optimization on the first thermal imaging sequence according to the field-of-view state change chain to acquire the second thermal imaging sequence. The third thermal imaging sequence acquisition module is used to perform frame rate constraint reconstruction on the second thermal imaging sequence according to a predetermined output frame rate to acquire the third thermal imaging sequence. The fourth thermal imaging sequence acquisition module is used to introduce a dual-field-of-view imaging loss recognition model to perform resolution feedback adjustment on the third thermal imaging sequence under different field-of-view imaging loss analysis, and to acquire the fourth thermal imaging sequence.

Images