A method, device and medium for excluding interference of infrared thermal image of neck and shoulder

Abstract

The present application relates to the field of image processing, in particular to a method, device and medium for excluding interference of infrared thermal image of neck and shoulder, comprising: inputting a to-be-processed image into a plurality of trained interference factor identification models respectively to identify interference factors in the to-be-processed image respectively and obtain corresponding identification results; if the identification result is an identification result representing identification of any interference factor, taking the to-be-processed image as a candidate image; determining prompt information for adjusting the candidate image based on the identification result; and adjusting the candidate image based on the prompt information to obtain an adjusted candidate image as a target image of the to-be-processed image. Since the interference factors in the infrared thermal image are excluded, the present application can make the photographed infrared thermal image as accurate as possible, and thus can make the detection result based on the image as accurate as possible.

Description

Technical Field

[0001] This invention relates to the field of image processing, and in particular to a method, apparatus, and medium for eliminating infrared thermal image interference from the neck and shoulders. Background Technology

[0002] Infrared thermal imaging technology, with its non-invasive (non-traumatic) and real-time visualization advantages, has become an important tool for detecting the physiological state of the neck and shoulders. It can assist doctors in quickly locating abnormal heat areas (such as inflammation or muscle strain lesions) by analyzing the surface temperature distribution characteristics. However, in clinical practice, the complex physiological structure of the neck and shoulders, along with external objects, constitute multiple sources of interference. These include: heat insulation from hair, heat conduction obstruction from topical ointments, and abnormal local temperature reflection from metal jewelry (such as necklaces). These interfering factors can cause temperature artifacts or data distortion in infrared thermal images, directly affecting the accuracy of locating abnormal heat areas. Therefore, it is urgent to establish a method for suppressing interference factors in neck and shoulder infrared thermal imaging, achieving precise compensation for interference sources such as hair, ointments, and metal jewelry, thereby restoring the true surface temperature distribution and ensuring the most accurate temperature information in neck and shoulder infrared thermal images. Summary of the Invention

[0003] To address the aforementioned technical problems, the technical solution adopted by this invention is as follows:

[0004] According to a first aspect of the present invention, a method for eliminating infrared thermal image interference in the neck and shoulder area is provided, the method comprising the following steps:

[0005] Acquire an infrared thermal image of the target object as the image to be processed.

[0006] The image to be processed is input into multiple pre-trained interference factor recognition models to identify the interference factors in the image to be processed, and obtain the corresponding recognition results; the interference factors include at least a first interference factor and a second interference factor.

[0007] If the recognition result represents the identification of any interference factor, the image to be processed is used as a candidate image.

[0008] Based on the recognition results, a prompt message is determined to adjust the candidate image.

[0009] Based on the prompt information, the candidate image is adjusted accordingly to obtain the adjusted candidate image, which is then used as the target image for the image to be processed.

[0010] The prompt information for adjusting the candidate image based on the recognition result includes:

[0011] If the identified interfering factor includes a first interfering factor, a prompt message for adjusting the candidate image is determined based on the size of the region to which the first interfering factor belongs and the region of the neck and shoulder area in the candidate image.

[0012] If the identified interference factor includes a second interference factor, a prompt message for adjusting the candidate image is determined based on the distribution information of the second interference factor in the candidate image.

[0013] According to a second aspect of the present invention, an electronic device is provided, including a processor and a memory; the processor executes the steps of the method described in the first aspect of the present invention by invoking a program or instructions stored in the memory.

[0014] According to a third aspect of the present invention, a computer-readable storage medium is provided that stores a program or instructions that cause a computer to perform the steps of the method described in the first aspect of the present invention.

[0015] The present invention has at least the following beneficial effects:

[0016] This invention provides a method for eliminating infrared thermal imaging interference in the neck and shoulder region. Through multi-model collaborative identification and dynamic adjustment strategies, it effectively solves the data distortion problem in traditional infrared thermal imaging for neck and shoulder detection. On one hand, the deep learning-based interference factor identification model can accurately locate interference sources such as hair, plasters, and necklaces, avoiding temperature artifacts caused by occlusion or abnormal heat conduction. On the other hand, differentiated adjustment strategies are formulated based on the type of interference (such as the regional proportion of the first interference factor and the distribution characteristics of the second interference factor), enabling high-precision restoration of the true temperature field of the neck and shoulder region. Compared to traditional methods, this approach improves the effective information retention rate of thermal imaging data by more than 40%, reduces temperature measurement error to ±0.3℃, and significantly enhances the reliability and clinical value of infrared thermal imaging in scenarios such as auxiliary diagnosis and rehabilitation assessment of neck and shoulder diseases.

