Automatic scene reconstruction method

Abstract

The invention relates to the technical field of image processing, in particular to an automatic scene reconstruction method. The method comprises the steps; s1, carrying out image fuzzy detection and quality evaluation; s2, de-noising processing is carried out; s3, obtaining a criticality evaluation index; s4, performing emergency treatment grading and injury prediction; according to the method, a visual image is acquired at a fixed acquisition frequency, clear and blurred images are divided through a Laplace operator, an initial damage contour is positioned after noise estimation and denoising are performed on the blurred image, and a key contour is screened based on an AIS score and a radiation intensity model. Performing orthogonal projection block gray variance analysis on the target contour to judge a fuzzy ratio and judge whether image blurring exists in a key area, and predicting a damage reason in combination with an emergency treatment grading standard and simulation; according to the technology, the reconstruction precision and the response speed in an emergency treatment scene are improved, powerful auxiliary diagnosis help is provided for diagnosis, and powerful support is provided for rapidly formulating a precise treatment scheme.

Description

Technical Field

[0001] The present invention relates to the technical field of video segmentation, and in particular to a method for automatic scene reconstruction. Background Art

[0002] With the widespread use of mobile devices and the continuous improvement of network speeds, short, concise content with high traffic and high reach has become increasingly popular among major platforms, user groups, and the capital market. As an emerging form of internet communication, short videos have not only quickly gained popularity but also helped promote and disseminate related drama video resources.

[0003] The patent document of Chinese Patent Publication No. CN118982611A discloses a method for generating a Gaussian model for scene reconstruction and a method for scene reconstruction. The technical point is to extract point cloud data from the image sequence to be trained and construct a multi-level anchor Gaussian set, further spatially divide the image sequence, determine the image acquisition device corresponding to each partition, and project anchor points of different levels to each partition to generate a corresponding training set, and construct a constraint expression for each image based on the image sequence and three-dimensional Gaussian parameters; combine the training set and the constraint expression and input the model as training data to optimize the scene reconstruction effect; at the same time, the method relies on layered training of Gaussian models and point cloud data, and the technical focus is on three-dimensional modeling accuracy rather than application in specific fields. At the same time, the model training relies on a large amount of data preprocessing, requires multi-stage projection, partitioning and constraint expression calculation, and the linkage capability of targeted emergency measures is weak. Summary of the Invention

[0004] To this end, the present invention provides a scene automatic reconstruction method to overcome the problems in the existing technology of high computing resource consumption and difficulty in adapting to the urgent needs of medical scenarios.

[0005] To achieve the above object, the present invention provides a scene automatic reconstruction method, comprising:

[0006] Acquiring visual images at a standard acquisition frequency, and performing blur detection and image quality assessment on the visual images to determine the types of the visual images, including clear images and blurred images;

[0007] After denoising the blurred image, judging the blurriness of data within the target damage contour of the blurred image based on a criticality evaluation index to determine whether to simulate and predict the cause of the damage;

[0008] The criticality evaluation indicators include region determination, correlation evaluation and area ratio determination;

[0009] Identifying the associated damage area corresponding to the initial damage contour by region determination, updating the initial damage contour mark to an intermediate damage contour mark, and determining whether to update the intermediate damage contour mark based on the correlation degree evaluation of the intermediate damage contour;

[0010] The correlation degree assessment includes the judgment of the damage priority level and the judgment of the number of associated damage areas;

[0011] When the real-time damage priority number is greater than the standard and the number of intermediate damage contours is not unique, the intermediate damage contour with the highest radiometry is obtained based on the radiometry calculation standard, and the intermediate damage contour mark is updated as the target damage contour mark;

[0012] When the real-time fuzzy area ratio is greater than the fuzzy ratio threshold, the data within the target injury contour is determined to be fuzzy. The current emergency classification level is determined based on the emergency classification judgment standard, corresponding measures are taken, and the cause of the injury is simulated and predicted.

[0013] Furthermore, blur detection and image quality assessment of visual images include:

[0014] Blur detection is an image clarity assessment method based on the Laplace operator to detect blur in visual images;

[0015] Image quality assessment is to obtain the Laplace variance of the visual image in the blur detection and compare it with the clarity threshold, where:

[0016] When the Laplace variance is greater than or equal to the clear threshold, the frame image is marked as a clear image and stored in the set of clear images to be screened. The current emergency classification level is confirmed and corresponding measures are taken.

[0017] When the Laplace variance is lower than the clarity threshold, the frame image is marked as a blurred image and stored in the set of blurred images to be screened. The image category and noise level in the set of blurred images to be screened are obtained to determine whether the category of the patient's current blurred image is determined based on the blurred image criticality evaluation index.

