Method and device for enhancing visual field in large vessel operation based on AR (Augmented Reality)

Abstract

The invention discloses an AR (Augmented Reality)-based large vessel intraoperative visual field enhancement method and device, and relates to the technical field of image processing. The method comprises the following steps: acquiring medical image data of the heart and great vessels of a patient, and constructing a cardiovascular three-dimensional model with space coordinate information through segmentation and three-dimensional reconstruction; collecting real-time surgical field images of the thoracoscope and acquiring pose parameters of the real-time surgical field images; performing image enhancement on the real-time picture to obtain a target picture; and carrying out spatial registration according to the thoracoscope pose parameters and the three-dimensional model coordinate information, and carrying out superposition display in an augmented reality form at a display end. By means of the method, three-dimensional visualization of the visual field in the operation is achieved, multi-source information is fused on a single interface, attention distraction caused by multi-screen switching is avoided, the visual perception of an operator on the large blood vessel and the branch space relation of the large blood vessel is improved, and the visualization effect in the operation is improved.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of image processing, and in particular to an AR-based intraoperative field of view enhancement method and device for large vessels. BACKGROUND

[0002] In recent years, with the rapid development of minimally invasive cardiovascular surgical techniques, thoracoscopy and robot-assisted surgery have been widely used in the treatment of large vessel diseases. However, compared with traditional open chest surgery, the operative field of minimally invasive surgery is significantly limited, and doctors mainly rely on two-dimensional video images for operation, lacking intuitive spatial perception of deep tissues and complex anatomical structures. In this case, the surgeon has difficulty accurately determining the spatial positional relationship of the large vessels and their branches, especially when operating in the vicinity of important structures such as the aortic arch, pulmonary trunk, and superior vena cava. Limited field of view or anatomical misjudgment can increase the risk of intraoperative complications.

[0003] Although the current thoracoscopy imaging system has improved in resolution and color restoration, it still cannot fully present the three-dimensional spatial relationship of complex anatomical structures. The surgeon needs to rely on experience to build a spatial model in the brain to determine the relative positions of deep blood vessels, nerves, and lesions. This approach is highly subjective, increases the burden of operation, and increases the risk of surgery. Existing intraoperative image-assisted techniques mainly include preoperative three-dimensional reconstruction navigation and intraoperative ultrasound navigation. Although these techniques have improved intraoperative positioning and visualization to some extent, they still have many shortcomings. The preoperative three-dimensional reconstruction model can only be viewed offline before surgery and cannot be corresponded in real time with the actual intraoperative image. The intraoperative information display under thoracoscopy is still in a single-screen or multi-screen manner, and the navigation image and the real operative field are usually displayed separately, requiring the surgeon to switch attention between multiple display interfaces, which not only reduces operational efficiency but also increases the risk of errors. In summary, the existing thoracoscopy surgery still has problems such as insufficient visualization of the operative field and fragmented information display, which cannot meet the needs of precise identification and intuitive navigation for complex minimally invasive surgery of large vessels. SUMMARY

[0004] The present application aims to solve the problems of insufficient visualization of the operative field and fragmented information display mentioned in the background art, and proposes an AR-based intraoperative field of view enhancement method and device for large vessels.

[0005] The first aspect of the present application provides an AR-based intraoperative field of view enhancement method for large vessels, which comprises:

[0006] Obtaining medical image data of the patient's heart and large vessels, constructing a cardiovascular three-dimensional model with spatial coordinate information through segmentation and three-dimensional reconstruction;

[0007] Collecting real-time operative field images through a thoracoscope and synchronously obtaining the pose parameters of the thoracoscope;

[0008] performing image enhancement processing on the real-time surgical field picture to obtain a target picture;

[0009] According to the pose parameter of the thoracoscope and the spatial coordinate information of the cardiovascular three-dimensional model, the target picture is spatially registered with the three-dimensional model, and the registration result is displayed in the form of augmented reality on a display end.