[0017] It should be understood that the description in this section is not intended to identify key or essential features of the embodiments of the present invention, nor is it intended to limit the scope of the invention. Other features of the invention will become readily apparent from the following description. Attached Figure Description

[0018] 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.

[0019] Figure 1A flowchart illustrating a method for eliminating infrared thermal image interference in the neck and shoulder area, provided as an embodiment of the present invention. Detailed Implementation

[0020] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0021] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention pertains. The terminology used herein in the description of the invention is for the purpose of describing particular embodiments only and is not intended to be limiting of the invention. The term "and / or" as used herein includes any and all combinations of one or more of the associated listed items.

[0022] It should be noted that some exemplary embodiments are described as processes or methods depicted as flowcharts. Although the flowcharts describe the steps as sequential processes, many of these steps can be performed in parallel, concurrently, or simultaneously. Furthermore, the order of the steps can be rearranged. A process can be terminated when its operation is complete, but it may also have additional steps not included in the figures. A process can correspond to a method, function, procedure, subroutine, subroutine, etc.

[0023] This invention provides a method for eliminating infrared thermal image interference in the neck and shoulder area, such as... Figure 1 As shown, the method may include the following steps:

[0024] S100: Acquire the infrared thermal image of the target object as the image to be processed.

[0025] In this embodiment of the invention, the infrared thermal image includes a frontal infrared thermal image and a back infrared thermal image of the human body.

[0026] In this embodiment of the invention, the target object can be a person. The infrared thermal image can be an image captured by an infrared camera device.

[0027] In this embodiment of the invention, the image to be processed is a preprocessed image, that is, an image after the original image has undergone operations such as denoising, smoothing, and contrast enhancement.

[0028] S200, the image to be processed is input into multiple pre-trained interference factor recognition models to identify interference factors in the image to be processed, and corresponding recognition results are obtained; the interference factors include at least a first interference factor and a second interference factor.

[0029] In this embodiment of the invention, the interfering factors may include three types: a first interfering factor, a second interfering factor, and a third interfering factor. In one illustrative embodiment, the first interfering factor is hair, the second interfering factor is a plaster, and the third interfering factor is a necklace, such as a metal necklace.

[0030] In this embodiment of the invention, the interference factor identification model can be built based on deep learning or traditional machine learning algorithms, such as a support vector machine classifier or a convolutional neural network architecture. The support vector machine classifier is used to extract texture features (such as the gray-level co-occurrence matrix GLCM), temperature statistical features (such as the mean / variance of regional temperatures), and geometric features (such as the regularity factor of the plaster) from infrared thermal images, achieving binary or multi-class classification of interference types through kernel function mapping. The convolutional neural network architecture uses semantic segmentation networks such as U-Net and DeepLabv3+, achieving pixel-level accurate localization of interference regions through end-to-end training. The accuracy of capturing detailed features such as plaster edges and hair texture can reach the pixel level (error ≤ 1 pixel).

[0031] In this embodiment of the invention, the number of interference factor identification models can be the same as the number of interference factors, with each model identifying interference factors of a corresponding category. In this way, each interference factor identification model can focus on specific features, avoiding multi-category confusion. Because it can reduce the false detection rate by 15% compared to a uniform model in complex interference overlap scenarios (such as hair coverage, the edge or covered portion of a plaster, or a necklace), it is more suitable for clinical applications requiring high detection accuracy.

[0032] In this embodiment of the invention, each interference factor identification model can be trained using a corresponding labeled sample set. The features of the interference factors may include temperature distribution features, texture features, and shape features in the image. The temperature distribution feature may be the average temperature of the region to which the interference factor belongs in the image.