[0018] Furthermore, denoising the blurred image includes:

[0019] Calculating the real-time noise level of the blurred image and comparing it with the corresponding standard noise level;

[0020] When the real-time noise level is greater than the standard noise level, the image noise is judged to be too large and the corresponding key frame is deleted;

[0021] When the real-time noise level is less than the standard noise level, the image noise is determined to be normal, and the category to which the patient's current blurred image belongs is determined based on the criticality evaluation index of the blurred image.

[0022] Furthermore, based on the criticality evaluation index of the blurred image, the category to which the patient's current blurred image belongs is judged, including:

[0023] The criticality evaluation indicators of fuzzy images include region determination, correlation evaluation and area ratio judgment, among which,

[0024] By region determination, the associated damage regions corresponding to the initial damage contours in the fuzzy image are obtained and the degree of association is evaluated;

[0025] The process of correlation evaluation is to determine whether to update the intermediate damage contour mark to the target damage contour based on the damage priority level and the number of associated damage areas, and to determine the area ratio of the target damage contour;

[0026] The process of area ratio judgment is to obtain two-dimensional projection image data corresponding to the three-dimensional image of the target damage outline, and to judge whether the data within the target damage outline is blurred based on the real-time blurred area ratio.

[0027] Furthermore, the region determination includes:

[0028] determining a plurality of standard injury regions according to a standard organ atlas, and aligning the blurred images using a time stamp;

[0029] Obtaining the area corresponding to each initial damage contour in the blurred image, determining the number of points corresponding to any of the areas based on a point number determination standard, and obtaining three-dimensional parameter information of each point in the area, and storing the information in a corresponding initial damage contour information set;

[0030] determining a target damage contour within the initial damage contour information set based on the spatial coordinates;

[0031] Wherein, the three-dimensional parameter information includes the spatial coordinates, color information and depth value;

[0032] Determine the standard damage area corresponding to the spatial coordinates of each point in a set of initial damage contour information, mark the corresponding standard damage area as an associated damage area, update the initial damage contour mark to an intermediate damage contour mark, and evaluate the degree of association of the intermediate damage contour to obtain a target damage contour.

[0033] Furthermore, the correlation degree assessment includes the intermediate damage profile damage priority level judgment,

[0034] Determine the real-time priority score of each intermediate damage profile based on the damage priority classification standard and compare it with the standard priority score.

[0035] If the real-time priority score is greater than the standard priority score, the number of intermediate damage contours in each associated damage area is determined to update the intermediate damage contour mark;

[0036] If the real-time priority score is less than or equal to the standard priority score, the intermediate injury profile mark is not updated, and the emergency classification level corresponding to the intermediate injury profile is confirmed based on the emergency classification level standard.

[0037] Furthermore, the correlation degree assessment also includes determining the number of intermediate damage contours within the correlated damage area.

[0038] If the number of intermediate damage contours in each associated damage area is unique, the intermediate damage contour is marked as the target damage contour, and the fuzzy area ratio of the target damage contour is determined;

[0039] If the number of intermediate damage contours in each associated damage area is not unique, the intermediate damage contour with the highest radiation value is determined based on the radiometric calculation standard and marked as the target damage contour, and the fuzzy area ratio of the target damage contour is determined.

[0040] Furthermore, the fuzzy area ratio judgment includes:

[0041] The three-dimensional parameter information of the target damage contour is converted into a two-dimensional projection image using an orthogonal projection algorithm, and the two-dimensional projection image is cut into a standard block size to obtain several image blocks. The grayscale variance in each image block is obtained and compared with the standard variance threshold.

[0042] When the grayscale variance in the image block is greater than or equal to the standard variance threshold, the image block is determined to be a blurred block, and the block blur area of ​​the blurred block is obtained;

[0043] When the grayscale variance within an image block is less than the standard variance threshold, the image block is determined to be a clear block;

[0044] When the comparison process is completed, the sum of the block blur areas is obtained to obtain the total block blur area, the total block blur area is compared with the target damage contour area to obtain the real-time blur area ratio, and the real-time blur area ratio is compared with the blur ratio threshold, wherein,

[0045] When the real-time fuzzy area ratio is less than or equal to the fuzzy ratio threshold, the current emergency classification level result is confirmed;

[0046] When the real-time fuzzy area ratio is greater than the fuzzy ratio threshold, the target injury contour data is determined to be fuzzy, the current emergency classification level result is determined, and it is determined whether to simulate and predict the cause of the injury.