[0010] Optionally, the image enhancement processing on the real-time surgical field picture to obtain a target picture includes:

[0011] The real-time surgical field picture is color corrected by using a white balance algorithm to obtain a first enhanced image;

[0012] The first enhanced image is converted to an HSV color space;

[0013] A plurality of first V-channel enhanced images are obtained by performing a plurality of differential enhancements on the V-channel image; the plurality of differential enhancements include logarithmic transformation, adaptive gamma correction, and contrast-limited adaptive histogram equalization;

[0014] The plurality of first V-channel enhanced images are fused to obtain a second V-channel enhanced image;

[0015] The second V-channel enhanced image and the original H-channel and S-channel are combined and mapped back to an RGB image to obtain a second enhanced image as the target picture.

[0016] Optionally, the fusing of the plurality of first V-channel enhanced images to obtain a second V-channel enhanced image includes:

[0017] According to an exposure, a first weight map corresponding to each first V-channel enhanced image is calculated;

[0018] The plurality of first weight maps are normalized at each pixel position to obtain a second weight map corresponding to each first V-channel enhanced image;

[0019] A corresponding Laplacian pyramid is established for each of the plurality of first V-channel enhanced images;

[0020] A corresponding Gaussian pyramid is established for the second weight map corresponding to each of the plurality of first V-channel enhanced images;

[0021] According to the Gaussian pyramid, each layer of the plurality of Laplacian pyramids is weighted and fused to obtain a new fused Laplacian pyramid;

[0022] According to the fused Laplacian pyramid, a reverse reconstruction is performed to obtain a fused image of the plurality of first V-channel enhanced images;

[0023] The fusion image is subjected to denoising processing to obtain a denoised fusion image as a second V channel enhanced image.

[0024] Optionally, the calculating of the first weight map corresponding to each first V channel enhanced image according to the exposure degree comprises:

[0025] The weight is calculated by using a Gaussian function:

[0026] ;

[0027] wherein, V m represents the mth first V channel enhanced image, (i, j) represents a pixel position; u m is the brightness mean value of the mth first V channel enhanced image; is a control parameter; exp is an exponential function with the natural constant e as the base; w m,1 is the first weight map corresponding to the mth first V channel enhanced image.

[0028] Optionally, the spatial registration of the target picture and the three-dimensional model according to the pose parameter of the thoracoscope and the spatial coordinate information of the cardiovascular three-dimensional model comprises:

[0029] According to the pose parameter of the thoracoscope, a transformation relationship between a thoracoscope coordinate system and a patient coordinate system is determined;

[0030] According to the spatial coordinate information of the cardiovascular three-dimensional model, the cardiovascular three-dimensional model is mapped to the patient coordinate system by point set registration to obtain an initial rough position of the cardiovascular three-dimensional model in the patient coordinate system;

[0031] According to the transformation relationship between the thoracoscope coordinate system and the patient coordinate system, the cardiovascular three-dimensional model is projected to the thoracoscope field of view to obtain an initial registration result;

[0032] The initial registration result is optimized by using a 2D-3D iterative closest point algorithm to obtain an accurate registration result.

[0033] The second aspect of the embodiment of the application provides a large blood vessel intraoperative field of view enhancement device based on AR, and the device comprises:

[0034] A model acquisition module is configured to acquire medical image data of a patient's heart and large blood vessels, and construct a cardiovascular three-dimensional model with spatial coordinate information through segmentation and three-dimensional reconstruction.

[0035] An operation field acquisition module is configured to acquire a real-time operation field picture through a thoracoscope and synchronously acquire a pose parameter of the thoracoscope.

[0036] A picture enhancement module is configured to perform image enhancement processing on the real-time operation field picture to obtain a target picture.

[0037] an AR display module configured to perform spatial registration of the target picture and the three-dimensional model according to the pose parameter of the thoracoscope and the spatial coordinate information of the three-dimensional model, and display a registration result in the form of augmented reality on a display end.