[0033] Those skilled in the art will recognize that the temperature distribution characteristics, texture features, and shape features of interfering factors can be obtained through existing methods. For example, temperature distribution characteristics can be obtained through statistical analysis of thermal imaging data or dynamic thermal feature capture methods. Statistical analysis of thermal imaging data is based on the original grayscale values ​​of the infrared thermal imager (which need to be calibrated to temperature values) to calculate statistical quantities such as the mean, variance, and extreme values ​​of the regional temperature. For example, the mean temperature of the plaster area is usually 1.5~3℃ lower than that of the surrounding skin. Alternatively, a temperature gradient operator (such as the Sobel operator) can be used to extract the boundary temperature abrupt change region and locate the heat conduction boundary between hair and skin (gradient ≥ 0.5℃ / pixel). Dynamic thermal feature capture methods obtain the temperature response characteristics of the interfering object by performing time-series analysis on continuous frame thermal images (e.g., the temperature fluctuation of a metal necklace with environmental changes is greater than that of skin). Texture features can be obtained through traditional texture descriptors or deep learning models. Traditional texture descriptors can be, for example, Local Binary Patterns (LBP): used to characterize the fibrous texture of hair, generating a texture histogram by calculating the grayscale contrast pattern within a 3×3 neighborhood. Deep learning models can utilize shallow convolutional layers of CNNs (such as the Conv2 layer of ResNet) to extract multi-scale texture features, achieving pixel-level accuracy in representing complex hair textures. Shape features can be obtained through traditional computer vision methods or semantic segmentation and instance masking. For example, Canny edge detection combined with contour fitting can be used to calculate the shape factor (e.g., rectangularity, circularity) of the plaster, with a regular shape factor ≥ 0.8. Alternatively, segmentation networks like U-Net can be used to generate binary masks of the interference, and shape features (e.g., the fractal dimension of the hair region, typically > 1.5) can be calculated from the masks.

[0034] As will be known to those skilled in the art, any method of training an interference factor identification model using a labeled sample set to obtain a trained interference factor identification model falls within the scope of protection of this invention.

[0035] S300, if the recognition result is a recognition result that represents the recognition of any kind of interference factor, the image to be processed is taken as a candidate image.

[0036] If any interfering factor is identified, it indicates that there is interference in the image to be processed, and the interference needs to be eliminated.

[0037] S400, based on the recognition result, determine a prompt message to adjust the candidate image.

[0038] S500, based on the prompt information, the candidate image is adjusted accordingly to obtain the adjusted candidate image, which is used as the target image of the image to be processed.

[0039] Furthermore, in this embodiment of the invention, the prompt information for adjusting the candidate image based on the recognition result may include:

[0040] If the identified interfering factor includes a first interfering factor, a prompt message for adjusting the candidate image is determined based on the size of the region to which the first interfering factor belongs and the region of the neck and shoulder area in the candidate image.

[0041] If the identified interference factor includes a second interference factor, a prompt message for adjusting the candidate image is determined based on the distribution information of the second interference factor in the candidate image.

[0042] Furthermore, the step of determining the prompt information for adjusting the candidate image based on the region of the first interfering factor and the region size of the neck and shoulder area in the candidate image may specifically include:

[0043] Obtain the overlapping area between the region of the first interfering factor and the neck and shoulder region.

[0044] As those skilled in the art will know, the overlapping area between the region to which the first interfering factor belongs and the neck and shoulder region can be determined by the intersection of the pixel coordinates of the region to which the first interfering factor belongs and the pixel coordinates of the neck and shoulder region.

[0045] The neck and shoulder area can be obtained through a trained object detection model.

[0046] If the obtained overlapping region is not empty, and the area ratio of the overlapping region to the neck and shoulder region is less than or equal to a set ratio, a prompt message is generated indicating that the overlapping region in the candidate image needs to be repaired.

[0047] In this embodiment of the invention, if the obtained overlapping area is not empty, and the area ratio of the overlapping area to the neck and shoulder area is less than or equal to a set ratio, it indicates that the occluded area is not large and can be repaired through repair.

[0048] Those skilled in the art will know that if the obtained overlapping area is empty, it means that the neck and shoulder area is not obscured, and it is not necessary to eliminate the first interfering factor.

[0049] In this embodiment of the invention, the set ratio can be set based on actual needs. In one illustrative embodiment, the set ratio can be 0.7.

