Image stitching method and device for X-ray photography, host, medium and product

Abstract

The embodiment of the invention discloses an X-ray photography image splicing method and device, a host, a medium and a product. The method comprises the following steps: determining a three-dimensional model of a photographic object based on sensing data of the photographic object; selecting a splicing area in the three-dimensional model; performing a plurality of exposures on the photographic object based on the stitching region to generate a plurality of X-ray images; and splicing a plurality of X-ray images. Based on the three-dimensional model of the photographic object, the motion control precision and the image splicing quality are improved. Accurate exposure dose control is realized by controlling the pixel activation state in the flat panel detector. And the foreground and the background are spliced and then fused, so that the image splicing quality is further improved.

Description

Technical Field

[0001] This invention relates to the field of medical equipment technology, and in particular to image stitching methods, devices, main units, media, and products for X-ray imaging. Background Technology

[0002] X-rays are electromagnetic radiation with wavelengths between ultraviolet and gamma rays. X-rays are penetrating, with varying penetration capabilities through materials of different densities. In medicine, X-rays are generally used to project images of human organs and bones to create medical images. Direct digital radiology (DR) technology, characterized by its fast imaging speed, ease of operation, and high imaging resolution, has become the dominant direction in X-ray imaging. An X-ray tube uses high voltage provided by a high-voltage generator to emit X-rays that pass through the target, forming a medical image of the target on a flat panel detector. The flat panel detector then transmits the medical image information to a remote control console. The target can stand near the chest X-ray stand assembly or lie on the examination table assembly, allowing for X-ray imaging of different parts of the body, such as the head, chest, abdomen, and joints.

[0003] In some cases, it is necessary to expose elongated imaging targets (such as the spine). However, the length of such targets may be much larger than the size of the X-ray detector. Long bone stitching techniques can be used to stitch multiple individual X-ray images together into a single image that displays the complete target.

[0004] Improving the quality of stitched images is one of the key concerns in the industry. Summary of the Invention

[0005] The present invention provides an image stitching method, apparatus, host, medium, and product for X-ray imaging.

[0006] An image stitching method for X-ray photography includes:

[0007] Based on the sensor data of the photographed object, a three-dimensional model of the photographed object is determined;

[0008] Select the splicing area in the 3D model;

[0009] Based on the stitching region, multiple exposures are performed on the photographic object to generate multiple X-ray images;

[0010] The multiple X-ray images are stitched together.

[0011] It is evident that generating multiple X-ray images based on the selected stitching area in the 3D model of the photographic object improves motion control precision, thereby improving the quality of the X-ray images to be stitched and further enhancing the stitching quality.

[0012] In one implementation, determining the three-dimensional model of the photographic object based on the sensor data of the photographic object includes:

[0013] The sensor data is input into a trained artificial intelligence model, which is adapted to determine the skeletal model of the photographed object based on the sensor data.

[0014] Extract the morphological parameters of the photographed object from the sensor data;

[0015] Based on the morphological parameters and the skeletal model, a three-dimensional model of the photographic object is obtained by rendering.

[0016] Therefore, introducing artificial intelligence technology into image stitching improves the accuracy of 3D models and reduces the difficulty of generating 3D models.

[0017] In one implementation, the method includes a training process for the artificial intelligence model;

[0018] The training process includes:

[0019] Training samples containing training sensor data and labels are input into the artificial intelligence model, wherein the labels contain a skeletal model of the training sensor data;

[0020] Receive a skeletal model predicted based on the training sensor data from the artificial intelligence model;

[0021] The loss function value of the artificial intelligence model is determined based on the difference between the predicted skeletal model and the label.

[0022] Configure the model parameters of the artificial intelligence model so that the loss function value is lower than a preset threshold.

[0023] Therefore, it is possible to quickly train artificial intelligence models.

[0024] In one implementation, performing multiple exposures on the photographic subject to generate multiple X-ray images includes:

[0025] In the three-dimensional model, multiple stitching sub-regions with multiple exposures are selected from the stitching region;

[0026] Identify the multiple pixel regions in the flat panel detector that surround each stitched sub-region and correspond to multiple exposures;

[0027] In each exposure, pixels in the corresponding pixel region are controlled to be in an active state, while pixels outside the pixel region are in an inactive state.