[0038] Optionally, the picture enhancement module comprises:

[0039] a color correction module configured to perform color correction on the real-time surgical field picture by using a white balance algorithm to obtain a first enhanced image;

[0040] a first conversion module configured to convert the first enhanced image to an HSV color space;

[0041] a differential enhancement module configured to perform a plurality of differential enhancements on the V channel image to obtain a plurality of first V channel enhanced images; the plurality of differential enhancements comprise logarithmic transformation, adaptive gamma correction and contrast limited adaptive histogram equalization;

[0042] a fusion module configured to fuse the plurality of first V channel enhanced images to obtain a second V channel enhanced image;

[0043] a second conversion module configured to combine the second V channel enhanced image and the original H channel and S channel, map back to an RGB image, and obtain a second enhanced image as the target picture.

[0044] Optionally, the fusion module comprises:

[0045] a weight calculation module configured to calculate a first weight map corresponding to each first V channel enhanced image according to an exposure, and normalize the plurality of first weight maps at each pixel position to obtain a second weight map corresponding to each first V channel enhanced image;

[0046] a first pyramid construction module configured to construct a corresponding Laplacian pyramid for each of the plurality of first V channel enhanced images;

[0047] a second pyramid construction module configured to construct a corresponding Gaussian pyramid for each of the plurality of first V channel enhanced image second weight maps;

[0048] a pyramid fusion module configured to perform weighted fusion on each layer of the plurality of Laplacian pyramids according to the Gaussian pyramids to obtain a new fused Laplacian pyramid;

[0049] an image reconstruction module configured to perform inverse reconstruction according to the fused Laplacian pyramid to obtain a fused image of the plurality of first V channel enhanced images;

[0050] The denoising module is used to denoise the fused image to obtain a denoised fused image, which serves as the second V-channel enhanced image.

[0051] Optionally, the weight calculation module uses a Gaussian function to calculate the weights, specifically:

[0052] ;

[0053] Among them, V m This represents the m-th image with the first V channel enhanced, where (i, j) represents the pixel position; u m It is the average brightness of the m-th first V channel enhanced image; It is a control parameter; exp is an exponential function with the natural constant e as its base; w m,1 It is the first weight map corresponding to the m-th first V channel enhanced image.

[0054] Optionally, the AR display module includes a spatial registration module for spatially registering the target image with the 3D model; the spatial registration module includes:

[0055] The first transformation module is used to determine the transformation relationship between the thoracoscope coordinate system and the patient coordinate system based on the pose parameters of the thoracoscope.

[0056] The second transformation module is used to map the cardiovascular 3D model to the patient coordinate system through point set registration based on the spatial coordinate information of the cardiovascular 3D model, so as to obtain the initial coarse position of the cardiovascular 3D model in the patient coordinate system.

[0057] The initial registration module is used to project the cardiovascular 3D model onto the thoracoscopic field of view according to the transformation relationship between the thoracoscopic coordinate system and the patient coordinate system, so as to obtain the initial registration result;

[0058] The fine registration module is used to optimize the initial registration result using a 2D-3D iterative nearest point algorithm to obtain a precise registration result.

[0059] The beneficial effects of this invention are:

[0060] By spatially registering and overlaying real-time surgical field images with cardiovascular 3D models, three-dimensional visualization of the intraoperative field of view is achieved. By integrating and displaying multi-source information on a single interface, attention distraction caused by switching between multiple screens is avoided, and the surgeon's intuitive perception of the spatial relationships of major blood vessels and their branches is enhanced. This effectively solves the problems of insufficient intraoperative field visualization and fragmented information display in existing major blood vessel surgeries. Attached Figure Description

[0061] Figure 1A flowchart of an AR-based intraoperative visual enhancement method for large vessel surgery is provided in an embodiment of the present invention;

[0062] Figure 2 This is a structural diagram of an AR-based intraoperative visual enhancement device for large blood vessels, provided as an embodiment of the present invention. Detailed Implementation

[0063] To further illustrate the technical means and effects of the present invention in achieving its intended purpose, the following detailed description of the specific implementation methods, structures, features, and effects of the present invention, in conjunction with the accompanying drawings and preferred embodiments, is provided below.