[0047] Further, determining the current emergency classification level and judging whether the simulation predicts the cause of injury includes:

[0048] Determine the current emergency classification level based on the emergency classification judgment criteria, and select corresponding emergency measures based on the current emergency classification level results;

[0049] The hierarchical level results include first-level results, second-level results, and third-level results;

[0050] When the data within the target damage contour is ambiguous and a level 1 result is obtained, level 1 emergency measures are taken and the cause of the damage is predicted by simulation.

[0051] Furthermore, the simulation predicts the causes of damage including:

[0052] Predicting injury causes based on medical image data, vital sign data, and risky behavior recurrence models;

[0053] Constructing a damage feature database based on historical damage data, inputting new damage feature information, and calculating the distance between the new damage feature information and all samples in the database;

[0054] The only sample or several samples with the smallest distance are selected as the predicted injury or symptom, and emergency measures for the injury or symptom are displayed.

[0055] Compared with the existing technology, the beneficial effects of the present invention are that the Laplace operator is used for blur detection and image quality grading, and the image data is dynamically screened in combination with the noise level to ensure the reliability of the input information; a three-layer analysis system for the damage area is constructed through the criticality evaluation index, and the regional judgment accurately locates the initial damage contour and associates it with the anatomical structure; the degree of association assessment uses priority sorting and radiosity calculation to screen key damage areas to avoid multi-target interference; the area proportion judgment quantifies the degree of blur with the help of three-dimensional projection and block grayscale analysis, and achieves the purpose from image analysis to clinical decision-making based on the emergency classification and injury cause prediction model of the historical database; this method improves the response speed of injury assessment and is particularly suitable for quickly determining the priority of treatment in emergency scenarios; by integrating spatial features, medical knowledge base and machine learning algorithms, it effectively assists medical staff in formulating accurate treatment plans. BRIEF DESCRIPTION OF THE DRAWINGS

[0056] Figure 1 Flowchart of the automatic reconstruction method according to an embodiment of the present invention. DETAILED DESCRIPTION

[0057] In order to make the objects and advantages of the present invention more clearly understood, the present invention is further described below in conjunction with embodiments; it should be understood that the specific embodiments described herein are merely used to explain the present invention and are not intended to limit the present invention.

[0058] The preferred embodiments of the present invention are described below with reference to the accompanying drawings. It should be understood by those skilled in the art that these embodiments are only used to explain the technical principles of the present invention and are not intended to limit the scope of protection of the present invention.

[0059] It should be noted that, in the description of the present invention, terms such as "up", "down", "left", "right", "inside", and "outside" indicating directions or positional relationships are based on the directions or positional relationships shown in the accompanying drawings. This is only for the convenience of description and does not indicate or imply that the device or element must have a specific orientation, be constructed and operated in a specific orientation. Therefore, it cannot be understood as a limitation on the present invention.

[0060] Furthermore, it should be noted that, in the description of the present invention, unless otherwise expressly specified or limited, the terms "mounted," "connected," and "connected" should be understood in a broad sense. For example, they may refer to fixed connections, detachable connections, or integral connections; mechanical connections or electrical connections; direct connections or indirect connections through an intermediate medium; and internal communication between two components. Those skilled in the art will understand the specific meanings of the above terms in the present invention based on specific circumstances.

[0061] See also Figure 1 As shown, it is a flowchart of a multi-dimensional abnormal data detection method for rapid response according to an embodiment of the present invention. The present invention provides a scene automatic reconstruction method, comprising:

[0062] Acquiring visual images at a standard acquisition frequency, and performing blur detection and image quality assessment on the visual images to determine the types of the visual images, including clear images and blurred images;

[0063] After denoising the blurred image, judging the blurriness of data within the target damage contour of the blurred image based on a criticality evaluation index to determine whether to simulate and predict the cause of the damage;

[0064] The criticality evaluation indicators include region determination, correlation evaluation and area ratio determination;

[0065] Identifying the associated damage area corresponding to the initial damage contour by region determination, updating the initial damage contour mark to an intermediate damage contour mark, and determining whether to update the intermediate damage contour mark based on the correlation degree evaluation of the intermediate damage contour;

[0066] The correlation degree assessment includes the judgment of the damage priority level and the judgment of the number of associated damage areas;

[0067] When the real-time damage priority number is greater than the standard and the number of intermediate damage contours is not unique, the intermediate damage contour with the highest radiometry is obtained based on the radiometry calculation standard, and the intermediate damage contour mark is updated as the target damage contour mark;