[0028] Therefore, precise exposure dose control is achieved by controlling the pixel activation state in the flat panel detector.

[0029] In one embodiment, determining the plurality of pixel regions in the flat panel detector that respectively surround each stitching sub-region and correspond to multiple exposures includes:

[0030] Determine the corresponding region in the flat panel detector for each stitched sub-region;

[0031] Determine the minimum bounding rectangle of the corresponding region;

[0032] The multiple smallest bounding rectangles corresponding to the multiple splicing sub-regions are determined as the multiple pixel regions.

[0033] Therefore, pixel regions can be quickly determined based on the minimum bounding rectangle.

[0034] In one embodiment, stitching together the multiple X-ray images includes:

[0035] Each of the multiple X-ray images is segmented into foreground and background;

[0036] In the background of each X-ray image, the contour lines of the selected area are determined.

[0037] Multiple contour lines from the multiple X-ray images are stitched together to form a stitched background;

[0038] Detect the foreground of each X-ray image within its selected region to serve as the stitching foreground;

[0039] The stitched background and the stitched foreground are merged to generate a stitched image of the multiple X-ray images.

[0040] Therefore, by separately stitching and then merging the foreground and background, the quality of the stitched image is improved.

[0041] An image stitching device for X-ray photography, comprising:

[0042] The determination module is used to determine the three-dimensional model of the photographic object based on the sensor data of the photographic object;

[0043] The selection module is used to select the stitching area in the 3D model;

[0044] An exposure module is used to perform multiple exposures on the photographic object based on the stitching area to generate multiple X-ray images;

[0045] The stitching module is used to stitch together the multiple X-ray images.

[0046] It is evident that generating multiple X-ray images based on the selected stitching area in the 3D model of the photographic object improves motion control precision, thereby improving the quality of the X-ray images to be stitched and further enhancing the stitching quality.

[0047] In one embodiment, the exposure module is configured to select multiple stitching sub-regions for multiple exposures from the stitching region in the three-dimensional model; determine multiple pixel regions in the flat panel detector that surround each stitching sub-region and correspond to the multiple exposures; and control the pixels in the corresponding pixel regions to be in an active state and the pixels outside the pixel regions to be in an inactive state during each exposure.

[0048] Therefore, precise exposure dose control is achieved by controlling the pixel activation state in the flat panel detector.

[0049] In one embodiment, the exposure module is configured to determine the corresponding region of each stitched sub-region in the flat panel detector; determine the minimum bounding rectangle of the corresponding region; and determine the multiple minimum bounding rectangles corresponding to the multiple stitched sub-regions as the multiple pixel regions.

[0050] Therefore, pixel regions can be quickly determined based on the minimum bounding rectangle.

[0051] In one embodiment, the stitching module is configured to segment each of the plurality of X-ray images into a foreground and a background; determine contour lines within a selected region of each X-ray image's background; stitch together multiple contour lines from the plurality of X-ray images to form a stitched background; detect the foreground of each X-ray image within its selected region to form a stitched foreground; and fuse the stitched background and the stitched foreground to generate a stitched image of the plurality of X-ray images.

[0052] Therefore, by separately stitching and then merging the foreground and background, the quality of the stitched image is improved.

[0053] A control unit for an X-ray imaging system includes:

[0054] processor;

[0055] Memory for storing the executable instructions of the processor;

[0056] The processor is configured to read the executable instructions from the memory and execute the executable instructions to implement the image stitching method of X-ray photography as described above.

[0057] A computer-readable storage medium having computer instructions stored thereon, which, when executed by a processor, implement the image stitching method of X-ray photography as described above.

[0058] A computer program product includes a computer program that, when executed by a processor, implements the image stitching method of X-ray photography as described above. Attached Figure Description

[0059] Preferred embodiments of the present invention will now be described in detail with reference to the accompanying drawings, which will make the above and other features and advantages of the present invention more apparent to those skilled in the art. In the drawings:

[0060] Figure 1 This is a flowchart of an image stitching method for X-ray photography according to an embodiment of the present invention.

