Image calibration method, image sensor, and storage medium

Abstract

The application relates to the field of radiation detection imaging technology and discloses an image calibration method, an image sensor and a storage medium, wherein the method comprises the following steps: using at least one radiation detector to shoot a plurality of parts of a calibration pattern at a plurality of shooting positions to obtain a plurality of partial images; based on each partial image, determining a calibration data set and a splicing position data set of the radiation detector for obtaining each partial image, which are related to the shooting position of the radiation detector when the radiation detector obtains the partial image, wherein the calibration data set can be used to calibrate the partial image obtained by the radiation detector when the radiation detector is located at the shooting position into a partial calibration image, and the splicing position data set can be used to determine the relative position of the partial calibration image relative to other partial calibration images when the partial calibration image is spliced with the other partial calibration images. The method of the application simplifies the operation during calibration, reduces the difficulty of a user in obtaining the calibration data set and the splicing position data set, and improves the calibration efficiency.

Description

Technical Field

[0001] This application relates to the field of radiation detection imaging technology, and in particular to an image calibration method, an image sensor, and a storage medium. Background Technology

[0002] A radiation detector is a device used to measure the flux, spatial distribution, spectrum, or other properties of radiation. One important application of radiation detectors is imaging. Radiation imaging is a radiographic technique used to reveal the internal structure of non-homogeneous and opaque objects (e.g., the human body).

[0003] In an image sensor with multiple radiation detectors, these detectors capture multiple local images of an object from multiple shooting positions, obtaining multiple local images. These local images are then stitched together to obtain a complete image of the object. To obtain an accurate complete image, the spatial positional relationships of the multiple radiation detectors need to be calculated beforehand, and then the stitching of the multiple local images needs to be calculated based on these relationships. Due to potential insufficient machining precision at the installation locations of the radiation detectors, and sometimes the detection surfaces of these detectors are tilted and arranged in a dome shape, the overall image after stitching may have a high distortion rate.

[0004] One method to reduce distortion is to capture multiple local images of an object using multiple radiometers at multiple shooting locations. After obtaining these local images, each local image is calibrated, and then the calibrated local images are stitched together. This eliminates the need to precisely measure the spatial relationships between the multiple radiometers, but requires knowledge of the calibration dataset for each radiometer relative to its shooting location.

[0005] In existing technologies, one method for obtaining a calibration dataset involves photographing a reference object with multiple identical icons arranged in two-dimensional space to obtain images of the icons, and then calibrating these images. During photography, each radiation detector captures exactly one of these identical icons, and then compares the image of that icon with its calibration reference image to obtain the calibration dataset for that radiation detector relative to its shooting position. In this approach, to accurately calculate the calibration dataset, the reference object needs to be precisely aligned relative to each radiation detector to ensure that each detector captures exactly one of the multiple identical icons, allowing for comparison with its calibration reference image. This results in a highly tedious and complex operation where operators must repeatedly image the reference object and adjust its alignment, requiring a high level of skill. Furthermore, since all icons captured by all radiation detectors are identical, the position of each icon captured by each detector within the reference object cannot be determined, making it difficult to conveniently determine the stitching position dataset for each radiation detector relative to its shooting position. Summary of the Invention

[0006] The embodiments of this application aim to provide an image calibration method, apparatus and storage medium to solve the problem that in the prior art, when calibrating an image sensor with multiple radiation detectors, the operator needs to make multiple adjustments and placements, resulting in cumbersome and time-consuming operations.

[0007] To address the aforementioned technical problems, this application provides the following technical solutions:

[0008] According to a first aspect of this application, an image calibration method is provided, the method comprising:

[0009] Using at least one radiation detector, multiple local images of a calibration pattern are captured at multiple shooting positions using radiation from a radiation source, wherein each local image of the calibration pattern is unique within the calibration pattern; and

[0010] Based on each of the plurality of local images, a calibration dataset and a stitching position dataset related to the shooting position of the radiation detector at the time the local image was obtained are determined for each local image. The calibration dataset related to the shooting position of the radiation detector can be used to calibrate the local image obtained by the radiation detector at the shooting position into a local calibration image. The stitching position dataset related to the shooting position of the radiation detector can be used to determine the relative position of the local calibration image with respect to one or more other local calibration images calibrated from one or more other local images when stitched together.

[0011] Optionally, the method further includes:

[0012] Using the at least one radiation detector, multiple local scenes are captured at the multiple shooting locations using the radiation, thereby obtaining multiple local scene images;

[0013] For each of the plurality of local scene images, the calibration dataset and the stitching position dataset related to the shooting position of the radiation detector at the time the local scene image was obtained are respectively used as the calibration dataset and the stitching position dataset related to the local scene image; and

[0014] Using multiple calibration datasets associated with the multiple local scene images, the multiple local scene images are calibrated into multiple local scene calibration images. Using multiple stitching position datasets associated with the multiple local scene images, the multiple local scene calibration images are stitched together to obtain the stitched calibration image of the scene.

[0015] Optionally, the calibration pattern is formed by arranging at least two types of icons in two mutually orthogonal directions in a two-dimensional space. The arrangement rule of the icons ensures that the m×n feature icon array at any position in the calibration pattern is unique within the calibration pattern, where m and n are positive integers.

[0016] The area of ​​each of the plurality of local parts of the calibration pattern is greater than the area of ​​the feature icon array.

[0017] Optionally, the area of ​​each of the plurality of localities of the calibration pattern is not less than the area of ​​the icon array of size (m+1)×(n+1).

[0018] Optionally, the calibration pattern is formed by arranging the at least two icons according to a two-dimensional de Bruin sequence.

[0019] Optionally, the calibration pattern consists of four icons arranged in a two-dimensional de Bruin sequence, and the feature icon array is 3×3 in size.

[0020] Optionally, the step of determining, based on each of the plurality of local images, a calibration dataset and a stitching location dataset related to the shooting position of the radiation detector at the time the local image was obtained, for each of the local images, includes:

[0021] Identify the original image of the feature icon array contained in each local image, and determine the shape and arrangement of the icons in the feature icon array based on the shape and arrangement of the icon images in the original image of the feature icon array;

[0022] A calibration reference image of the feature icon array is generated based on the shape and arrangement of the icons in the feature icon array;

[0023] Based on the original image of the feature icon array and the calibration reference image of the feature icon array, a transformation dataset is determined that can be used to transform the original image of the feature icon array into the calibration reference image of the feature icon array. The transformation dataset is used as the calibration dataset of the radiation detector that obtained the local image, which is related to the shooting position of the radiation detector when the local image was obtained.

[0024] Using the calibration dataset of the radiation detector related to the shooting location, the local image is calibrated into the local calibration image, wherein the original image of the feature icon array contained in the local image is calibrated into the calibration image of the feature icon array contained in the local calibration image;

[0025] Generate a global reference image of the calibration pattern; and

[0026] Based on the position of the calibration image of the feature icon array and / or the calibration reference image of the feature icon array in the global reference image of the calibration pattern and the local calibration image, a stitching position dataset related to the shooting position of the radiation detector at the time of obtaining the local image is determined, wherein the stitching position dataset represents the position of the local calibration image in the global reference image of the calibration pattern.