[0064] This invention provides an AR-based method for enhancing intraoperative visualization of large blood vessels. See also... Figure 1 , Figure 1 A flowchart illustrating an AR-based intraoperative visual enhancement method for large vessel surgery, provided as an embodiment of the present invention. The method includes the following steps:

[0065] S101: Acquire medical imaging data of the patient's heart and major blood vessels, and construct a cardiovascular 3D model with spatial coordinate information through segmentation and 3D reconstruction.

[0066] S102 acquires real-time surgical field images via thoracoscopy and simultaneously obtains the thoracoscopy's position and orientation parameters.

[0067] S103, perform image enhancement processing on the real-time surgical field image to obtain the target image.

[0068] S104, based on the pose parameters of the thoracoscope and the spatial coordinate information of the cardiovascular 3D model, spatially registers the target image with the 3D model, and displays the registration result in an augmented reality format on the display end.

[0069] This invention provides an AR-based method for enhancing the intraoperative field of vision in large blood vessels. By spatially registering and overlaying a 3D cardiovascular model with a real-time surgical field image, and displaying it using augmented reality, the method achieves 3D visualization of the intraoperative field of vision. This allows surgeons to clearly see the location of the target blood vessel and its surrounding structures, thereby reducing the risk of accidental injury to large blood vessels and important tissues. By integrating and displaying multi-source information on a single interface, the method avoids distraction caused by switching between multiple screens, enhances the surgeon's intuitive perception of the spatial relationships between large blood vessels and their branches, and effectively solves the problems of insufficient visualization of the surgical field and fragmented information display in existing large blood vessel surgeries.

[0070] In one embodiment, step S103 includes:

[0071] Step 1: Use a white balance algorithm to perform color correction on the real-time surgical field image to obtain the first enhanced image.

[0072] Step two: Convert the first enhanced image to the HSV color space.

[0073] Step 3: Perform various differential enhancements on the V channel image to obtain multiple first V channel enhanced images.

[0074] Step four: Fuse multiple first V channel enhanced images to obtain a second V channel enhanced image.

[0075] Step 5: Combine the enhanced image of the second V channel with the original H and S channels, and map it back to the RGB image to obtain the second enhanced image, which serves as the target image.

[0076] This embodiment improves the visualization of the surgical field by performing HSV spatial separation and multi-V channel differential enhancement on the white balance-corrected surgical field image, and then mapping it back to RGB after fusion processing. This makes the intraoperative image more realistic in color, more balanced in brightness, and clearer in local details and blood vessel edges, thereby significantly improving the visualization effect of the surgical field and helping doctors to accurately locate blood vessels and perform operations.

[0077] In one implementation, multiple differential enhancements include logarithmic transformation, adaptive gamma correction, and contrast-limited adaptive histogram equalization. Fusing multiple first V-channel enhanced images to obtain a second V-channel enhanced image includes:

[0078] Step 1: Calculate the first weight map corresponding to each enhanced image in the first V channel based on the exposure. Specifically, the weights are calculated using a Gaussian function:

[0079] Among them, V m This represents the m-th image with the first V channel enhanced, where (i, j) represents the pixel position; u m It is the average brightness of the m-th first V channel enhanced image; It is a control parameter, which can be set to 0.25; exp is an exponential function with the natural constant e as its base; w m,1 It is the first weight map corresponding to the m-th first V channel enhanced image.

[0080] Step 2: Normalize the multiple first weight maps at each pixel position to obtain the second weight map corresponding to each first V channel enhanced image:

[0081] Among them, w m,2 It is the second weight map corresponding to the m-th first V channel enhanced image; w k,1 It is the first weight map corresponding to the k-th first V channel enhanced image.

[0082] Step 3: For each of the multiple enhanced images of the first V channel, construct a corresponding Laplacian pyramid. The pyramid has 5 layers.

[0083] Step 4: For the second weight maps corresponding to the multiple first V channel enhanced images, establish corresponding Gaussian pyramids respectively.