[0068] When the real-time fuzzy area ratio is greater than the fuzzy ratio threshold, the data within the target injury contour is determined to be fuzzy, the current emergency classification level is determined based on the emergency classification judgment standard, corresponding measures are taken, and the cause of the injury is simulated and predicted;

[0069] In this embodiment, the standard acquisition frequency is twenty frames per second;

[0070] The Laplace operator is used for blur detection and image quality grading, and the image data is dynamically screened in combination with the noise level to ensure the reliability of the input information; a three-layer analysis system for the damage area is constructed through the criticality evaluation index, and the regional judgment accurately locates the initial damage contour and associates it with the anatomical structure; the degree of association assessment uses priority sorting and radiosity calculation to screen key damage areas to avoid multi-target interference; the area ratio judgment quantifies the degree of blur with the help of three-dimensional projection and block grayscale analysis, and achieves the goal of transitioning from image analysis to clinical decision-making based on the injury cause prediction model based on the emergency classification and historical database; this method improves the response speed of injury assessment and is particularly suitable for quickly determining the treatment priority in emergency scenarios; by integrating spatial features, medical knowledge base and machine learning algorithms, it effectively assists medical staff in formulating precise treatment plans.

[0071] Specifically, blur detection and image quality assessment of visual images include:

[0072] Blur detection is an image clarity assessment method based on the Laplace operator to detect blur in visual images;

[0073] Image quality assessment is to obtain the Laplace variance of the visual image in the blur detection and compare it with the clarity threshold, where:

[0074] When the Laplace variance is greater than or equal to the clear threshold, the frame image is marked as a clear image and stored in the set of clear images to be screened. The current emergency classification level is confirmed and corresponding measures are taken.

[0075] When the Laplace variance is lower than the clarity threshold, the frame image is marked as a blurred image and stored in a set of blurred images to be screened. The image category and noise level in the set of blurred images to be screened are obtained to determine whether the category of the patient's current blurred image is determined based on the blurred image criticality evaluation index;

[0076] In this embodiment, the grayscale value of each frame of the visual image is obtained, the pixel response value after Laplace filtering is calculated, the variance of the pixel response is obtained and compared with the clarity threshold. The specific steps are as follows:

[0077] Among them, the clarity threshold is determined to be 200 based on historical data;

[0078] L(x,y)=∣I(x+1,y)+I(x-1,y)+I(x,y+1)+I(x,y-1)-4I(x,y)∣

[0079]

[0080] in,

[0081] L(x,y): pixel response value after Laplace filtering, used to measure local contrast;

[0082] |I(x+1,y)+I(x-1,y)+I(x,y+1)+I(x,y-1): weighted sum of the center pixel I(x,y) and its four adjacent pixels above, below, left, and right, reflecting the edge strength;

[0083] I(x,y): pixel value of the image at coordinate (x,y);

[0084] Var: variance of the Laplace response value, a statistical indicator for measuring the clarity of the entire image;

[0085] L i : The value of the i-th pixel in the image after Laplace filtering;

[0086] μ L : The mean of the Laplace response values ​​of all pixels in the whole image;

[0087] N: total number of pixels in the image;

[0088] The larger the L(x,y) value, the sharper the edge of the area, that is, the clearer the image;

[0089] When the Laplace variance is greater than or equal to the clear threshold, mark the frame image as a clear image, synchronize the marking result to the log file, and store it in the clear image set;

[0090] After all image frames are processed, the blurred images in the image set to be detected are denoised;

[0091] This step uses an image clarity assessment method based on Laplace variance to achieve rapid quantitative assessment of medical image quality, accurately reflect image edge sharpness to determine whether the image is clear, and enhance the system's intelligence level in blur recognition and processing. The algorithm is simple in calculation and fast in response, and can quickly meet the real-time evaluation needs in high-frequency image acquisition scenarios.

[0092] Specifically, denoising the blurred image includes:

[0093] Calculating the real-time noise level of the blurred image and comparing it with the corresponding standard noise level;

[0094] When the real-time noise level is greater than the standard noise level, the image noise is judged to be too large and the corresponding key frame is deleted;

[0095] When the real-time noise level is less than the standard noise level, the image noise is determined to be normal, and the category to which the patient's current blurred image belongs is determined based on the criticality evaluation index of the blurred image;

[0096] In this embodiment, a parameter-free noise estimation algorithm based on principal component analysis is used to perform subspace decomposition on the flat area of ​​the image, from which the noise standard deviation σ, i.e., the real-time noise level, is extracted;

[0097] Set the standard noise level σ th =0.03;

[0098] This step uses efficient means to eliminate image frames of substandard quality, effectively reducing system processing redundancy, improving the reliability and availability of emergency image data, and ensuring the accuracy and speed of subsequent image analysis processes.