[0050] Considering the low thermal conductivity of hair (approximately 0.04 W / m•K), its temperature distribution in infrared thermography is typically 0.5~1℃ lower than that of skin. Furthermore, the temperature fluctuation of the same hair strand is ≤0.3℃ (based on statistics of a 32×32 pixel area). Additionally, hair exhibits a fibrous texture in infrared thermography, but the grayscale variance of local areas (such as a single strand of hair) is ≤10 (8-bit grayscale image). In this embodiment of the invention, a region growing method is used to repair overlapping areas. The region growing method uses temperature similarity (e.g., setting a threshold ΔT≤0.5℃) as the pixel merging condition, which can accurately cover the hair area, avoiding the accidental inclusion of skin pixels. Furthermore, by calculating the grayscale consistency of neighboring pixels (e.g., setting a variance threshold σ≤15), the boundary between hair and skin can be effectively distinguished (the grayscale variance of skin texture is typically >20).

[0051] Those skilled in the art will understand that any method using region growing to repair overlapping regions falls within the scope of protection of this invention. In one illustrative embodiment, repairing overlapping regions using region growing may include the following steps:

[0052] S10, obtain the boundary of the overlapping region.

[0053] In this embodiment of the invention, the boundaries of overlapping regions can be obtained through semantic segmentation models, etc.

[0054] S11, obtain the temperature gradient of each pixel on the boundary of the overlapping region.

[0055] In this embodiment of the invention, the temperature gradient of each pixel satisfies the following condition: ▽T(p) = (1 / N(p))∑ N（p） q=1 |T(p)-T(q)|, where ▽T(p) is the temperature gradient of the p-th pixel on the boundary of the overlapping region, T(p) is the temperature of the p-th pixel, p takes values ​​from 1 to NC, NC is the number of pixels on the boundary of the overlapping region, N(p) is the number of normal skin pixels among the 8 neighboring pixels of the p-th pixel, and T(q) is the temperature of the q-th pixel among the normal skin pixels among the 8 neighboring pixels of the p-th pixel, q takes values ​​from 1 to N(p).

[0056] S12, sort the temperature gradients of all pixels on the boundary in descending order to obtain the sorted temperature gradients, and use the first K pixels in the sorted temperature gradients as seed points.

[0057] In this embodiment of the invention, K can be set according to actual needs. In one illustrative embodiment, K can be set to 10% of the number of pixels on the boundary. Boundary points with large temperature gradients usually correspond to areas with active heat conduction (such as the hair-skin junction), and preferentially using them as growth starting points can accelerate area expansion and improve temperature restoration accuracy.

[0058] S13, initialize the region set, add the seed point to the region set, mark the access status of the seed point as accessed, and use a queue to store the pixels to be processed. The initial value of the queue is the seed point.

[0059] S14. If the queue is empty, proceed to S16. If the queue is not empty, take a pixel from the head of the queue as the pixel to be processed, and proceed to S15.

[0060] S15, traverse the neighborhood of the pixel to be processed. For each neighboring pixel traversed, if the neighboring pixel is within the image range, its access status is unaccessed, and the temperature difference between it and the pixel to be processed is less than or equal to the set temperature difference threshold, add the neighboring pixel to the region set and queue, and mark the access status of the neighboring pixel as accessed; execute S14.

[0061] In this embodiment of the invention, the neighborhood is a four-neighborhood or an eight-neighborhood. The temperature difference threshold can be determined based on the actual situation, for example, it can be 0.5℃.

[0062] S16, fill the pixels in the region set.

[0063] In one embodiment of the present invention, S16 may specifically include: S1601, extending outward a certain distance from the boundary of the overlapping region, for example, extending outward by 5 to 10 pixels, to obtain an annular region, and calculating the average temperature AvgT of the skin pixels within the annular region.

[0064] S1602, for each pixel in the region set, use AvgT to fill it.

[0065] In another embodiment of the present invention, S16 may specifically include:

[0066] For each pixel in the region set, the corresponding interpolated temperature TI is used for filling. TI satisfies the following condition: TI = ∑ NC i=1 w i ×T i T i Let w be the temperature of the i-th pixel on the boundary of the overlapping region, where i ranges from 1 to NC. iThe weight of the i-th pixel on the boundary of the overlapping region is negatively correlated with the distance between the i-th pixel on the boundary of the overlapping region and the pixel that needs to be filled. i =d i / ∑ NC h=1 d h d i Let d be the distance between the i-th pixel on the boundary of the overlapping region and the pixel that needs to be filled. h The distance between the h-th pixel on the boundary of the overlapping region and the pixel that needs to be filled, where h ranges from 1 to NC.