[0084] Step 5: Based on the Gaussian pyramid, weighted merge of each layer of multiple Laplace pyramids to obtain a new merged Laplace pyramid.

[0085] Step 6: Based on the fused Laplacian pyramid, perform reverse reconstruction to obtain a fused image of multiple first V channel enhanced images.

[0086] Step seven involves denoising the fused image to obtain a denoised fused image, which serves as the second V-channel enhanced image. Specifically, noise may be introduced during the V-channel fusion process, and wavelet denoising algorithms can be used to optimize image quality.

[0087] Logarithmic transformation rapidly expands the dynamic range of an image by nonlinearly stretching dark areas and compressing bright areas, addressing the problem of low global brightness. Adaptive gamma correction dynamically adjusts the enhancement intensity based on the average brightness of the image, ensuring sufficient brightening of dark areas while avoiding overexposure in bright areas. Contrast-limited adaptive histogram equalization enhances details in dark areas through local equalization, preserving texture and edges while suppressing halo and noise amplification. However, logarithmic transformation alone can easily lose local details, adaptive gamma correction alone is insufficient in extremely dark areas, and contrast-limited adaptive histogram equalization alone may lead to overexposure in bright areas. This implementation combines these three strategies to simultaneously address global brightness, adaptive local enhancement, and detail preservation, avoiding the limitations of a single method and achieving a brightening effect that is rich in detail and naturally balanced.

[0088] Using pyramid fusion enables the layered processing and fusion of images generated by different enhancement strategies across multiple scales, balancing global brightness consistency with local detail clarity. By smoothing the weight distribution in the Gaussian pyramid and preserving detail information in the Laplacian pyramid, the fusion result effectively avoids edge abruptness, artifacts, and unnatural transitions caused by direct pixel-level fusion, thus achieving high-quality enhanced images with balanced brightness, rich detail, and distinct layers.

[0089] In one embodiment, step S104 includes:

[0090] Step 1: Determine the transformation relationship between the thoracoscope coordinate system and the patient coordinate system based on the thoracoscope's position parameters.

[0091] Step 2: Based on the spatial coordinate information of the cardiovascular 3D model, the cardiovascular 3D model is mapped to the patient coordinate system through point set registration to obtain the initial rough position of the cardiovascular 3D model in the patient coordinate system.

[0092] Step 3: Based on the transformation relationship between the thoracoscopic coordinate system and the patient coordinate system, the cardiovascular 3D model is projected onto the thoracoscopic field of view to obtain the initial registration result.

[0093] Step four: The initial registration result is optimized using the 2D-3D iterative nearest point algorithm to obtain the accurate registration result.

[0094] This embodiment achieves precise spatial registration between the intraoperative cardiovascular 3D model and the real-time surgical field image by transforming the thoracoscopic pose and the patient coordinate system, registering the 3D model point set, and optimizing the 2D-3D iterative nearest point. This enables accurate superposition of the patient's vascular 3D structure in the thoracoscopic field of view, improving surgical positioning accuracy, reducing the risk of vascular injury, enhancing intraoperative visualization, and providing intuitive navigation for minimally invasive procedures.

[0095] This invention provides an AR-based intraoperative visual enhancement device for large vessel surgery. See also... Figure 2 , Figure 2 This is a structural diagram of an AR-based intraoperative visual enhancement device for large vessel surgery, provided as an embodiment of the present invention. The device includes:

[0096] The model acquisition module is used to acquire multimodal medical image data of the patient's heart and major blood vessels, and construct a cardiovascular 3D model with spatial coordinate information through segmentation and 3D reconstruction.

[0097] The surgical field acquisition module is used to acquire real-time surgical field images through thoracoscopy and simultaneously acquire the thoracoscopy's pose parameters.

[0098] The image enhancement module is used to perform image enhancement processing on the real-time surgical field image to obtain the target image.