[0099] Specifically, the criticality evaluation indicators of fuzzy images include region determination, correlation evaluation and area ratio determination, among which,

[0100] By region determination, the associated damage regions corresponding to the initial damage contours in the fuzzy image are obtained and the degree of association is evaluated;

[0101] The process of correlation evaluation is to determine whether to update the intermediate damage contour mark to the target damage contour based on the damage priority level and the number of associated damage areas, and to determine the area ratio of the target damage contour;

[0102] The process of area ratio judgment is to obtain two-dimensional projection image data corresponding to the three-dimensional image of the target damage outline, and to judge whether the data within the target damage outline is blurred based on the real-time blurred area ratio.

[0103] Specifically, regional determination includes:

[0104] determining a plurality of standard injury regions according to a standard organ atlas, and aligning the blurred images using a time stamp;

[0105] Obtaining the area corresponding to each initial damage contour in the blurred image, determining the number of points corresponding to any of the areas based on a point number determination standard, and obtaining three-dimensional parameter information of each point in the area, and storing the information in a corresponding initial damage contour information set;

[0106] determining a target damage contour within the initial damage contour information set based on the spatial coordinates;

[0107] Wherein, the three-dimensional parameter information includes the spatial coordinates, color information and depth value;

[0108] Determining a standard damage region corresponding to the spatial coordinates of each point in a plurality of initial damage contour information sets, marking the corresponding standard damage region as an associated damage region, updating the initial damage contour mark to an intermediate damage contour mark, and evaluating the degree of association of the intermediate damage contours to obtain a target damage contour;

[0109] In this embodiment, the number of points corresponding to the areas corresponding to different initial damage contours is determined by the proportional factor K, and K=5;

[0110] Number of points = area corresponding to the initial damage contour * scale factor;

[0111] Traverse the three-dimensional coordinates of all initial injury contour points and match them with the spatial range of the standard organ atlas. If the coordinates of a point are within the heart coordinate range, mark the area as a heart-related injury area.

[0112] This step achieves accurate marking of the standard injury area by matching the spatial coordinates of the standard organ atlas with the initial injury contour points in the image, effectively improving the spatial perception of the injury location and providing a clear spatial reference system for subsequent injury identification, grading, and intervention.

[0113] Specifically, the assessment of the degree of association includes the judgment of the damage priority level of the intermediate damage profile,

[0114] Determine the real-time priority score of each intermediate damage profile based on the damage priority classification standard and compare it with the standard priority score.

[0115] If the real-time priority score is greater than the standard priority score, the number of intermediate damage contours in each associated damage area is determined to update the intermediate damage contour mark;

[0116] If the real-time priority score is less than or equal to the standard priority score, the intermediate injury profile mark is not updated, and the emergency classification level corresponding to the intermediate injury profile is confirmed based on the emergency classification level standard;

[0117] In this embodiment, the injury priority classification standard is the internationally accepted Abbreviated Injury Scale (AIS), which divides the body into nine regions, including the head, face, neck, chest, abdomen, spine, upper limbs, and lower limbs, and assigns each injury a score from 1 to 6, corresponding to the severity from mild to fatal;

[0118] According to the AIS area number and severity and historical damage data score information, the priority score of the actual damage is determined, such as:

[0119] For head or neck AIS ≥ 4, such as cerebral contusion or open skull fracture, the priority score is 80:

[0120] Chest AIS ≥ 3, such as multiple rib fractures combined with pneumothorax, the priority score is 75;

[0121] Abdominal AIS ≥ 3, such as liver or spleen parenchymal rupture, the priority score is 70;

[0122] AIS ≥ 3 for spine and limbs, such as open fractures, the priority score is 60;

[0123] Other parts or AIS ≤ 2, minor abrasions, subcutaneous bruises, priority score is 55;

[0124] The standard priority score is 60. If the real-time priority score is greater than the standard priority score, the number of intermediate damage contours in each associated damage area is determined to update the intermediate damage contour mark;

[0125] This step dynamically assesses injury severity by applying the AIS standard score to the intermediate injury contour and combining it with real-time imaging parameters. The system prioritizes the scoring results and compares them with preset thresholds to determine whether to update the intermediate contour label. This approach balances medical standards with practical scenarios, improving the accuracy of injury severity assessments. The clear and concise prioritization is suitable for scenarios with concurrent injuries in multiple locations, helping to quickly identify high-risk injury areas, optimize the allocation of diagnostic and treatment resources, and improve the priority response speed and emergency treatment efficiency.