[0027] Optionally, the calibration reference image of the feature icon array is an image generated by simulating the projection of the feature icon array onto an image plane, and the global reference image of the calibration pattern is an image generated by simulating the projection of the calibration pattern onto the image plane, wherein the image plane is parallel to the surface of the calibration pattern facing the radiation source.

[0028] Optionally, before determining the stitching position dataset of the radiation detector that obtained the local image, which is related to the shooting position of the radiation detector when the local image was obtained, based on the position of the calibration image of the feature icon array and / or the calibration reference image of the feature icon array in the global reference image of the calibration pattern and the local calibration image,

[0029] The position of the feature icon array within the calibration pattern is determined based on the shape and arrangement of the icons in the feature icon array.

[0030] Based on the position of the feature icon array in the calibration pattern, determine the position of the calibration image of the feature icon array and / or the calibration reference image of the feature icon array in the global reference image of the calibration pattern.

[0031] Optionally, the stitching position dataset of the radiation detector related to the imaging position includes a first stitching position data subset and a second stitching position data subset. The first stitching position data subset represents the position of the calibration image of the feature icon array and / or the calibration reference image of the feature icon array in the global reference image of the calibration pattern. The second stitching position data subset represents the position of the portion of the local calibration image other than the calibration image of the included feature icon array in the global reference image of the calibration pattern.

[0032] The step of determining the stitching position dataset, which is related to the shooting position of the radiation detector at the time of obtaining the local image, based on the position of the calibration image of the feature icon array and / or the calibration reference image of the feature icon array in the global reference image of the calibration pattern and the local calibration image, includes:

[0033] The first splicing position data subset is determined based on the position of the calibration image of the feature icon array and / or the calibration reference image of the feature icon array in the global reference image of the calibration pattern;

[0034] Based on the position of the calibration image of the feature icon array within the local calibration image and the positions of the other portions of the local calibration image within the local calibration image, determine the relative positional relationship between the other portions of the local calibration image and the calibration image of the feature icon array; and

[0035] A second splicing position data subset is determined based on the position of the calibration image of the feature icon array and / or the calibration reference image of the feature icon array in the global reference image of the calibration pattern, as well as the relative positional relationship between the other parts of the local calibration image and the calibration image of the feature icon array.

[0036] Optionally, after calibrating the local image into the local calibration image using the calibration dataset, it is determined whether the calibration image of the feature icon array is substantially the same as the calibration reference image of the feature icon array.

[0037] Optionally, multiple radiation detectors are used, and each of the multiple radiation detectors remains fixed relative to the calibration pattern.

[0038] Alternatively, only one radiation detector may be used, and the radiation detector may be moved relative to the calibration pattern to the plurality of shooting positions.

[0039] Optionally, multiple radiation detectors are used, and at least one of the multiple radiation detectors is movable relative to the calibration pattern to the multiple shooting positions.

[0040] According to a second aspect of this application, an image sensor is provided, including at least one radiation detector, the image sensor being configured to perform the steps of the image calibration method described in any of the preceding claims.

[0041] According to a third aspect of this application, a computer-readable storage medium is provided, the computer-readable storage medium storing a computer program, which, when executed by a processor, performs the steps of the image calibration method described in any of the preceding claims.

[0042] The beneficial effects of the embodiments of this application are as follows: Unlike the prior art, the image calibration method provided in this application uses at least one radiation detector to capture multiple localities of a calibration pattern at multiple shooting positions, obtaining multiple local images. After obtaining multiple local images, since each locality of the calibration pattern is unique within the calibration pattern, based on each local image, the calibration dataset and stitching position dataset related to the shooting position of the radiation detector at the time of obtaining the local image can be determined. The image calibration method of this application does not require prior calculation of the relative spatial positions of multiple radiation detectors, nor does it require ensuring the precise alignment of the calibration pattern with respect to each radiation detector. It only requires one imaging operation after placing the calibration pattern to automatically analyze and calculate the calibration dataset and stitching position dataset related to the shooting position of each radiation detector. Therefore, it simplifies the calibration operation, reduces the difficulty for users to obtain the calibration dataset and stitching position dataset, and improves calibration efficiency. Attached Figure Description

[0043] One or more embodiments are illustrated by way of example with reference numerals in the accompanying drawings. These illustrations do not constitute a limitation on the embodiments. Elements with the same reference numerals in the drawings are denoted as similar elements. Unless otherwise stated, the figures in the drawings are not to be limited by scale.

[0044] Figure 1A A cross-sectional view of an image sensor, a radiation source, and a scene according to an embodiment is schematically shown;

[0045] Figure 1B A perspective view schematically illustrating an image sensor, radiation source, and calibration components according to an embodiment is shown.

[0046] Figure 1C Another perspective view schematically illustrates the image sensor, radiation source, and calibration components according to an embodiment;

[0047] Figure 2 The image sensor schematically illustrates multiple parts of a scene captured by an embodiment;

[0048] Figure 3A and Figure 3B Each schematically illustrates the movement of the radiation detector of the image sensor relative to the radiation source according to an embodiment;

[0049] Figures 4A-4C The arrangement of the radiation detector in the image sensor according to an embodiment is schematically shown;

[0050] Figure 5 An image sensor with multiple hexagonal radiation detectors according to an embodiment is schematically shown;

[0051] Figure 6 A radiation detector with a pixel array according to an embodiment is schematically shown;

[0052] Figure 7A A cross-sectional view of a radiation detector according to an embodiment is schematically shown;

[0053] Figure 7B A detailed cross-sectional view of a radiation detector according to an embodiment is schematically shown;

[0054] Figure 7C An alternative detailed cross-sectional view of the radiation detector according to an embodiment is schematically shown;

[0055] Figure 8 A calibration pattern according to an embodiment is schematically shown;

[0056] Figure 9 A flowchart illustrating an image calibration method according to an embodiment is shown schematically.

[0057] Figure 10 for Figure 9 A detailed flowchart of step S902 in the process;

[0058] Figure 11 The process of determining the calibration dataset and stitching location dataset based on local images is illustrated schematically.

[0059] Figure 12 A medical imaging system including an image sensor is schematically illustrated according to an embodiment;

[0060] Figure 13 A system for dental radiography, including an image sensor, is schematically illustrated according to an embodiment.