[0099] The AR display module is used to spatially register the target image with the 3D model based on the pose parameters of the thoracoscope and the spatial coordinate information of the cardiovascular 3D model, and then overlay the registration result on the display end in the form of augmented reality.

[0100] This invention provides an AR-based intraoperative visualization device for large blood vessels. By spatially registering and overlaying a 3D cardiovascular model with a real-time surgical field image, it achieves 3D visualization of the intraoperative field, allowing surgeons to clearly see the location of the target blood vessel and its surrounding structures, thereby reducing the risk of accidental injury to large blood vessels and important tissues. By integrating and displaying multi-source information on a single interface, it avoids the distraction caused by switching between multiple screens, enhancing the surgeon's intuitive perception of the spatial relationships between large blood vessels and their branches. This effectively solves the problems of insufficient visualization of the surgical field and fragmented information display in existing large blood vessel surgeries.

[0101] In one embodiment, the image enhancement module includes:

[0102] The color correction module is used to perform color correction on the real-time surgical field image using a white balance algorithm to obtain the first enhanced image.

[0103] The first conversion module is used to convert the first enhanced image to the HSV color space.

[0104] The differential enhancement module is used to perform various differential enhancements on the V channel image to obtain multiple first V channel enhanced images; the various differential enhancements include logarithmic transformation, adaptive gamma correction and contrast-limited adaptive histogram equalization.

[0105] The fusion module is used to fuse multiple first V channel enhanced images to obtain a second V channel enhanced image.

[0106] The second conversion module is used to combine the enhanced image of the second V channel with the original H and S channels and map it back to an RGB image to obtain the second enhanced image, which serves as the target image.

[0107] In one implementation, the fusion module includes:

[0108] The weight calculation module is used to calculate the first weight map corresponding to each first V channel enhanced image based on the exposure; and to normalize the multiple first weight maps at each pixel position to obtain the second weight map corresponding to each first V channel enhanced image.

[0109] The first pyramid construction module is used to build corresponding Laplacian pyramids for multiple first V channel enhanced images.

[0110] The second pyramid construction module is used to build corresponding Gaussian pyramids for the second weight maps corresponding to multiple first V channel enhanced images.

[0111] The pyramid fusion module is used to weightedly fuse each layer of multiple Laplace pyramids based on the Gaussian pyramid to obtain a new fused Laplace pyramid.

[0112] The image reconstruction module is used to perform reverse reconstruction based on the fused Laplacian pyramid to obtain a fused image of multiple first V channel enhanced images.

[0113] The denoising module is used to denoise the fused image to obtain a denoised fused image, which serves as the second V channel enhancement image.

[0114] In one implementation, the weight calculation module uses a Gaussian function to calculate the weights, specifically:

[0115] ;

[0116] Among them, V mThis represents the m-th image with the first V channel enhanced, where (i, j) represents the pixel position; u m It is the average brightness of the m-th first V channel enhanced image; It is a control parameter; exp is an exponential function with the natural constant e as its base; w m,1 It is the first weight map corresponding to the m-th first V channel enhanced image.

[0117] In one embodiment, the AR display module includes a spatial registration module for spatially registering the target image with a 3D model; the spatial registration module includes:

[0118] The first transformation module is used to determine the transformation relationship between the thoracoscope coordinate system and the patient coordinate system based on the thoracoscope's pose parameters.

[0119] The second transformation module is used to map the cardiovascular 3D model to the patient coordinate system through point set registration based on the spatial coordinate information of the cardiovascular 3D model, thereby obtaining the initial rough position of the cardiovascular 3D model in the patient coordinate system.

[0120] The initial registration module is used to project the cardiovascular 3D model onto the thoracoscopic field of view based on the transformation relationship between the thoracoscopic coordinate system and the patient coordinate system, and obtain the initial registration result.

[0121] The fine registration module is used to optimize the initial registration result using a 2D-3D iterative nearest point algorithm to obtain a precise registration result.