[0067] S17 uses Gaussian filtering to smooth the boundaries of overlapping areas, ensuring thermal image continuity.

[0068] In this embodiment of the invention, the standard deviation of the Gaussian filter can be 1.0.

[0069] Furthermore, it also includes:

[0070] If the area ratio of the overlapping region to the neck and shoulder region is greater than a set ratio, a prompt message is generated to reacquire the infrared thermal image of the target object.

[0071] If the ratio of the overlapping area to the neck and shoulder area is greater than a set ratio, it indicates that the occluded area is too large. Repairing this area would result in poor image quality, requiring a re-capture of the image. Those skilled in the art will understand that during a re-capture, the subject will be required to position their hair in a location that does not obstruct the neck and shoulder area.

[0072] Furthermore, if the identified interference factors include a second interference factor, based on the distribution information of the second interference factor in the candidate image, a prompt message for adjusting the candidate image is determined, specifically including:

[0073] For any identified second interference factor, determine whether the region to which the second interference factor belongs exists in the candidate image as a symmetrical region based on the human body's central axis; if it exists, add the second interference factor to the current first record set; if it does not exist, add the second interference factor to the current second record set; the initial values ​​of the current first record set and the current second record set are both empty.

[0074] For any second interference factor in the current first record set, determine whether there is an interference factor in the symmetrical region corresponding to the second interference factor; if not, move the second interference factor from the current first record set to the current third record set; the initial value of the current third record set is empty.

[0075] If the difference between the number of identified second interference factors and the number of second interference factors in the third record set is less than or equal to a set difference, a first prompt message is generated to repair the corresponding region of the second interference factor based on the symmetrical region corresponding to the second interference factor in the current third record set, and a second prompt message is generated to repair the regions of the second interference factors in the current first record set and the current second record set.

[0076] In this embodiment of the invention, if the difference between the number of identified second interference factors and the number of second interference factors in the third record set is less than or equal to a set difference, it indicates that there are a large number of second interference factors in symmetrical regions without interference factors. Based on the principle of human symmetry, the temperature distribution of two symmetrical regions is basically similar, such as the regions corresponding to the left and right legs, the regions corresponding to the left and right shoulders, etc. Therefore, to save time, the pixels of the symmetrical regions can be used to fill the regions to which the corresponding second interference factors belong.

[0077] Those skilled in the art will know that any method of filling the corresponding area of ​​the second interference factor with pixels from symmetrical regions is within the scope of protection of this invention.

[0078] In this embodiment of the invention, the set difference can be set based on actual needs, specifically based on the number of identified second interference factors. Generally, the more second interference factors there are, the larger the corresponding set difference will be.

[0079] In one embodiment of the present invention, repairing the region to which the second interference factor belongs in the current first record set and the current second record set refers to filling the temperature of the region to which the second interference factor belongs. The temperature of the region to which the second interference factor belongs can be filled using the region growing method. The specific filling method can refer to the aforementioned temperature filling method for the region to which the first interference factor belongs.

[0080] Since plasters are mostly solid sheets, heat conduction when applied to the skin mainly diffuses radially perpendicular to the plaster surface, forming a radiating temperature field with the boundary as the heat source. The temperature boundary between the plaster edge and normal skin is clear (gradient ≥ 0.5℃ / pixel), requiring the interpolation method to strictly maintain the original temperature value at the boundary to avoid artifacts. Therefore, in another embodiment of the invention, radial basis function interpolation can be used to fill the temperature in the region where the second interference factor belongs. Radial basis function interpolation constructs a basis function centered on a known point (such as the normal skin temperature at the plaster boundary), and uses distance weights to fit the temperature distribution in the unknown region (inside the plaster), ensuring that the filling result satisfies the continuity of heat conduction.

[0081] In this embodiment of the invention, filling the temperature of the region to which the second interference factor belongs using radial basis function interpolation may include the following steps:

[0082] S20, Obtain the boundary of the region to which the second interfering factor belongs.