[0061] Figure 14 A cargo scanning or non-invasive inspection (NII) system including an image sensor is schematically illustrated according to an embodiment;

[0062] Figure 15 Another cargo scanning or non-invasive inspection (NII) system including an image sensor is schematically illustrated according to an embodiment;

[0063] Figure 16 A full-body scanning system including an image sensor is schematically illustrated according to an embodiment;

[0064] Figure 17 A radiation computed tomography (radiation CT) system including an image sensor is schematically illustrated according to an embodiment;

[0065] Figure 18An electron microscope including an image sensor is schematically shown according to an embodiment. Detailed Implementation

[0066] To make the objectives, technical solutions, and advantages of the embodiments of this application clearer, the technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0067] Figure 1A A cross-sectional view of an image sensor 9000, a radiation source 109, and a scene 50 according to an embodiment is schematically shown. The image sensor 9000 includes a plurality of radiation detectors 100 (e.g., Figure 1A The first radiation detector 100A and the second radiation detector 100B shown in the figure Figure 1A (Radiation detectors 100C, 100D, etc., not shown). The image sensor 9000 may include a bracket 107 having a curved surface 102 (or a bracket 107 having a horizontal surface). Figure 1A As shown, multiple radiation detectors 100 can be arranged on a support 107, for example, on a curved surface 102. A first radiation detector 100A may have a first flat surface 103A configured to receive radiation from a radiation source 109. A second radiation detector 100B may have a second flat surface 103B configured to receive radiation from a radiation source 109. The first flat surface 103A and the second flat surface 103B may not be coplanar or parallel.

[0068] Radiation source 109 is a radiation source that emits a cone-shaped radiation beam. In some embodiments, radiation source 109 may be a radiation source that emits radiation beams of other shapes, such as a linear radiation beam, a fan-shaped radiation beam, a parallel radiation beam, etc.

[0069] Radiation source 109 is located outside scene 50. Radiation from radiation source 109 may pass through, before reaching at least one of the flat surfaces 103A of the first radiation detector 100A and 103B of the second radiation detector 100B. Figure 1A The scenario shown in scene 50 (e.g., a part of the human body) or as... Figure 1B The calibration component 101 is shown. In some embodiments, scene 50 or calibration component 101 is located between radiation source 109 and image sensor 9000. In some embodiments, the relative position of scene 50 or calibration component 101 with respect to radiation source 109 remains unchanged.

[0070] The calibration component 101 can be plate-shaped. The calibration component 101 can have a spatially uneven distribution of radiation absorption. Figure 1B In the example, calibration component 101 includes a first region 161 located at the center of calibration component 101 and a second region 162 located around the periphery of the first region 161. The first region 161 includes, for example, Figure 8 The calibration pattern shown has a second region 162 for securing the calibration component 101 in the appropriate position, for example, between the radiation source 109 and the image sensor 9000.

[0071] In some embodiments, the calibration pattern consists of at least two types of icons arranged in two mutually orthogonal directions in two-dimensional space, with an arrangement rule that makes the m×n feature icon array at any position in the calibration pattern unique within the calibration pattern, where m and n are positive integers. In this document, the m×n size means that each row of the array consists of m elements and each column consists of n elements, or that each column consists of m elements and each row consists of n elements. Furthermore, the area of ​​the local portion of the calibration pattern captured by each radiation detector is larger than the area of ​​the feature icon array. Therefore, the local portion of the calibration pattern captured by each radiation detector is unique within the calibration pattern. In some embodiments, the area of ​​the local portion of the calibration pattern captured by each radiation detector is not less than the area of ​​the (m+1)×(n+1) size icon array to ensure that the entirety of at least one m×n size feature icon array can be captured. In this document, the area of ​​the icon array or feature icon array includes not only the area of ​​all icons in the icon array or feature icon array but also the area of ​​the gaps between these icons.

[0072] In some embodiments, the calibration pattern consists of at least two icons arranged according to a two-dimensional de Bruijn sequence. A de Bruijn sequence, also known as a complete cycle, is a special type of combinatorial sequence that can be constructed using methods such as Hamiltonian circuits and PCR (Pure Cycling Register).

[0073] Specifically, a one-dimensional de Brouin sequence is a one-dimensional cyclic sequence consisting of k elements. All one-dimensional sequences of length h can be contained within this one-dimensional de Brouin sequence and appear only once. This one-dimensional de Brouin sequence is denoted as B(k,h). For example, one permutation of the one-dimensional de Brouin sequence B(2,4), consisting of the digits 0 and 1 and containing all one-dimensional sequences of length 4, is 0000101101001111. The permutations of de Brouin sequences are usually not unique; for example, B(2,4) can also be permuted as 0000111101001011.

[0074] A two-dimensional De Bruijn sequence is obtained by extending a one-dimensional De Bruijn sequence into a two-dimensional space. It is a two-dimensional sequence of size x×y composed of k elements, and any two-dimensional array of size w×h (w < x, h < y) appears only once. For example, in the following 4×6 two-dimensional De Bruijn sequence composed of the two digits 0 and 1, each 2×2 two-dimensional array is unique.

[0075]

[0076] The icons constituting the calibration pattern can be geometric figures or other graphic symbols with referential meanings, such as letters, mathematical symbols, etc. As Figure 8 shown, the icons of the calibration pattern are geometric figures of 4 different shapes (triangle, circle, square, X shape), and the 3×3 size feature icon array at any position in the calibration pattern is unique in the calibration pattern. It can be understood that those skilled in the art can also adopt other icons or other values of m and n and construct them according to the construction method of the two-dimensional De Bruijn sequence to obtain other calibration patterns.

[0077] Figure 1C A perspective view schematically shows an image sensor 9000, a radiation source 109, and a calibration component 101 according to an embodiment. The image sensor 9000 uses radiation from the radiation source 109 to capture multiple local parts of the calibration pattern at multiple shooting positions. In Figure 1C the example, the radiation incident on the calibration pattern in the first region 161 of the calibration component 101 can pass through the calibration pattern and be received by radiation detectors (such as 100A, 100B, Figure 1C 100C, 100D, etc. not shown in

[0078] As Figure 1CAs shown, the first partial image 104A contains an original image of a 4×4 icon array. If the feature icon array in the calibration pattern is 3×3, then the first partial image 104A contains an original image of at least one feature icon array. Since the first partial image 104A is distorted, the original image of the feature icon array is also distorted. However, based on the original image of the feature icon array, the shape and arrangement of the icons in the feature icon array can be determined. In some embodiments, multiple icon images in the first partial image 104A are first identified. Then, based on the shape and arrangement of these icon images, the original image of the feature icon array can be determined. Furthermore, based on the shape of the icon images in the original image of the feature icon array, the shape of the icons in the feature icon array can be determined. And based on the arrangement of the icon images in the original image of the feature icon array, the arrangement of the icons in the feature icon array can be determined.

[0079] Based on the shape and arrangement of the icons in the feature icon array and the original image of the feature icon array, the calibration dataset and stitching location dataset related to the first imaging position of the first radiation detector 100A can be determined.