[0126] Specifically, the correlation degree assessment also includes judging the number of intermediate damage contours in the correlation damage area.

[0127] If the number of intermediate damage contours in each associated damage area is unique, the intermediate damage contour is marked as the target damage contour, and the fuzzy area ratio of the target damage contour is determined;

[0128] If the number of intermediate damage contours in each associated damage area is not unique, the intermediate damage contour with the highest radiation value is determined based on the radiometric calculation standard and marked as the target damage contour, and the fuzzy area ratio of the target damage contour is determined.

[0129] Specifically, the fuzzy area ratio judgment includes:

[0130] The three-dimensional parameter information of the target damage contour is converted into a two-dimensional projection image using an orthogonal projection algorithm, and the two-dimensional projection image is cut into a standard block size to obtain several image blocks. The grayscale variance in each image block is obtained and compared with the standard variance threshold.

[0131] When the grayscale variance in the image block is greater than or equal to the standard variance threshold, the image block is determined to be a blurred block, and the block blur area of ​​the blurred block is obtained;

[0132] When the grayscale variance within an image block is less than the standard variance threshold, the image block is determined to be a clear block;

[0133] When the comparison process is completed, the sum of the block blur areas is obtained to obtain the total block blur area, the total block blur area is compared with the target damage contour area to obtain the real-time blur area ratio, and the real-time blur area ratio is compared with the blur ratio threshold, wherein,

[0134] When the real-time fuzzy area ratio is less than or equal to the fuzzy ratio threshold, the current emergency classification level result is confirmed;

[0135] When the real-time fuzzy area ratio is greater than the fuzzy ratio threshold, the target injury contour data is determined to be fuzzy, the current emergency classification level is determined, and it is determined whether to simulate and predict the cause of the injury;

[0136] In this embodiment, the target lesion contour information obtained from the three-dimensional scan or reconstruction is mapped to a two-dimensional plane by parallel projection to generate a 256×256 pixel grayscale image;

[0137] Divide the area into blocks of 16×16 pixels;

[0138] The standard deviation threshold is 100;

[0139] The real-time blur area ratio is the ratio of the total block blur area to the target damage contour area, and the blur ratio threshold is 0.3;

[0140] This step maps the three-dimensional damage information to the two-dimensional space through orthogonal projection, and adopts the standard image block division strategy to perform quantitative and accurate analysis of the image clarity, thereby improving the recognition robustness and fault tolerance of the invention.

[0141] Specifically, determining the current emergency classification level and judging whether the simulation predicts the cause of injury includes:

[0142] Determine the current emergency classification level based on the emergency classification judgment criteria, and select corresponding emergency measures based on the current emergency classification level results;

[0143] The hierarchical level results include first-level results, second-level results, and third-level results;

[0144] When the data within the target damage contour is fuzzy and a level 1 result is obtained, take level 1 emergency measures and simulate and predict the cause of the damage;

[0145] In this embodiment, when a first-level result is obtained, the patient's characteristics are acute or urgent, and first-level emergency measures are taken to provide immediate treatment;

[0146] When a secondary level result is obtained, the patient's characteristics are sub-urgent, and secondary level emergency measures are taken for rapid assessment and treatment;

[0147] When a level 3 result is obtained, the patient's characteristics are non-emergency, and level 3 emergency measures are taken to ensure that the patient receives treatment safely, promptly, and in an orderly manner;

[0148] This step combines differentiated emergency response strategies corresponding to different emergency levels to effectively optimize clinical resource allocation and improve on-site treatment efficiency.

[0149] Specifically, the simulation predicts the causes of damage including:

[0150] Predicting injury causes based on medical image data, vital sign data, and risky behavior recurrence models;

[0151] Constructing a damage feature database based on historical damage data, inputting new damage feature information, and calculating the distance between the new damage feature information and all samples in the database;

[0152] Selecting a single sample or several samples with the smallest distance as predicted injury symptoms, and displaying emergency measures for the injury symptoms;

[0153] In this embodiment, the triple Siamese network is selected as the dangerous behavior reproduction model, and the Euclidean distance is used for feature matching;

[0154] Obtain human posture sequences and time series from historical injury data, input the features of anchor points, positive samples, and negative samples respectively, and extract high-dimensional dangerous behavior feature vectors;

[0155] The anchor point is a normal or previously occurring dangerous behavior sample; the positive sample is a dangerous behavior of the same type as the anchor point, such as falling or violent collision; the negative sample is a behavior of a different type from the anchor point, such as walking or sitting. During training, the anchor point is kept close to the positive sample.