[0061] Figure 2 This is an exemplary schematic diagram of the training process of an artificial intelligence model according to an embodiment of the present invention.

[0062] Figure 3 This is an exemplary schematic diagram illustrating the generation of a three-dimensional model of a photographic object according to an embodiment of the present invention.

[0063] Figure 4 This is an exemplary schematic diagram illustrating the determination of a pixel region according to an embodiment of the present invention.

[0064] Figure 5 This is an exemplary schematic diagram of image stitching according to an embodiment of the present invention.

[0065] Figure 6 This is a schematic diagram of the image stitching process of X-ray photography according to an embodiment of the present invention.

[0066] Figure 7 This is an exemplary structural diagram of an image stitching apparatus for X-ray photography according to an embodiment of the present invention.

[0067] Figure 8 This is an exemplary structural diagram of the control host of an X-ray imaging system according to an embodiment of the present invention.

[0068] The reference numerals in the attached figures are as follows:

[0069]

[0070]

[0071] Detailed Implementation

[0072] To make the objectives, technical solutions, and advantages of this invention clearer, the following embodiments are provided to further illustrate the invention in detail. The nouns and pronouns referring to "person" in this patent application are not limited to specific genders.

[0073] For the sake of brevity and intuitiveness, the following description uses several representative embodiments to illustrate the solution of the present invention. Numerous details in the embodiments are only used to aid in understanding the solution of the present invention. However, it is obvious that the technical solution of the present invention can be implemented without being limited to these details. To avoid unnecessarily obscuring the solution of the present invention, some embodiments are not described in detail, but only a framework is given. In the following text, "comprising" means "including but not limited to," and "according to..." means "at least according to..., but not limited to only according to...". Due to Chinese language habits, unless the quantity of a component is specifically indicated below, it means that the component can be one or more, or can be understood as at least one.

[0074] The applicant found that there is room for improvement in both image quality and dose control in the current image stitching application of X-ray imaging systems. For example: (1) due to insufficient motion control precision, the quality of the image to be stitched is poor (e.g., white edges may appear in the image); (2) it is difficult to achieve accurate dose control; (3) due to defects in the stitching algorithm, there is room for improvement in the quality of the stitched image.

[0075] In this embodiment of the invention, precise motion control is performed on the actuators driving the X-ray tube and flat panel detector based on a precise three-dimensional model of the photographic object. This improves motion control accuracy, avoids overexposure or underexposure defects, and effectively overcomes or reduces potential white edges in the image. Furthermore, by selecting stitching sub-regions from the three-dimensional model, the state control of corresponding pixels in the flat panel detector is achieved, improving the accuracy of dose control. Additionally, the process of first distinguishing and stitching the foreground and background before merging improves the quality of the stitched image.

[0076] The above disclosure details the technical defects existing in the prior art, the causes of these defects, and the analytical process for overcoming them. In fact, the understanding of these technical defects is not common knowledge in the field, but rather a novel discovery made by the applicant during their research. Furthermore, the tracing of the causes of these defects and the analytical process for overcoming them are also the results of the applicant's gradual analysis during the actual research process, and are not common knowledge in the field.

[0077] Figure 1 This is a flowchart of an image stitching method for X-ray photography according to an embodiment of the present invention. Preferably, it can be executed by a controller. Figure 1 The method is shown. The controller can be implemented as a control host integrated into the X-ray imaging system, or as a control unit independent of the control host.

[0078] like Figure 1 As shown, the method includes:

[0079] Step 101: Determine the three-dimensional model of the photographic object based on the sensor data of the photographic object.

[0080] Here, various types of sensors can be used to acquire sensory data of the subject being photographed (e.g., a patient or person undergoing a medical examination). For example, sensory data can include: images (e.g., two-dimensional or three-dimensional images), depth information, and point cloud data, etc. Accordingly, sensors can be implemented as cameras (two-dimensional or three-dimensional cameras), depth sensors, and point cloud scanning devices (e.g., point cloud scanners), etc. Sensors can be positioned in the examination room where the subject is located, at locations suitable for performing sensing processing on the subject (e.g., ceiling or walls, etc.). Sensors can also be positioned in an X-ray system, such as on an X-ray tube.