[0080] Specifically, a calibration reference image for the feature icon array can be generated based on the shape and arrangement of the icons in the feature icon array. In some embodiments, the calibration reference image for the feature icon array is an image generated by simulating the projection of the feature icon array onto an image plane parallel to the surface of the calibration pattern facing the radiation source 109. Furthermore, based on the original image of the feature icon array and the calibration reference image of the feature icon array, for example using an image alignment algorithm, a transformation dataset that can be used to transform the original image of the feature icon array into the calibration reference image of the feature icon array can be determined, namely, the calibration dataset of the first radiation detector 100A related to the first shooting position.

[0081] Using the calibration dataset, the first local image 104A (the original image containing the feature icon array) can be calibrated to a first local calibration image. Clearly, while calibrating the first local image 104A to the first local calibration image, the original image of the feature icon array contained in the first local image 104A is also calibrated to a calibration image of the feature icon array; that is, the first local calibration image contains a calibration image of the feature icon array. In some embodiments, the calibration image of the feature icon array is substantially the same as the calibration reference image of the feature icon array.

[0082] Since the feature icon array is unique within the calibration pattern, its position within the calibration pattern can be determined based on the shape and arrangement of the icons. Furthermore, based on the position of the feature icon array within the calibration pattern, the position of its calibration image and / or calibration reference image within the global reference image of the calibration pattern can be determined. In some embodiments, the global reference image of the calibration pattern is an image generated by simulating the projection of the calibration pattern onto an image plane.

[0083] Then, based on the positions of the calibration image of the feature icon array and / or the calibration reference image of the feature icon array in the global reference image of the calibration pattern, and the first local calibration image, a stitching position dataset related to the first imaging position of the first radiation detector 100A can be determined. This stitching position dataset represents the position of the first local calibration image in the global reference image of the calibration pattern. Specifically, the stitching position dataset includes a first stitching position data subset and a second stitching position data subset. The first stitching position data subset represents the position of the calibration image of the feature icon array and / or the calibration reference image of the feature icon array in the global reference image of the calibration pattern, and the second stitching position data subset represents the position of the portion of the local calibration image other than the included calibration image of the feature icon array in the global reference image of the calibration pattern. The first stitching position data subset can be determined based on the positions of the calibration image of the feature icon array and / or the calibration reference image of the feature icon array in the global reference image of the calibration pattern. Then, based on the position of the calibration image of the feature icon array in the local calibration image and the positions of other parts of the local calibration image in the local calibration image, the relative positional relationship between the other parts of the local calibration image and the calibration image of the feature icon array can be determined. Next, based on the position of the calibration image of the feature icon array and / or the calibration reference image of the feature icon array in the global reference image of the calibration pattern and the relative positional relationship between the other parts of the local calibration image and the calibration image of the feature icon array, the second stitching position data subset can be determined.

[0084] The above, using the first radiation detector 100A as an example, illustrates how to determine the calibration dataset and stitching position dataset related to the first imaging position of the first radiation detector 100A. Similarly, the calibration dataset and stitching position dataset related to the second imaging position of the second radiation detector 100B, as well as multiple calibration datasets and multiple stitching position datasets related to multiple other imaging positions for multiple other radiation detectors (e.g., radiation detectors 100C, 100D, etc.), can also be determined.

[0085] When determining the calibration dataset and stitching position dataset of each radiation detector 100 in relation to its imaging position as described above, it is not necessary to calculate the relative positions of multiple radiation detectors 100 in space in advance, nor is it necessary to ensure the precise alignment of the calibration pattern with respect to each radiation detector 100. Only one imaging is required after placing the calibration pattern, and the calibration dataset and stitching position dataset of each radiation detector 100 in relation to its imaging position can be automatically analyzed and calculated. Therefore, the calibration operation is simplified, the difficulty for users to obtain the calibration dataset and stitching position dataset is reduced, and the calibration efficiency is improved.

[0086] After determining the calibration dataset and stitching location dataset associated with each radiation detector and its respective shooting position, the multiple radiation detectors 100 of the image sensor 9000 use radiation from the radiation source 109 to capture multiple local scenes of scene 50 at multiple shooting positions, obtaining multiple local scene images. For each local scene image, the calibration dataset and stitching location dataset associated with the shooting position of the radiation detector that acquired the local scene image are respectively used as the calibration dataset and stitching location dataset associated with that local scene image. Using the multiple calibration datasets associated with the multiple local scene images obtained above, the multiple local scene images are calibrated into multiple local scene calibration images. Then, using the multiple stitching location datasets associated with the multiple local scene images, the multiple local scene calibration images are stitched together to obtain a stitched calibration image of scene 50. Specifically, the first radiation detector 100A of the image sensor 9000 captures a first portion of scene 50 at a first shooting position, obtaining a first partial scene image. The second radiation detector 100B captures a second portion of scene 50 at a second shooting position, obtaining a second partial scene image. Multiple other radiation detectors capture multiple other portions of scene 50 at multiple other shooting positions, obtaining multiple other partial scene images. The calibration dataset and stitching position dataset related to the first shooting position of the first radiation detector 100A are used as the calibration dataset and stitching position dataset related to the first partial scene image. The calibration dataset and stitching position dataset related to the second shooting position of the second radiation detector 100B are used as the calibration dataset and stitching position dataset related to the second partial scene image. The multiple calibration datasets and stitching position datasets related to the multiple other shooting positions of the multiple other radiation detectors are used as multiple calibration datasets and stitching position datasets related to the multiple other partial scene images. Image sensor 9000 calibrates the first local scene image into a first local scene calibration image using a calibration dataset associated with the first local scene image, calibrates the second local scene image into a second local scene calibration image using a calibration dataset associated with the second local scene image, and calibrates multiple other local scene images into multiple other local scene calibration images using multiple calibration datasets associated with multiple other local scene images. Image sensor 9000 then stitches the first local scene calibration image, the second local scene calibration image, and the multiple other local scene calibration images using a stitching position dataset associated with the first local scene image, a stitching position dataset associated with the second local scene image, and multiple stitching position datasets associated with multiple other local scene images, respectively, to obtain a stitched calibration image of scene 50.