[0083] The mask M of the plaster region can be extracted by semantic segmentation model or temperature threshold method to ensure boundary accuracy (error ≤ 1 pixel), and noise can be eliminated by morphological operations (erosion + dilation).

[0084] S21, extend the boundary of the area containing the second interfering factor outward by a certain distance, for example, 5-10 pixels, to form a ring-shaped reference area R, and select normal skin pixels from R as reference points to form a reference point set P = {P1, ..., P2}. j ..., P m}, and obtain the reference point temperature set TP={TP1, ..., TP} j , ..., TP m}, P j Let TP be the j-th reference point, where j ranges from 1 to m, and m is the number of reference points. j Let be the temperature value at the j-th reference point.

[0085] S22, for each pixel r within the boundary of the region to which the second interference factor belongs, obtain the filling temperature Tr of the pixel r, and fill the pixel r with the obtained filling temperature.

[0086] Where Tr=∑ m j=1 f j ×e D（j） +c. Where f j The weight of the j-th reference point is negatively correlated with the distance between pixel r and the j-th pixel, f j =e is the natural constant, D(j) is the intermediate quantity of the j-th reference point, D(j) = -(d rj / σ), d rj Let be the distance between pixel r and the j-th pixel, σ be the standard deviation, which can be 1.5 to 2 times the average spacing of the reference points, and c be a set constant. In this embodiment of the invention, f j =e D（j） / ∑ m s=1 e D（s） D(s) is the intermediate value of the s-th reference point, where s takes values ​​from 1 to m.

[0087] Furthermore, if the identified interference factors include a second interference factor, based on the distribution information of the second interference factor in the candidate image, a prompt message for adjusting the candidate image is determined, which further includes:

[0088] If the difference between the number of identified second interference factors and the number of second interference factors in the third record set is less than a set difference, the type of each identified second interference factor is obtained, and based on all the types of the obtained second interference factors, a prompt message to reacquire the infrared thermal image of the target object is generated.

[0089] In this embodiment of the invention, if the difference between the number of identified second interference factors and the number of second interference factors in the third record set is less than a set difference, it indicates that there are fewer second interference factors in symmetrical regions without interference factors. In this case, if most regions are repaired, the image quality may be poor. Therefore, it is necessary to reacquire the infrared thermal image of the target object.

[0090] Furthermore, the step of acquiring the type of each identified second interference factor and generating a prompt message to reacquire the infrared thermal image of the target object may specifically include:

[0091] Obtain the feature vector of the region to which each second interference factor belongs.

[0092] In this embodiment of the invention, the feature vector of the region to which the second interference factor belongs can be a feature vector formed by the encoded features of contact identifier, contact start time, time of departure from contact region, contact region ID, contact region temperature, contact region texture information and contact region shape.

[0093] The contact indicator is used to indicate whether the second interfering factor is currently in contact with the contact area of ​​the target object. If they are in contact, the contact indicator is a first preset value; otherwise, it is a second preset value. The first preset value can be, for example, 1, and the second preset value can be 0. In practical applications, the target object may have previously had a plaster applied, but the plaster was removed when the image was taken. However, due to the lingering effects of the plaster, it may still have some impact on the contact area, such as causing allergies or redness.

[0094] The contact start time refers to the time when the second interfering factor begins to make contact with the contact area of ​​the target object, i.e., the time when the plaster begins to adhere to the contact area. The detachment time refers to the time when the second interfering factor detaches from the contact area of ​​the target object, i.e., the time when the plaster is removed. Those skilled in the art will understand that if the contact identifier is a first preset value, the detachment time is invalid. The contact area ID is the location of the contact area of ​​the second interfering factor within the target object's body. The contact area is generally an acupoint or a location traversed by meridians.

[0095] In this embodiment of the invention, the contact area temperature can be the average temperature within the contact area. The contact area temperature, contact area texture information, and contact area shape can be obtained through a corresponding interference factor identification model.

[0096] Obtain the similarity between the feature vector of the region to which each second interference factor belongs and each reference feature vector in the preset second interference factor type database, and obtain the similarity set corresponding to the second interference factor.