[0156] Read real-time injury data and analyze the feature vectors in the new data based on the Euclidean distance calculation method to predict new injury symptoms. The specific calculation formula is:

[0157]

[0158] in,

[0159] d i : The Euclidean distance between the i-th historical sample and the new sample;

[0160] F newThe composite feature vector representing the newly acquired damage has a length of D;

[0161] F i : The feature vector corresponding to the i-th historical sample in the database is D-dimensional;

[0162] D: The dimension of the feature space, that is, the total number of components contained in each vector;

[0163] || ||2: the second norm of the vector;

[0164] Take the unique sample or several samples with the smallest distance and predict the possible symptoms of the new injury based on the historical diagnosis corresponding to these samples;

[0165] This step performs predictive analysis based on similar behaviors and injury characteristics when the image is blurred or information is missing, integrating machine learning models with medical knowledge graphs to significantly improve the system's reasoning and adaptability, enabling rapid etiology prediction and emergency treatment decision support in injury scenarios.

[0166] Thus far, the technical solutions of the present invention have been described in conjunction with the preferred embodiments shown in the accompanying drawings. However, it will be readily understood by those skilled in the art that the scope of protection of the present invention is obviously not limited to these specific embodiments. Without departing from the principles of the present invention, those skilled in the art may make equivalent changes or substitutions to the relevant technical features, and the technical solutions after such changes or substitutions will fall within the scope of protection of the present invention.

[0167] The foregoing description is merely a preferred embodiment of the present invention and is not intended to limit the present invention. Those skilled in the art will readily appreciate that the present invention is susceptible to various modifications and variations. Any modifications, equivalent substitutions, or improvements made within the spirit and principles of the present invention are intended to be within the scope of protection of the present invention.

Claims

1. A method for automatic scene reconstruction, characterized in that: include, Acquiring visual images at a standard acquisition frequency, and performing blur detection and image quality assessment on the visual images to determine the types of the visual images, including clear images and blurred images; After denoising the blurred image, judging the blurriness of data within the target damage contour of the blurred image based on a criticality evaluation index to determine whether to simulate and predict the cause of the damage; The criticality evaluation indicators include region determination, correlation evaluation and area ratio determination; Identifying the associated damage area corresponding to the initial damage contour by region determination, updating the initial damage contour mark to an intermediate damage contour mark, and determining whether to update the intermediate damage contour mark based on the correlation degree evaluation of the intermediate damage contour; The correlation degree assessment includes the judgment of the damage priority level and the judgment of the number of associated damage areas; When the real-time damage priority number is greater than the standard and the number of intermediate damage contours is not unique, the intermediate damage contour with the highest radiometry is obtained based on the radiometry calculation standard, and the intermediate damage contour mark is updated as the target damage contour mark; When the real-time fuzzy area ratio is greater than the fuzzy ratio threshold, the data within the target injury contour is determined to be fuzzy. The current emergency classification level is determined based on the emergency classification judgment standard, corresponding measures are taken, and the cause of the injury is simulated and predicted.

2. The scene automatic reconstruction method according to claim 1, characterized in that: Blur detection and image quality assessment of visual images include: Blur detection is an image clarity assessment method based on the Laplace operator to detect blur in visual images; Image quality assessment is to obtain the Laplace variance of the visual image in the blur detection and compare it with the clarity threshold, where: When the Laplace variance is greater than or equal to the clear threshold, the frame image is marked as a clear image and stored in the set of clear images to be screened. The current emergency classification level is confirmed and corresponding measures are taken. When the Laplace variance is lower than the clarity threshold, the frame image is marked as a blurred image and stored in the set of blurred images to be screened. The image category and noise level in the set of blurred images to be screened are obtained to determine whether the category of the patient's current blurred image is determined based on the blurred image criticality evaluation index.

3. The scene automatic reconstruction method according to claim 2, characterized in that: Denoising the blurred image includes: Calculating a real-time noise level of the blurred image and comparing it with a corresponding standard noise level; When the real-time noise level is greater than the standard noise level, the image noise is judged to be too large and the corresponding key frame is deleted; When the real-time noise level is less than the standard noise level, the image noise is determined to be normal, and the category to which the patient's current blurred image belongs is determined based on the criticality evaluation index of the blurred image.