[0087] Figure 2The image sensor 9000 is schematically shown to be capable of capturing multiple portions of scene 50. The image sensor 9000 includes multiple (in...) Figure 2 There are eight radiation detectors 100 in the image sensor 9000. The image sensor 9000 can be moved relative to the scene 50 to three positions A, B, and C via the actuator 500. The image sensor 9000 captures three parts of the scene 50 at positions A, B, and C, respectively, obtaining a first partial scene image 51A, a second partial scene image 51B, and a third partial scene image 51C. The first partial scene image 51A, the second partial scene image 51B, and the third partial scene image 51C each include multiple partial scene images obtained by the multiple radiation detectors 100 of the image sensor 9000 capturing multiple parts of the scene 50 at multiple shooting positions when the image sensor 9000 is located at positions A, B, and C, respectively. For each partial scene image, a calibration dataset and a stitching position dataset related to the shooting position of the radiation detector 100 at the time of acquiring the partial scene image are respectively used as the calibration dataset and stitching position dataset associated with that partial scene image. Using multiple calibration datasets associated with multiple local scene images, the first scene image 51A, the second scene image 51B, and the third scene image 51C can be calibrated into a first scene calibration image, a second scene calibration image, and a third scene calibration image, respectively. Specifically, the multiple local scene images included in the first scene image 51A, the second scene image 51B, and the third scene image 51C can be calibrated into multiple local scene calibration images included in the first scene calibration image, the second scene calibration image, and the third scene calibration image, respectively. Furthermore, using multiple stitching position datasets associated with multiple local scene images, the first scene calibration image, the second scene calibration image, and the third scene calibration image can be stitched together to form a stitched calibration image of scene 50. Specifically, the multiple local scene calibration images included in the first scene calibration image, the second scene calibration image, and the third scene calibration image can be stitched together to obtain the stitched calibration image of scene 50. These first scene calibration images, second scene calibration images, and third scene calibration images can overlap each other for stitching purposes.

[0088] In some embodiments, each region of scene 50 is captured by image sensor 9000 at one of multiple locations A, B, and C. That is, when the first part of the scene calibration image, the second part of the scene calibration image, and the third part of the scene calibration image are stitched together, the resulting stitched calibration image of scene 50 can display the entire scene 50.

[0089] In some embodiments, actuator 500 may include a controller that can determine multiple locations of image sensor 9000. The controller may be configured to determine and store calibration datasets and stitching location datasets for each radiation detector 100 of image sensor 9000, respectively, associated with one or more imaging locations it can be located at or moved to. After multiple local scene images are obtained by the multiple radiation detectors 100 of image sensor 9000 capturing multiple local scenes of scene 50, the controller calibrates these local scene images and then stitches them together.

[0090] In some embodiments, the radiation detectors 100 of the image sensor 9000 are all fixedly positioned, while the scene 50 can be moved to multiple positions relative to the image sensor 9000, allowing the image sensor 9000 to capture multiple portions of the scene 50 and obtain multiple partial scene images. In other embodiments, some of the radiation detectors 100 of the image sensor 9000 can be moved to multiple shooting positions relative to the scene 50, while other radiation detectors 100 are fixedly positioned.

[0091] In the embodiments described above, where the image sensor 9000 is moved to multiple positions relative to the scene 50 by the actuator 500, or the scene 50 is moved to multiple positions relative to the image sensor 9000 by the actuator 500, or a portion of the radiation detectors 100 of the image sensor 9000 is moved to multiple shooting positions relative to the scene 50 by the actuator 500, the method for determining the calibration dataset and stitching position dataset associated with a certain shooting position that each radiation detector 100 can be located at or moved to is the same as the method described above for determining the calibration dataset and stitching position dataset associated with the first shooting position of the first radiation detector 100A using the calibration component 101.

[0092] Figure 3A and Figure 3B Each schematically illustrates an image sensor 9000 relative to calibration component 101 or scene 50 according to an embodiment. Figure 3A and Figure 3B (Movement not shown in the image). Figure 3A and Figure 3BIn the example, only the first radiation detector 100A and the second radiation detector 100B of the image sensor 9000 are shown. The first radiation detector 100A and the second radiation detector 100B can be arranged on the bracket 107. The relative position of the first radiation detector 100A with respect to the second radiation detector 100B remains the same at multiple locations. When the first radiation detector 100A moves from one of the multiple locations to another, the relative position of the first radiation detector 100A with respect to the second radiation detector 100B may, but does not need to, remain the same. The image sensor 9000 can rotate about a first axis 501 relative to the calibration component 101 or the scene 50. Figure 3A As shown, the image sensor 9000 rotates from position 503A to position 503B relative to the calibration component 101 or scene 50 about a first axis 501. The first axis 501 may be parallel to the first flat surface 103A of the first radiation detector 100A and the second flat surface 103B of the second radiation detector 100B. The radiation source 109 may be on the first axis 501. The image sensor 9000 may also rotate about a second axis 502 relative to the calibration component 101 or scene 50. The second axis 502 is different from the first axis 501. For example, the second axis 502 may be perpendicular to the first axis 501. Figure 3A As shown, the image sensor 9000 can rotate from position 503A to position 503C about the second axis 502. The radiation source 109 can be on the second axis 502.

[0093] like Figure 3B As shown, the image sensor 9000 can be translated from position 506A to position 506B relative to the calibration component 101 or scene 50 along a first direction 504. The image sensor 9000 can also be translated along a second direction 505. The second direction 505 is different from the first direction 504. For example, the second direction 505 can be perpendicular to the first direction 504. Figure 3B As shown, the image sensor 9000 can be translated from position 506A to position 506C along the second direction 505. The first direction 504 or the second direction 505 can be parallel to either the first flat surface 103A or the second flat surface 103B, or parallel to both, or not parallel to either. For example, the first direction 504 can be parallel to the first flat surface 103A but not parallel to the second flat surface 103B.

[0094] The radiation detector 100 can be arranged in the image sensor 9000 in various ways. Figure 4AAn arrangement according to an embodiment is schematically illustrated, in which radiation detectors 100 are arranged in staggered rows. For example, radiation detectors 100A and 100B are in the same row, aligned in the Y direction, and of the same size; radiation detectors 100C and 100D are in the same row, aligned in the Y direction, and of the same size. Radiation detectors 100A and 100B are staggered in the X direction relative to radiation detectors 100C and 100D. In some embodiments, the distance X2 between two adjacent radiation detectors 100A and 100B in the same row is greater than the width X1 (i.e., the X-direction dimension, i.e., the direction of row extension) of a single radiation detector in the same row and less than twice the width X1. Radiation detectors 100A and 100E are in the same column, aligned in the X direction, and of the same size; the distance Y2 between two adjacent radiation detectors 100A and 100E in the same column is less than the width Y1 (i.e., the Y-direction dimension) of a single radiation detector in the same column. This arrangement allows for... Figure 2 The scene is imaged as shown, and three parts of the scene are photographed at three locations spaced apart along the X direction. After obtaining three images, a stitched calibration image of the scene can be obtained by calibrating and stitching these three images.

[0095] Figure 4B Another arrangement according to an embodiment is schematically shown, wherein the radiation detector 100 is arranged in a rectangular grid. For example, the radiation detector 100 may include, for example, Figure 4A The radiation detectors 100A, 100B, 100E, and 100F are precisely positioned in the middle, without... Figure 4A The radiation detectors are 100C, 100D, 100G, or 100H. This arrangement allows for imaging of a scene by photographing multiple portions of the scene from six positions. For example, these six positions are three spaced apart along the X-direction and three more spaced apart along the X-direction and spaced apart from the first three positions along the Y-direction.