[0081] In one implementation, step 101 includes: inputting sensor data into a trained artificial intelligence model, the artificial intelligence model being adapted to determine the skeletal model of the photographic object based on the sensor data; extracting morphological parameters of the photographic object from the sensor data; and obtaining a three-dimensional model of the photographic object by rendering based on the morphological parameters and the skeletal model.

[0082] For example, morphological parameters can include the overall height of the photographic subject, limb length, arm span to height ratio, or sitting height to height ratio, etc. Based on these morphological parameters, the subject's skin system is determined, and then the skin system is rendered onto the skeletal model to form a 3D model of the photographic subject.

[0083] Therefore, introducing artificial intelligence technology into image stitching improves the accuracy of 3D models and reduces the difficulty of generating 3D models.

[0084] In one implementation, the method includes a training process for an artificial intelligence (AI) model. The training process includes: inputting training samples containing training sensor data and labels into the AI ​​model, wherein the labels contain a skeletal model of the training sensor data; receiving a predicted skeletal model from the AI ​​model based on the training sensor data; determining a loss function value for the AI ​​model based on the difference between the predicted skeletal model and the labels; and configuring the model parameters of the AI ​​model so that the loss function value is below a preset threshold. Therefore, rapid training of the AI ​​model is achieved.

[0085] The network architecture of the artificial intelligence model can be implemented as an encoder-decoder architecture, a convolutional neural network (CNN), a recurrent neural network (RNN), or a long short-term memory network (LSTM), etc. In fact, the embodiments of this invention do not limit the network architecture of the artificial intelligence model.

[0086] Figure 2 This is an exemplary schematic diagram illustrating the training process of an artificial intelligence model according to an embodiment of the present invention. Figure 2As shown, training sample 30 includes training sensor data 20 and labels 21. Training sensor data 20 can be historical sensor data of the training photographic object (which may be different from the training object in step 101). Label 21 is the skeletal model of the labeled training photographic object. Training sample 30 is input into artificial intelligence model 40. Artificial intelligence model 40 predicts skeletal model 41 based on training sensor data 20. The difference between the predicted skeletal model 41 and label 21 is calculated to obtain the loss function value 42 of artificial intelligence model 40. Then, based on the loss function value 42 of artificial intelligence model 40, backpropagation is performed on artificial intelligence model 401 to update the model parameters of artificial intelligence model 40 until the loss function value 42 of artificial intelligence model 40 is less than a preset value, thus completing the training process of artificial intelligence model 40. The trained artificial intelligence model 40 has the ability to predict skeletal models based on input sensor data.

[0087] Step 102: Select the splicing area in the 3D model.

[0088] For example, it can respond to a user-triggered selection command to select a stitching area in a 3D model. For instance, the 3D model is displayed on a screen (e.g., the display of a control unit). Then, it receives a selection command from the user based on the screen to select a stitching area in the 3D model.

[0089] In addition to the selection commands triggered by the display interface as described above, selection commands can also be triggered by various other methods such as voice control and motion control. The embodiments of the present invention are not limited in this regard.

[0090] Step 103: Based on the stitching region, perform multiple exposures on the photographic object to generate multiple X-ray images.

[0091] Based on the selected stitching area on the digital 3D model of the photographic object, multiple exposures can be performed on the object to generate multiple X-ray images. The 3D model accurately represents the photographic object, thus enabling precise digital control.

[0092] For example, motion control commands can be generated based on the stitching area. These commands drive the X-ray tube and flat panel detector to move synchronously within a movement range corresponding to the stitching area. Here, motion control commands are generated based on the stitching area selected in the 3D model to drive the X-ray tube and flat panel detector to move synchronously within a movement range corresponding to the stitching area. The actuators of the X-ray tube and flat panel detector execute the motion control commands to drive the X-ray tube and flat panel detector to move synchronously within the movement range corresponding to the stitching area.