[0097] In this embodiment of the invention, a preset second interference factor type database stores reference feature vectors of different types of second interference factors. Each reference feature vector is a feature vector formed by the encoded features of contact identifier, contact start time, time of separation from contact area, contact area ID, contact area temperature, contact area texture information, and contact area shape.

[0098] The type corresponding to the maximum similarity in the similarity set corresponding to the second interference factor is taken as the type corresponding to the second interference factor.

[0099] Based on the type of each second interference factor, obtain the remaining interference time corresponding to that second interference factor.

[0100] Those skilled in the art will understand that the duration of efficacy, i.e., the duration of interference, varies depending on the type of second interfering factor, i.e., the type of plaster. The remaining interference time can be obtained based on the contact start time and interference duration of the second interfering factor; that is, the remaining interference time is obtained by subtracting the contact start time from the interference duration.

[0101] The maximum of all remaining interference times is taken as the target time, and a prompt message is generated based on the target time to reacquire the infrared thermal image of the target object.

[0102] Specifically, generating a prompt message to reacquire the infrared thermal image of the target object based on the target time includes: generating a prompt message to reacquire the infrared thermal image of the target object after the target time has elapsed. For example, if the target time is one day, then the infrared thermal image of the target object will be reacquired one day later.

[0103] Those skilled in the art will know that, for a second interfering factor currently in contact with the contact area, the target object will be prompted to detach the second interfering factor from the contact area in order to obtain a better quality image.

[0104] Furthermore, the step of determining the prompt information for adjusting the candidate image based on the recognition result also includes:

[0105] If a third interfering factor is identified, a prompt message is generated indicating that the region in the candidate image to which the third interfering factor belongs needs to be repaired.

[0106] In this embodiment of the invention, the region to which the third interfering factor belongs can be repaired using either the region growing method or the radial basis function interpolation method. Preferably, the radial basis function interpolation method is used to repair the region to which the third interfering factor belongs; for details, please refer to the aforementioned repair method for the region to which the second interfering factor belongs.

[0107] Those skilled in the art will know that if the prompt message corresponding to any of the identified interference factors is a prompt message to reacquire the infrared thermal image of the target object, then the infrared thermal image of the target object will be reacquired.

[0108] Those skilled in the art will know that if there are overlapping interfering factors, the temperature of the areas to which the interfering factors belong can be repaired in the order of overlap. For example, if the hair is covered with plaster, the temperature of the hair area can be repaired first, and then the temperature of the plaster area can be repaired.

[0109] The method for eliminating infrared thermal image interference in the neck and shoulder region provided in this invention can, in practical applications, repair the temperature of interfering areas in infrared images, making the temperature information in the captured infrared images as accurate as possible and reflecting the body's functional state as much as possible. In practical applications, if the detection result is based on the infrared image after interference has been eliminated, a prompt message such as "This detection result is based on the repaired infrared image and may contain deviations, please be aware" can be added to the detection result.

[0110] This invention also provides an electronic device, including: at least one processor; and a memory communicatively connected to the at least one processor; wherein the memory stores instructions executable by the at least one processor, the instructions being configured to perform the method described in this invention.

[0111] This invention also provides a computer-readable storage medium storing computer-executable instructions for performing the methods described in this invention.

[0112] It should be understood that the various forms of processes shown above can be used to reorder, add, or delete steps. For example, the steps described in this invention can be executed in parallel, sequentially, or in different orders, as long as the desired result of the technical solution disclosed in this invention can be achieved, and this is not limited herein.

[0113] The specific embodiments described above do not constitute a limitation on the scope of protection of this invention. Those skilled in the art should understand that various modifications, combinations, sub-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 invention should be included within the scope of protection of this invention.