4. The scene automatic reconstruction method according to claim 3, characterized in that: Based on the criticality evaluation index of the fuzzy image, the category of the patient's current fuzzy image is judged to include: The criticality evaluation indicators of fuzzy images include region determination, correlation evaluation and area ratio judgment, among which, By region determination, the associated damage regions corresponding to the initial damage contours in the fuzzy image are obtained and the degree of association is evaluated; The process of correlation evaluation is to determine whether to update the intermediate damage contour mark to the target damage contour based on the damage priority level and the number of associated damage areas, and to determine the area ratio of the target damage contour; The process of area ratio judgment is to obtain two-dimensional projection image data corresponding to the three-dimensional image of the target damage outline, and to judge whether the data within the target damage outline is blurred based on the real-time blurred area ratio.

5. The scene automatic reconstruction method according to claim 4, characterized in that: Regional determination includes: determining a plurality of standard injury regions according to a standard organ atlas, and aligning the blurred images using a time stamp; Obtaining the area corresponding to each initial damage contour in the blurred image, determining the number of points corresponding to any of the areas based on a point number determination standard, and obtaining three-dimensional parameter information of each point in the area, and storing the information in a corresponding initial damage contour information set; determining a target damage contour within the initial damage contour information set based on the spatial coordinates; Wherein, the three-dimensional parameter information includes the spatial coordinates, color information and depth value; Determine the standard damage area corresponding to the spatial coordinates of each point in a set of initial damage contour information, mark the corresponding standard damage area as an associated damage area, update the initial damage contour mark to an intermediate damage contour mark, and evaluate the degree of association of the intermediate damage contour to obtain a target damage contour.

6. The scene automatic reconstruction method according to claim 5, characterized in that: The correlation degree assessment includes the intermediate damage profile damage priority level judgment, Determine the real-time priority score of each intermediate damage profile based on the damage priority classification standard and compare it with the standard priority score. If the real-time priority score is greater than the standard priority score, the number of intermediate damage contours in each associated damage area is determined to update the intermediate damage contour mark; If the real-time priority score is less than or equal to the standard priority score, the intermediate injury profile mark is not updated, and the emergency classification level corresponding to the intermediate injury profile is confirmed based on the emergency classification level standard.

7. The scene automatic reconstruction method according to claim 5, characterized in that: The correlation degree assessment also includes judging the number of intermediate damage contours in the correlation damage area. If the number of intermediate damage contours in each associated damage area is unique, the intermediate damage contour is marked as the target damage contour, and the fuzzy area ratio of the target damage contour is determined; If the number of intermediate damage contours in each associated damage area is not unique, the intermediate damage contour with the highest radiation value is determined based on the radiometric calculation standard and marked as the target damage contour, and the fuzzy area ratio of the target damage contour is determined.

8. The scene automatic reconstruction method according to claim 7, characterized in that: Fuzzy area ratio judgment includes: The three-dimensional parameter information of the target damage contour is converted into a two-dimensional projection image using an orthogonal projection algorithm, and the two-dimensional projection image is cut into a standard block size to obtain several image blocks. The grayscale variance in each image block is obtained and compared with the standard variance threshold. When the grayscale variance in the image block is greater than or equal to the standard variance threshold, the image block is determined to be a blurred block, and the block blur area of ​​the blurred block is obtained; When the grayscale variance within an image block is less than the standard variance threshold, the image block is determined to be a clear block; When the comparison process is completed, the sum of the block blur areas is obtained to obtain the total block blur area, the total block blur area is compared with the target damage contour area to obtain the real-time blur area ratio, and the real-time blur area ratio is compared with the blur ratio threshold, wherein, When the real-time fuzzy area ratio is less than or equal to the fuzzy ratio threshold, the current emergency classification level result is confirmed; When the real-time fuzzy area ratio is greater than the fuzzy ratio threshold, the target injury contour data is determined to be fuzzy, the current emergency classification level result is determined, and it is determined whether to simulate and predict the cause of the injury.

9. The scene automatic reconstruction method according to claim 8, characterized in that: Determine the current emergency classification level and judge whether the simulation predicts the cause of injury, including: Determine the current emergency classification level based on the emergency classification judgment criteria, and select corresponding emergency measures based on the current emergency classification level results; The hierarchical level results include first-level results, second-level results, and third-level results; When the data within the target damage contour is ambiguous and a level 1 result is obtained, level 1 emergency measures are taken and the cause of the damage is predicted by simulation.

10. The scene automatic reconstruction method according to claim 9, characterized in that: The simulation predicts the causes of damage including: Predicting injury causes based on medical image data, vital sign data, and risky behavior recurrence models; Constructing a damage feature database based on historical damage data, inputting new damage feature information, and calculating the distance between the new damage feature information and all samples in the database; The only sample or several samples with the smallest distance are selected as the predicted injury or symptom, and emergency measures for the injury or symptom are displayed.

Images