[0093] For example: A transformation matrix between the coordinate system of the 3D model and the real-world coordinate system is pre-established. Then, based on the transformation matrix and the stitching region, the movement range of the X-ray tube and flat panel detector suitable for performing X-ray imaging on that stitching region is determined. Next, based on the determined movement range, motion control commands are generated to control the actuators of the X-ray tube and flat panel detector. The actuators execute these motion control commands to drive the X-ray tube and flat panel detector to move synchronously within that movement range.

[0094] It is evident that generating commands to drive the synchronous movement of the X-ray tube and the flat panel detector based on the selected stitching area in the 3D model of the photographic object improves motion control accuracy, enhances the quality of the X-ray image to be stitched, and improves the quality of the stitched image.

[0095] Here, multiple exposures can be performed to generate multiple X-ray images during the synchronized movement of the X-ray tube and the flat panel detector.

[0096] In one implementation, step 103 includes: selecting multiple stitching sub-regions for multiple exposures from the stitching region in the 3D model; determining multiple pixel regions in the flat panel detector that surround each stitching sub-region and correspond to the multiple exposures; and controlling the pixels in the corresponding pixel region to be in an active state during each exposure, while pixels outside the pixel region are in an inactive state. Pixels in the active state can detect X-rays; pixels in the inactive state do not detect X-rays. When the received dose of the pixel region of the flat panel detector reaches a preset dose, the flat panel detector reports to the control host, and the control host controls the X-ray tube to stop emitting X-rays. Here, the corresponding pixel region is: the pixel region in the flat panel detector used to receive the dose from the stitching sub-region of that exposure during each exposure.

[0097] Therefore, by controlling the pixel activation state in the flat panel detector, only pixels in the pixel area (surrounding the splicing sub-area) detect X-rays, avoiding overexposure and causing unnecessary radiation to patients, and also achieving precise exposure dose control.

[0098] In one implementation, determining the multiple pixel regions in the flat panel detector that surround each stitched sub-region and correspond to multiple exposures includes: determining the corresponding region of each stitched sub-region in the flat panel detector; determining the minimum bounding rectangle (MBR) of the corresponding region; and determining the multiple minimum bounding rectangles corresponding to the multiple stitched sub-regions as multiple pixel regions.

[0099] Step 104: Stitch together multiple X-ray images.

[0100] In one implementation, step 104 includes: segmenting each of the multiple X-ray images into a foreground and a background; determining contour lines within a selected region of each X-ray image's background; stitching together multiple contour lines from the multiple X-ray images to form a stitched background; detecting the foreground of each X-ray image within its selected region to form a stitched foreground; and fusing the stitched background and foreground to generate a stitched image of the multiple X-ray images. Therefore, by separately stitching and fusing the foreground and background, the quality of the stitched image is improved.

[0101] Figure 3 This is an exemplary schematic diagram illustrating the generation of a three-dimensional model of a photographic object according to an embodiment of the present invention. Figure 3 In this process, a skeletal model 50 can be obtained based on the sensor data and artificial intelligence model of the photographic object. Then, based on the morphological parameters of the photographic object and the skeletal model 50, a 3D model 51 of the photographic object is obtained through rendering. By using the transformation matrix between the coordinate system of the 3D model 51 and the real-world coordinate system, the movement range of the X-ray tube and flat panel detector performing X-ray imaging on the corresponding stitching area of ​​the photographic object in the real-world coordinate system can be determined based on the coordinate range of the selected stitching area in the 3D model 51. Then, motion control commands are generated to control the actuators of the X-ray tube and flat panel detector. The actuators execute the motion control commands to drive the X-ray tube and flat panel detector to move synchronously within the movement range.

[0102] Figure 4 This is an exemplary schematic diagram illustrating the determination of a pixel region according to an embodiment of the present invention.

[0103] During each exposure, the following sub-steps are performed:

[0104] Sub-step (1): In the 3D model, determine the stitching sub-regions in this exposure from the stitching regions. For example, such as... Figure 4 The splicing sub-region 52 is shown.

[0105] Sub-step (2): Determine the minimum bounding rectangle 53 of the flat panel detector that surrounds the splicing sub-region 52.