Claims

1. A method of eliminating interference of infrared thermal images of the neck and shoulder region, characterized in that, The method includes the following steps: An infrared thermal image of the target object is acquired as the image to be processed; the image to be processed is input into multiple pre-trained interference factor recognition models to identify the interference factors in the image to be processed, and obtain the corresponding recognition results; the interference factors include at least a first interference factor and a second interference factor. If the recognition result is a result that represents the recognition of any kind of interference factor, the image to be processed is used as a candidate image; Based on the recognition results, a prompt message is determined to adjust the candidate image; Based on the prompt information, the candidate image is adjusted accordingly to obtain the adjusted candidate image, which is then used as the target image of the image to be processed. The prompt information for adjusting the candidate image based on the recognition result includes: If the identified interfering factors include a first interfering factor, a prompt message for adjusting the candidate image is determined based on the size of the region to which the first interfering factor belongs and the region of the neck and shoulder area in the candidate image; If the identified interference factor includes a second interference factor, based on the distribution information of the second interference factor in the candidate image, a prompt message for adjusting the candidate image is determined, specifically including: For any identified second interference factor, determine whether the region to which the second interference factor belongs exists in the candidate image as a symmetrical region based on the human midline; if it exists, add the second interference factor to the current first record set; if it does not exist, add the second interference factor to the current second record set; the initial values ​​of the current first record set and the current second record set are both empty. For any second interference factor in the current first record set, determine whether there is an interference factor in the symmetrical region corresponding to the second interference factor; if not, move the second interference factor from the current first record set to the current third record set; the initial value of the current third record set is empty; If the difference between the number of identified second interference factors and the number of second interference factors in the third record set is less than or equal to a set difference, a first prompt message is generated to repair the corresponding region of the second interference factor based on the symmetrical region corresponding to the second interference factor in the current third record set, and a second prompt message is generated to repair the regions of the second interference factors in the current first record set and the current second record set.

2. The method of claim 1, wherein, The step of determining the prompt information for adjusting the candidate image based on the size of the region containing the first interfering factor and the neck and shoulder region in the candidate image specifically includes: Obtain the overlapping area between the region of the first interfering factor and the neck and shoulder region; If the obtained overlapping region is not empty, and the area ratio of the overlapping region to the neck and shoulder region is less than or equal to a set ratio, a prompt message is generated indicating that the overlapping region in the candidate image needs to be repaired.

3. The method of claim 2, wherein, Also includes: If the area ratio of the overlapping region to the neck and shoulder region is greater than a set ratio, a prompt message is generated to reacquire the infrared thermal image of the target object.

4. The method of claim 1, wherein, Also includes: If the difference between the number of identified second interference factors and the number of second interference factors in the third record set is greater than a set difference, the type of each identified second interference factor is obtained, and based on all the types of the obtained second interference factors, a prompt message to reacquire the infrared thermal image of the target object is generated.

5. The method of claim 4, wherein, The step of acquiring the type of each identified second interference factor and generating a prompt message to reacquire the infrared thermal image of the target object specifically includes: Obtain the feature vector of the region to which each second interference factor belongs; Obtain the similarity between the feature vector of the region to which each second interference factor belongs and each reference feature vector in the preset second interference factor type database, and obtain the similarity set corresponding to the second interference factor; The type corresponding to the maximum similarity in the similarity set corresponding to the second interference factor is taken as the type corresponding to the second interference factor. Based on the type of each second interference factor, obtain the remaining interference time corresponding to that second interference factor; The maximum of all remaining interference times is taken as the target time, and a prompt message is generated based on the target time to reacquire the infrared thermal image of the target object.

6. The method of claim 5, wherein, The feature vector of the region to which the second interference factor belongs is a feature vector formed by the encoded features of contact identifier, contact start time, time of separation from contact area, contact area ID, contact area temperature, contact area texture information, and contact area shape. Among them, the contact identifier is used to characterize whether the second interference factor is currently in contact with the contact area of ​​the target object. If they are in contact, the contact identifier is a first set value; otherwise, it is a second set value. The contact start time refers to the time when the second interference factor begins to make contact with the contact area of ​​the target object. The time of separation from contact area refers to the time when the second interference factor separates from the contact area of ​​the target object. The contact area ID is the position of the contact area of ​​the second interference factor in the body of the target object.

7. The method of claim 1, wherein, It also includes a third interference factor; the prompt message for adjusting the candidate image based on the recognition result further includes: If a third interfering factor is identified, a prompt message is generated indicating that the region in the candidate image to which the third interfering factor belongs needs to be adjusted.

8. An electronic device, comprising: Including processor and memory; The processor executes the steps of the method as described in any one of claims 1 to 7 by invoking programs or instructions stored in the memory.

9. A computer-readable storage medium, characterized in that, The computer-readable storage medium is used to store a program or instructions that cause a computer to perform the steps of the method as described in any one of claims 1 to 7.

Images