[0096] Other arrangements are also possible. For example, in Figure 4C In this configuration, the radiation detector 100 can cover the entire width of the image sensor 9000 in the X direction, and the distance Y2 between two adjacent radiation detectors 100 is less than the width Y1 of one radiation detector. Assuming that the width of the radiation detector in the X direction is greater than the width of the scene in the X direction, two parts of the scene are captured at two positions spaced apart along the Y direction. After obtaining two images, these two images can be calibrated and then stitched together to form a stitched calibrated image of the scene.

[0097] The radiation detector 100 described above can have any suitable size and shape. In some embodiments, at least some of the radiation detectors are rectangular in shape. In some embodiments, such as Figure 5 As shown, at least some of the radiation detectors are hexagonal in shape.

[0098] Figure 6 The diagram schematically illustrates that a radiation detector 100 according to an embodiment (e.g., a first radiation detector 100A, a second radiation detector 100B, other radiation detectors 100C, 100D, etc.) may have an array of pixels 150. The array may be a rectangular array, a cellular array, a hexagonal array, or any other suitable array. Each pixel 150 may be configured to detect radiating particles incident thereon, measure the energy of the radiating particles, or both. For example, each pixel 150 may be configured to count the number of radiating particles incident thereon and whose energy falls into multiple bins over a period of time. All pixels 150 may be configured to count the number of radiating particles incident thereon in multiple energy bins within the same time period. Each pixel 150 may have its own analog-to-digital converter (ADC) configured to digitize an analog signal representing the energy of the incident radiating particles into a digital signal. The ADC may have a resolution of 10 bits or higher. Each pixel 150 may be configured to measure its dark current, for example, before or simultaneously with each radiating particle incident thereon. Each pixel 150 can be configured to subtract the contribution of dark current from the energy of the radiating particle incident upon it. Pixels 150 can be configured to operate in parallel. For example, while one pixel 150 is measuring an incident radiating particle, another pixel 150 may be waiting for a radiating particle to arrive. Pixels 150 may, but do not necessarily, be individually addressable.

[0099] Figure 7A A cross-sectional view of a radiation detector 100 according to an embodiment is schematically shown. The radiation detector 100 may be used in an image sensor 9000. The radiation detector 100 may include a radiation absorption layer 110 and an electronic layer 120 (e.g., an ASIC) for processing or analyzing electrical signals generated in the radiation absorption layer 110 by incident radiation. In some embodiments, the radiation detector 100 does not include a scintillator. The radiation absorption layer 110 may include a semiconductor material, such as silicon, germanium, GaAs, CdTe, CdZnTe, or combinations thereof. The semiconductor may have a high mass attenuation coefficient for the radiation energy of interest. A surface 103 of the radiation absorption layer 110, remote from the electronic layer 120, is configured to receive radiation.

[0100] like Figure 7BA detailed cross-sectional view of the radiation detector 100 is shown. According to an embodiment, the radiation absorption layer 110 may include one or more diodes (e.g., pin or pn) consisting of one or more discrete regions 114 of a first doped region 111 and a second doped region 113. The second doped region 113 may be separated from the first doped region 111 by an optional intrinsic region 112. The discrete regions 114 are separated from each other by either the first doped region 111 or the intrinsic region 112. The first doped region 111 and the second doped region 113 have opposite types of doping (e.g., the first doped region 111 is p-type and the second doped region 113 is n-type, or the first doped region 111 is n-type and the second doped region 113 is p-type). Figure 7B In the example, each discrete region 114 of the second doped region 113, together with the first doped region 111 and the optional intrinsic region 112, forms a diode. That is, in Figure 7B In the example, the radiation-absorbing layer 110 has multiple diodes, which have a first doped region 111 as a common electrode. The first doped region 111 may also have multiple discrete portions.

[0101] When a radiating particle strikes the radiation-absorbing layer 110, which includes a diode, the radiating particle can be absorbed and generate one or more charge carriers through several mechanisms. A single radiating particle can generate 10 to 100,000 charge carriers. The charge carriers can drift toward the electrodes of one of the diodes under an electric field. The electric field can be an external electric field. The electrical contact 119B can include multiple discrete portions, each of which is in electrical contact with a discrete region 114. In some embodiments, the charge carriers can drift in multiple directions such that the charge carriers generated by a single radiating particle are substantially not shared by two different discrete regions 114 (“substantially not shared” here means that less than 2%, less than 0.5%, less than 0.1%, or less than 0.01% of these charge carriers flow toward a different discrete region 114 compared to the rest of these charge carriers). The charge carriers generated by a radiating particle incident around the footprint of one of these discrete regions 114 are substantially not shared with the other of these discrete regions 114. A pixel 150 associated with discrete region 114 can be a region surrounding discrete region 114 in which substantially all (more than 98%, more than 99.5%, more than 99.9%, or more than 99.99%) of the charge carriers generated by incident radiating particles flow into discrete region 114. That is, less than 2%, less than 1%, less than 0.1%, or less than 0.01% of the charge carriers flow out of the pixel.

[0102] like Figure 7CAn alternative detailed cross-sectional view of the radiation detector 100 is shown. According to an embodiment, the radiation absorption layer 110 may include resistors made of semiconductor materials such as silicon, germanium, GaAs, CdTe, CdZnTe, or combinations thereof, but not diodes. Semiconductors may have a high mass attenuation coefficient for the radiation energy of interest.

[0103] When a radiating particle strikes a radiation-absorbing layer 110 that includes resistors but not diodes, the radiating particle can be absorbed and generate one or more charge carriers through several mechanisms. A single radiating particle can generate 10 to 100,000 charge carriers. The charge carriers can drift toward electrical contacts 119A and 119B under an electric field. The electric field can be an external electric field. Electrical contact 119B includes multiple discrete portions. In some embodiments, charge carriers can drift in multiple directions such that the charge carriers generated by a single radiating particle are substantially not shared by two different discrete portions of electrical contact 119B (“substantially not shared” here means that less than 2%, less than 0.5%, less than 0.1%, or less than 0.01% of these charge carriers flow toward a different discrete portion compared to the rest of these charge carriers). The charge carriers generated by a radiating particle incident on the space occupied by one of these discrete portions of electrical contact 119B are substantially not shared with the other discrete portion of electrical contact 119B. A pixel 150 associated with a discrete portion of an electrical contact 119B may be a region surrounding that discrete portion, in which substantially all (more than 98%, more than 99.5%, more than 99.9%, or more than 99.99%) of the charge carriers generated by incident radiating particles flow to that discrete portion of the electrical contact 119B. That is, less than 2%, less than 0.5%, less than 0.1%, or less than 0.01% of these charge carriers flow out of the pixel associated with that discrete portion of the electrical contact 119B.