[0106] Sub-step (3): Control the pixels within the minimum bounding rectangle 53 to be in an active state, while the remaining pixels outside the minimum bounding rectangle 53 are in an inactive state.

[0107] Sub-step (4) The X-ray tube emits X-rays and begins exposure. At this time, only the pixels within the minimum bounding rectangle 53 detect X-rays and form medical image information within the minimum bounding rectangle 53.

[0108] Sub-step (5): When the received dose in the pixel area reaches the preset dose, the flat panel detector reports to the control host, and the control host controls the X-ray tube to stop emitting X-rays.

[0109] Therefore, for each exposure, overexposure can be avoided, preventing unnecessary radiation to patients, and precise exposure dose control is achieved.

[0110] Figure 5 This is an exemplary schematic diagram of image stitching according to an embodiment of the present invention. For example, image stitching is performed on a first X-ray image 61 generated by a first exposure and a second X-ray image 62 generated by a second exposure.

[0111] Specifically, the image stitching process includes the following sub-steps:

[0112] Sub-step (1): Perform downsampling processing on the first X-ray image 61 and the second X-ray image 62 respectively.

[0113] Sub-step (2): Perform wavelet transform on the downsampled first X-ray image 61 and the downsampled second X-ray image 62 respectively.

[0114] Sub-step (3): Segment (e.g., using a U-Net network) the foreground and background of the first X-ray image 61 and the second X-ray image 62 by wavelet transform.

[0115] Sub-step (4): Perform foreground stitching and background stitching respectively.

[0116] In background stitching: In the background 63 of the first X-ray image 61, a first contour line within a first selected region 65 is determined using the Canny operator. In the background 64 of the second X-ray image 62, a second contour line within a second selected region 66 is determined using the Canny operator. Then, the first and second contour lines are stitched together to obtain the stitched background.

[0117] In foreground stitching: stitch the foreground within the first selected area 65 and the foreground within the second selected area 66 to obtain the stitched foreground.

[0118] Sub-step (5): Fuse the stitched background and stitched foreground to generate a stitched image of the first X-ray image 61 and the second X-ray image 62.

[0119] Therefore, by separately stitching and then merging the foreground and background, the quality of the stitched image is improved.

[0120] Figure 6 This is a schematic diagram of the image stitching process in X-ray photography according to an embodiment of the present invention. Figure 6As shown, the image stitching process includes: generating a skeletal model 72 of the photographic object based on the visible light image 70 and depth information 71 of the photographic object; generating a three-dimensional model 73 of the photographic object based on the rendering processing of the skeletal model 72; determining the stitching region 74 by selection operations on the three-dimensional model 73; in motion control 75, driving the X-ray tube and the flat panel detector to move synchronously within a range of motion corresponding to the stitching region 74, wherein information provided by the angle sensor 76 and the position encoder 77 is used to confirm the accuracy of the synchronous movement; in the exposure process 78 of the synchronous movement, multiple exposures are performed, wherein each exposure is precisely controlled by controlling the pixel activation state in the flat panel detector; in the stitching process 79, the foreground and background are stitched and then fused separately to obtain the stitched image 80.

[0121] Figure 7 This is an exemplary structural diagram of an image stitching apparatus for X-ray photography according to an embodiment of the present invention. Figure 7 As shown, the X-ray imaging stitching device 700 includes: a determining module 701 for determining a three-dimensional model of the photographic object based on sensor data of the photographic object; a selecting module 702 for selecting a stitching area in the three-dimensional model; an exposure module 703 for performing multiple exposures on the photographic object based on the stitching area to generate multiple X-ray images; and a stitching module 704 for stitching the multiple X-ray images.

[0122] In one embodiment, the exposure module 703 is used to select multiple stitching sub-regions for multiple exposures from the stitching region in the three-dimensional model; determine multiple pixel regions in the flat panel detector that surround each stitching sub-region and correspond to the multiple exposures; and control the pixels in the corresponding pixel regions to be in an active state and the pixels outside the pixel regions to be in an inactive state during each exposure.

[0123] In one embodiment, the exposure module 703 is used to determine the corresponding region of each stitched sub-region in the flat panel detector; determine the minimum bounding rectangle of the corresponding region; and determine the multiple minimum bounding rectangles corresponding to multiple stitched sub-regions as multiple pixel regions.