[0122] The above description is merely a preferred embodiment of the present invention and is not intended to limit the present invention in any way. Although the present invention has been disclosed above with reference to preferred embodiments, it is not intended to limit the present invention. Any person skilled in the art can make some modifications or alterations to the above-disclosed technical content to create equivalent embodiments without departing from the scope of the present invention. Any simple modifications, equivalent changes and alterations made to the above embodiments based on the technical essence of the present invention without departing from the scope of the present invention shall fall within the scope of the present invention.

Claims

1. An AR-based intraoperative field of view enhancement method for large vessels, characterized in that, The method comprises: acquiring medical image data of a patient's heart and large blood vessels, constructing a cardiovascular three-dimensional model with spatial coordinate information through segmentation and three-dimensional reconstruction; collecting a real-time surgical field picture through a thoracoscope and synchronously acquiring a pose parameter of the thoracoscope; performing image enhancement processing on the real-time surgical field picture to obtain a target picture; according to the pose parameter of the thoracoscope and the spatial coordinate information of the cardiovascular three-dimensional model, performing spatial registration on the target picture and the three-dimensional model, and displaying the registration result in the form of augmented reality on a display end.

2. The AR-based intraoperative large vessel field enhancement method of claim 1, wherein, The image enhancement processing on the real-time surgical field picture to obtain a target picture comprises: performing color correction on the real-time surgical field picture by using a white balance algorithm to obtain a first enhanced image; converting the first enhanced image to an HSV color space; performing a plurality of differential enhancement on a V channel image to obtain a plurality of first V channel enhanced images; the plurality of differential enhancement comprises logarithmic transformation, adaptive gamma correction and contrast limited adaptive histogram equalization; fusing the plurality of first V channel enhanced images to obtain a second V channel enhanced image; combining the second V channel enhanced image and original H and S channels to map back to an RGB image to obtain a second enhanced image as the target picture.

3. The AR-based intraoperative large vessel field enhancement method of claim 2, wherein, The fusing of the plurality of first V channel enhanced images to obtain a second V channel enhanced image comprises: calculating a first weight map corresponding to each first V channel enhanced image according to an exposure; normalizing the plurality of first weight maps at each pixel position to obtain a second weight map corresponding to each first V channel enhanced image; establishing a corresponding Laplacian pyramid for each of the plurality of first V channel enhanced images; establishing a corresponding Gaussian pyramid for the second weight map corresponding to each of the plurality of first V channel enhanced images; weighting and fusing each layer of the plurality of Laplacian pyramids according to the Gaussian pyramids to obtain a new fused Laplacian pyramid; performing reverse reconstruction according to the fused Laplacian pyramid to obtain a fused image of the plurality of first V channel enhanced images; performing denoising processing on the fused image to obtain a denoised fused image as the second V channel enhanced image.

4. The AR-based intraoperative large vessel field enhancement method of claim 3, wherein, The calculation of the first weight map corresponding to each first V channel enhanced image according to the exposure comprises: calculating the weight by using a Gaussian function: ； wherein V m represents the mth first V-channel enhanced image, (i, j) represents a pixel position; u m is the brightness mean value of the mth first V-channel enhanced image; is a control parameter; exp is an exponential function with the natural constant e as the base; w m,1 is the first weight map corresponding to the mth first V-channel enhanced image.

5. The AR-based intraoperative large vessel field enhancement method of claim 1, wherein, The spatial registration of the target picture and the three-dimensional model according to the pose parameter of the thoracoscope and the spatial coordinate information of the cardiovascular three-dimensional model comprises: determining a transformation relationship between a thoracoscope coordinate system and a patient coordinate system according to the pose parameter of the thoracoscope; mapping the cardiovascular three-dimensional model to the patient coordinate system by point set registration to obtain an initial rough position of the cardiovascular three-dimensional model in the patient coordinate system according to the spatial coordinate information of the cardiovascular three-dimensional model; projecting the cardiovascular three-dimensional model to a thoracoscope field of view according to the transformation relationship between the thoracoscope coordinate system and the patient coordinate system to obtain an initial registration result; optimizing the initial registration result by using a 2D-3D iterative closest point algorithm to obtain an accurate registration result.