[0104] Electronic layer 120 may include electronic system 121 adapted to process or interpret signals generated by radiating particles incident on radiation-absorbing layer 110. Electronic system 121 may include analog circuitry such as filter networks, amplifiers, integrators, and comparators, or digital circuitry such as microprocessors and memory. Electronic system 121 may include components shared by multiple pixels or components dedicated to a single pixel. For example, electronic system 121 may include an amplifier dedicated to each pixel and a microprocessor shared among all pixels. Electronic system 121 may be electrically connected to pixels via vias 131. The space between vias may be filled with filler material 130, which may increase the mechanical stability of the connection between electronic layer 120 and radiation-absorbing layer 110. Other bonding techniques may connect electronic system 121 to pixels without using vias.

[0105] Figure 9A flowchart illustrating an image calibration method according to an embodiment is shown, the image calibration method specifically including the following steps:

[0106] Step S901: Using at least one radiation detector, multiple local images of the calibration pattern are captured at multiple shooting positions using radiation from a radiation source, thereby obtaining multiple local images, wherein each local image of the calibration pattern is unique within the calibration pattern.

[0107] In some embodiments, the calibration pattern consists of at least two types of icons arranged in two mutually orthogonal directions in two-dimensional space, with an arrangement rule that ensures the m×n feature icon array at any position within the calibration pattern is unique, where m and n are positive integers. Furthermore, the area of ​​each of the multiple localities of the calibration pattern captured by multiple radiation detectors is greater than the area of ​​the feature icon array. In some embodiments, the area of ​​each of the multiple localities of the calibration pattern captured by each radiation detector is not less than the area of ​​a (m+1)×(n+1) size icon array, ensuring that the entirety of at least one m×n size feature icon array can be captured. The area of ​​the icon array or feature icon array includes not only the area of ​​all icons in the array but also the area of ​​the gaps between them.

[0108] In some embodiments, the calibration pattern consists of at least two icons arranged in a two-dimensional de Bruijn sequence. The icons constituting the calibration pattern can be geometric figures or other graphic symbols with denotative meaning, such as letters, mathematical symbols, etc. Figure 8 The calibration pattern shown consists of four geometric shapes (triangle, circle, square, and X-shape) arranged in a two-dimensional de Bruin sequence. Its feature icon array is 3×3 in size, meaning that any 3×3 icon array is unique in the global arrangement of the calibration pattern.

[0109] In some embodiments, multiple radiation detectors are used, and each radiation detector remains fixed (i.e., does not move) relative to the calibration pattern. In other embodiments, only one radiation detector is used, and this radiation detector is movable relative to the calibration pattern to multiple imaging positions. In this case, multiple portions of the calibration pattern can also be captured, obtaining multiple partial images. In still other embodiments, multiple radiation detectors are used, and at least one of the multiple radiation detectors is movable relative to the calibration pattern to multiple imaging positions.

[0110] In some embodiments, one or more radiation detectors may be moved relative to a calibration pattern to multiple imaging positions via an actuator. In other embodiments, the calibration pattern may be moved relative to one or more radiation detectors to multiple positions via an actuator, such that each radiation detector is moved relative to the calibration pattern to multiple imaging positions.

[0111] In some embodiments, multiple portions of the calibration pattern captured by different radiation detectors may be at least partially identical to each other. In other embodiments, multiple portions of the calibration pattern captured by the same radiation detector at different shooting positions may be at least partially identical to each other.

[0112] In some embodiments, some or all of the multiple shooting locations may at least partially overlap with each other.

[0113] Step S902: Based on each of the multiple local images, determine the calibration dataset and stitching location dataset of the radiation detector that obtained each local image, which are related to the shooting position of the radiation detector when the local image was obtained.

[0114] The location-dependent calibration dataset of the radiation detector can be used to calibrate a local image obtained by the radiation detector at that location into a local calibration image. The location-dependent stitching location dataset of the radiation detector can be used to determine the relative position of the local calibration image with respect to one or more other local calibration images when stitched together with one or more other local calibration images calibrated from one or more other local images.

[0115] In some embodiments, when a radiation detector can be moved to multiple shooting locations, the calibration dataset and stitching location dataset associated with one of the shooting locations are typically different from the calibration dataset and stitching location dataset associated with another shooting location.

[0116] In other embodiments, some or all of the multiple radiation detectors may be moved to the same shooting position at different times. However, due to differences in the orientation, shape and area of ​​the effective detection area of ​​these radiation detectors, the calibration dataset and stitching position dataset associated with the shooting position of one radiation detector are usually different from those associated with the shooting position of another radiation detector.

[0117] Step S903: Using at least one radiation detector, multiple local scenes are captured at multiple shooting positions using radiation to obtain multiple local scene images.

[0118] In some embodiments, multiple portions of a scene captured by different radiation detectors may be at least partially identical to each other to facilitate subsequent stitching. In other embodiments, multiple portions of a scene captured by the same radiation detector at different shooting positions may be at least partially identical to each other to facilitate subsequent stitching.

[0119] Step S904: For each of the plurality of local scene images, the calibration dataset and stitching position dataset of the radiation detector that obtained the local scene image, which are related to the shooting position of the radiation detector when the local scene image was obtained, are respectively used as the calibration dataset and stitching position dataset related to the local scene image.

[0120] Step S905: Using multiple calibration datasets associated with multiple local scene images, the multiple local scene images are calibrated into multiple local scene calibration images. Using multiple stitching position datasets associated with multiple local scene images, the multiple local scene calibration images are stitched together to obtain a stitched calibration image of the scene.

[0121] Figure 10 for Figure 9 A detailed flowchart of step S902 in the process. Figure 11 This schematically illustrates the process of determining the calibration dataset and stitching location dataset based on local images. See also... Figure 10 and Figure 11 Step S902 specifically includes:

[0122] Step S9021: Identify the original image of the feature icon array contained in each local image, and determine the shape and arrangement of the icons in the feature icon array based on the shape and arrangement of the icon images in the original image of the feature icon array.

[0123] In some embodiments, the calibration pattern consists of at least two (e.g., four) icons arranged in two mutually orthogonal directions in a two-dimensional space. The arrangement rule ensures that an m×n (e.g., 3×3) feature icon array at any position in the calibration pattern is unique within the calibration pattern, where m and n are positive integers. For each local image, multiple icon images contained in the local image are first identified, and the shapes and positions of these icon images in the local image are determined. Then, based on the positions of the icon images in the local image, the relative positional relationships between the icon images are determined, thereby determining the arrangement of the icon images. Finally, based on the shapes and arrangement of the icon images, the original image of the feature icon array contained in the local image is determined.

[0124] For example, refer to Figure 11 For filming Figure 8To improve the accuracy of icon recognition, the local image obtained from the calibration pattern shown is preprocessed before identifying the shape and arrangement of the icon images. This preprocessing includes removing bad pixels (i.e., isolated pixels) from the local image, and binarizing and segmenting the local image after removing bad pixels. Through segmentation, icon images that touch the edges of the local image (usually not displaying complete icons) are removed, resulting in a segmented local image. Connected regions contained in the local image can be identified using OpenCV's connected component detection algorithm. For each connected region, the shape of the connected region is determined based on its area ratio with the bounding rectangle and / or the number of vertices in each connected region, thereby determining the shape of the icon image contained in the local image. Furthermore, the position of the icon image in the local image can be determined based on the position of the connected region. Then, based on the position of the icon image in the local image, the relative positional relationship between the icon images is determined, thus determining the arrangement of the icon images. Finally, based on the shape and arrangement of the icon images in the local image, the original image of the feature icon array contained in the local image is determined. Thus, the original image of the feature icon array contained in the local image is identified.