[0124] In one embodiment, the stitching module 704 is used to segment each of the multiple X-ray images into a foreground and a background; determine the contour lines within a selected region of each X-ray image in the background; stitch together the contour lines of the multiple X-ray images to form a stitched background; detect the foreground of each X-ray image within its selected region to form a stitched foreground; and fuse the stitched background and the stitched foreground to generate a stitched image of the multiple X-ray images.

[0125] The present invention also proposes a control host for an X-ray imaging system with a processor-memory architecture. Figure 8 This is an exemplary structural diagram of the control host of an X-ray imaging system according to an embodiment of the present invention. Figure 8 As shown, the control host 800 includes a processor 801, a memory 802, and a computer program stored in the memory 802 and executable on the processor 801. When executed by the processor 801, the computer program implements any of the above-described X-ray imaging image stitching methods. Specifically, the memory 802 can be implemented as various storage media such as electrically erasable programmable read-only memory (EEPROM), flash memory, and programmable programmable read-only memory (PROM). The processor 801 can be implemented as including one or more central processing units (CPUs) or one or more field-programmable gate arrays (FPGAs), wherein the FPGA integrates one or more CPU cores. Specifically, the CPU or CPU core can be implemented as a CPU, MCU, or DSP, etc.

[0126] It should be noted that not all steps and modules in the above processes and structural diagrams are mandatory; some steps or modules can be omitted as needed. The execution order of the steps is not fixed and can be adjusted as required. The division of modules is merely for the convenience of description and functional division. In actual implementation, a module can be implemented by multiple modules, and the functions of multiple modules can also be implemented by the same module. These modules can be located in the same device or in different devices.

[0127] The hardware modules in each embodiment can be implemented mechanically or electronically. For example, a hardware module may include specially designed permanent circuitry or logic devices (such as dedicated processors, such as FPGAs or ASICs) to perform specific operations. A hardware module may also include programmable logic devices or circuitry (such as general-purpose processors or other programmable processors) temporarily configured by software to perform specific operations. The choice between mechanical implementation, dedicated permanent circuitry, or temporarily configured circuitry (such as software-configured circuitry) can be made based on cost and time considerations.

[0128] The present invention also provides a machine-readable storage medium storing instructions for causing a machine to perform the methods described in this application. Specifically, a system or apparatus equipped with a storage medium storing software program code that implements the functions of any of the embodiments described above, and causing a computer (e.g., CPU, MCU, or MPU) of the system or apparatus to read and execute the program code stored in the storage medium. Furthermore, an operating system or similar device operating on a computer can perform some or all of the actual operations through instructions based on the program code. The program code read from the storage medium can also be written to a memory located in an expansion board inserted into a computer or to a memory located in an expansion unit connected to the computer. Subsequently, a control unit or similar device installed on the expansion board or expansion unit can perform some or all of the actual operations based on the instructions in the program code, thereby implementing the functions of any of the embodiments described above. Storage medium embodiments for providing program code include floppy disks, hard disks, magneto-optical disks, optical disks (such as CD-ROM, CD-R, CD-RW, DVD-ROM, DVD-RAM, DVD-RW, DVD+RW), magnetic tapes, non-volatile memory cards, and ROMs. Alternatively, program code can be downloaded from a server computer or the cloud via a communication network.

[0129] The above description is merely a preferred embodiment of the present invention and is not intended to limit the scope of protection of the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.

Claims

1. An image stitching method for X-ray photography, characterized by, comprising: determining a three-dimensional model of the photographic object based on sensor data of the photographic object (101); selecting a stitching region in the three-dimensional model (102); performing multiple exposures on the photographic object based on the stitching region to generate multiple X-ray images (103); stitching the multiple X-ray images (104).

2. The method of claim 1, wherein, The determining a three-dimensional model of the photographic object based on sensor data of the photographic object (101) comprises: inputting the sensor data into a trained artificial intelligence model, the artificial intelligence model being adapted to determine a skeletal model of the photographic object based on the sensor data; extracting morphological parameters of the photographic object from the sensor data; obtaining a three-dimensional model of the photographic object in a rendering manner based on the morphological parameters and the skeletal model.