6. An AR-based intraoperative field of view enhancement device for large vessels, characterized in that, The device comprises: The model obtaining module is configured to obtain medical image data of a heart and large blood vessels of a patient, and construct a three-dimensional cardiovascular model with spatial coordinate information through segmentation and three-dimensional reconstruction; The operation field obtaining module is configured to collect a real-time operation field image through a thoracoscope and synchronously obtain a pose parameter of the thoracoscope; The image enhancement module is configured to perform image enhancement processing on the real-time operation field image to obtain a target image; The AR display module is configured to perform spatial registration on the target image and the three-dimensional cardiovascular model according to the pose parameter of the thoracoscope and the spatial coordinate information of the three-dimensional cardiovascular model, and display the registration result in the form of augmented reality on a display terminal.

7. The AR-based intraoperative large vessel field augmentation device of claim 6, wherein, The image enhancement module includes: The color correction module is configured to perform color correction on the real-time operation field image by using a white balance algorithm to obtain a first enhanced image; The first conversion module is configured to convert the first enhanced image to an HSV color space; The differential enhancement module is configured to perform a plurality of differential enhancements on a V channel image to obtain a plurality of first V channel enhanced images; the plurality of differential enhancements include logarithmic transformation, adaptive gamma correction and contrast limited adaptive histogram equalization; The fusion module is configured to fuse the plurality of first V channel enhanced images to obtain a second V channel enhanced image; The second conversion module is configured to combine the second V channel enhanced image and original H and S channels, map back to an RGB image, and obtain a second enhanced image as the target image.

8. The AR-based intraoperative large vessel field augmentation device of claim 7, wherein, The fusion module includes: The weight calculation module is configured to calculate a first weight map corresponding to each first V channel enhanced image according to an exposure, and normalize a plurality of first weight maps at each pixel position to obtain a second weight map corresponding to each first V channel enhanced image; The first pyramid construction module is configured to construct a corresponding Laplacian pyramid for each of the plurality of first V channel enhanced images; The second pyramid construction module is configured to construct a corresponding Gaussian pyramid for each of the plurality of first V channel enhanced images; The pyramid fusion module is configured to perform weighted fusion on each layer of the plurality of Laplacian pyramids according to the Gaussian pyramids to obtain a new fused Laplacian pyramid; The image reconstruction module is configured to perform inverse reconstruction according to the fused Laplacian pyramid to obtain a fused image of the plurality of first V channel enhanced images; The denoising module is configured to perform denoising processing on the fused image to obtain a denoised fused image as the second V channel enhanced image.

9. The AR-based intraoperative large-vessel visual field augmentation device of claim 8, wherein, The weight calculation module calculates the weight by using a Gaussian function, and specifically: ； wherein V m represents the mth first V-channel enhanced image, (i, j) represents a pixel position; u m is the brightness mean value of the mth first V-channel enhanced image; is a control parameter; exp is an exponential function with base of natural constant e; w m,1 is the first weight map corresponding to the mth first V-channel enhanced image.

10. The AR-based intraoperative large-vessel visual field augmentation device of claim 6, wherein, The AR display module includes a spatial registration module configured to perform spatial registration on the target image and the three-dimensional model; The spatial registration module includes: The first transformation module is configured to determine a transformation relationship between a thoracoscope coordinate system and a patient coordinate system according to the pose parameter of the thoracoscope; The second transformation module is configured to map the three-dimensional cardiovascular model to the patient coordinate system by point set registration according to the spatial coordinate information of the three-dimensional cardiovascular model to obtain an initial rough position of the three-dimensional cardiovascular model in the patient coordinate system; An initial registration module is configured to project the cardiovascular three-dimensional model to the thoracoscope visual field according to the transformation relationship between the thoracoscope coordinate system and the patient coordinate system, and obtain an initial registration result. A fine registration module is configured to optimize the initial registration result by using a 2D-3D iterative closest point algorithm, and obtain a fine registration result.

Images