3. The method of claim 2, wherein, The method comprises a training process of the artificial intelligence model; wherein the training process comprises: inputting a training sample comprising training sensor data and a label into the artificial intelligence model, wherein the label comprises a skeletal model of the training sensor data; receiving a predicted skeletal model based on the training sensor data from the artificial intelligence model; determining a loss function value of the artificial intelligence model based on a difference between the predicted skeletal model and the label; configuring model parameters of the artificial intelligence model to make the loss function value lower than a preset threshold.

4. The method of claim 1, wherein The performing multiple exposures on the photographic object to generate multiple X-ray images (103) comprises: selecting multiple stitching sub-regions for multiple exposures from the stitching region in the three-dimensional model; determining multiple pixel regions in a flat panel detector respectively surrounding each stitching sub-region corresponding to multiple exposures; controlling pixels in the corresponding pixel region to be in an active state and pixels other than the pixel region to be in an inactive state in each exposure.

5. The method of claim 4, wherein, The determining multiple pixel regions in a flat panel detector respectively surrounding each stitching sub-region corresponding to multiple exposures comprises: determining a corresponding region of each stitching sub-region in the flat panel detector; determining a minimum bounding rectangle of the corresponding region; determining multiple minimum bounding rectangles corresponding to multiple stitching sub-regions as the multiple pixel regions.

6. The method according to any one of claims 1-5, characterized in that, The stitching the multiple X-ray images (104) comprises: segmenting each X-ray image in the multiple X-ray images into foreground and background; determining contour lines in a respective selected region in the background of each X-ray image; stitching multiple contour lines of the multiple X-ray images as a stitched background; detecting the foreground of each X-ray image in the respective selected region as a stitched foreground; fusing the stitched background and the stitched foreground to generate a stitched image of the multiple X-ray images.

7. An image stitching apparatus for X-ray photography, characterized by comprising: a determining module (701) configured to determine a three-dimensional model of the photographic object based on sensor data of the photographic object; a selecting module (702) configured to select a stitching region in the three-dimensional model; an exposure module (703) configured to perform multiple exposures on the photographic object based on the stitching region to generate multiple X-ray images. The splicing module (704) is configured to splice the multiple X-ray images.

8. The apparatus of claim 7, wherein, The exposure module (703) is configured to select multiple splicing sub-regions of multiple exposures from the splicing region in the three-dimensional model, and determine multiple pixel regions in the flat panel detector that respectively surround each splicing sub-region and correspond to the multiple exposures. In each exposure, the pixels in the corresponding pixel region are in an active state, and the pixels other than the pixel region are in an inactive state.

9. The apparatus of claim 8, wherein, The exposure module (703) is configured to determine the corresponding region of each splicing sub-region in the flat panel detector, determine the minimum circumscribed rectangle of the corresponding region, and determine multiple minimum circumscribed rectangles corresponding to multiple splicing sub-regions as the multiple pixel regions.

10. The apparatus of any one of claims 7-9, wherein, The splicing module (704) is configured to divide each X-ray image in the multiple X-ray images into foreground and background, determine the contour line in the respective selected region in the background of each X-ray image, splice the multiple contour lines of the multiple X-ray images as the spliced background, detect the foreground of each X-ray image in the respective selected region as the spliced foreground, and fuse the spliced background and the spliced foreground to generate the spliced image of the multiple X-ray images. comprising: a processor (801); a memory (802) configured to store executable instructions of the processor (801); 11. A control host of an X-ray imaging system, characterized in that the processor (801) is configured to read the executable instructions from the memory (802) and execute the executable instructions to implement the image splicing method of X-ray photography according to any one of claims 1-6. the computer instructions, when executed by the processor, implement the image splicing method of X-ray photography according to any one of claims 1-6. a computer program, when executed by the processor, implements the image splicing method of X-ray photography according to any one of claims 1-6. ​ 12. A computer readable storage medium having stored thereon computer instructions, wherein, ​ 13. A computer program product, characterised in that, ​

Images