[0125] In some embodiments, the shape of the icons in the feature icon array can be determined based on the shape of the icon images in the original image of the feature icon array, and the arrangement of the icons in the feature icon array can be determined based on the arrangement of the icon images in the original image of the feature icon array.

[0126] Step S9022: Generate a calibration reference image of the feature icon array based on the shape and arrangement of the icons in the feature icon array.

[0127] In some embodiments, the calibration reference image of the feature icon array is an image generated by simulating the projection of the feature icon array onto an image plane parallel to the surface of the calibration pattern facing the radiation source.

[0128] Step S9023: Based on the original image of the feature icon array and the calibration reference image of the feature icon array, determine a transformation dataset that can be used to transform the original image of the feature icon array into the calibration reference image of the feature icon array, and use the transformation dataset as a calibration dataset for the radiation detector that obtains the local image, which is related to the shooting position of the radiation detector when obtaining the local image.

[0129] In some embodiments, based on the original image of the feature icon array and a calibration reference image of the feature icon array, such as using an image alignment algorithm from OpenCV, a transformation dataset that can be determined to transform the original image of the feature icon array into a calibration reference image of the feature icon array can be used as a calibration dataset for the radiometer that acquires the local image, related to the shooting position of the radiometer when acquiring the local image. In some embodiments, referencing Figure 11 The calibration dataset is the deformation registration matrix T.

[0130] Step S9024: Using the calibration dataset of the radiation detector related to the shooting location, calibrate the local image into a local calibration image.

[0131] Obviously, while calibrating a local image into a local calibration image, the original image of the feature icon array contained in the local image is also calibrated into a calibration image of the feature icon array; that is, the local calibration image contains a calibration image of the feature icon array. In some embodiments, the calibration image of the feature icon array is substantially the same as the calibration reference image of the feature icon array.

[0132] Step S9025: Generate a global reference image of the calibration pattern.

[0133] In some embodiments, the global reference image of the calibration pattern is an image generated by simulating the projection of the calibration pattern onto the image plane.

[0134] Step S9026: Based on the position of the calibration image of the feature icon array and / or the calibration reference image of the feature icon array in the global reference image of the calibration pattern and the local calibration image, determine the stitching position dataset of the radiation detector that obtained the local image, which is related to the shooting position of the radiation detector when the local image was obtained.

[0135] The stitching location dataset represents the position of the local calibration image within the global reference image of the calibration pattern.

[0136] Since the feature icon array is unique within the calibration pattern, its position within the calibration pattern can be determined based on the shape and arrangement of the icons within it. Figure 11 The position P in the calibration pattern. Furthermore, based on the position of the feature icon array in the calibration pattern, the position of the calibration image of the feature icon array and / or the calibration reference image of the feature icon array in the global reference image of the calibration pattern can be determined.

[0137] Then, based on the positions of the calibration image of the feature icon array and / or the calibration reference image of the feature icon array in the global reference image of the calibration pattern, and the local calibration image, a stitching position dataset related to the imaging position of the radiation detector at the time of acquiring the local image is determined. Specifically, the stitching position dataset of the radiation detector related to the imaging position includes a first stitching position data subset and a second stitching position data subset. The first stitching position data subset represents the positions of the calibration image of the feature icon array and / or the calibration reference image of the feature icon array in the global reference image of the calibration pattern, and the second stitching position data subset represents the positions of the parts of the local calibration image other than the included calibration image of the feature icon array in the global reference image of the calibration pattern. The first stitching position data subset can be determined based on the positions of the calibration image of the feature icon array and / or the calibration reference image of the feature icon array in the global reference image of the calibration pattern. Then, based on the positions of the calibration image of the feature icon array in the local calibration image and the positions of the other parts of the local calibration image in the local calibration image, the relative positional relationship between the other parts of the local calibration image and the calibration image of the feature icon array can be determined. Next, based on the position of the calibration image of the feature icon array and / or the calibration reference image of the feature icon array in the global reference image of the calibration pattern, as well as the relative positional relationship between other parts of the local calibration image and the calibration image of the feature icon array, the second splicing position data subset can be determined.

[0138] In some embodiments, refer to Figure 11 After calibrating a local image to a local calibration image using the calibration dataset (deformation registration matrix T), the calibration dataset (deformation registration matrix T) can be validated. For example, it can be determined whether the calibration image of the feature icon array is substantially the same as the calibration reference image of the feature icon array. If they are substantially the same, the calibration dataset (deformation registration matrix T) passes validation; if they are different, the calibration dataset (deformation registration matrix T) fails validation, and a new calibration dataset (deformation registration matrix T) needs to be determined or corrected subsequently.

[0139] In some embodiments, a computer-readable storage medium is provided, which stores a computer program or instructions that, when executed by a processor, cause the processor to perform the image calibration method described above.

[0140] The image sensor 9000 described above can be used in a variety of systems, including those described above.

[0141] Figure 12A system is schematically illustrated, comprising an image sensor 9000 as shown in Figures 1-7. This system can be used for medical imaging, such as chest radiography and abdominal radiography. The system includes a radiation source 1201. Radiation emitted from the radiation source 1201 penetrates an object 1202 (e.g., a human body part such as the chest, limbs, or abdomen), is attenuated to varying degrees by the internal structures of the object 1202 (e.g., bones, muscles, fat, and organs), and is projected onto the image sensor 9000. The image sensor 9000 forms an image by detecting the intensity distribution of the radiation.

[0142] Figure 13 A system is schematically illustrated, comprising an image sensor 9000 as shown in Figures 1-7. This system can be used for medical imaging, such as dental radiography. The system includes a radiation source 1301. Radiation emitted from the radiation source 1301 penetrates an object 1302 that is part of the oral cavity of a mammal (e.g., a human). The object 1302 may include the maxilla, palate, teeth, mandible, or tongue. The radiation is attenuated to varying degrees by different structures of the object 1302 and is projected onto the image sensor 9000. The image sensor 9000 forms an image by detecting the intensity distribution of the radiation. Teeth absorb more radiation than cavities, infections, or periodontal ligaments. The radiation dose received by dental patients is typically very small (approximately 0.150 mSv for a full-mouth series).

[0143] Figure 14 A cargo scanning or non-invasive inspection (NII) system is schematically illustrated, comprising an image sensor 9000 as shown in Figures 1-7. This system can be used to inspect and identify cargo in a transportation system, such as containers, vehicles, ships, luggage, etc. The system includes a radiation source 1401. Radiation emitted from the radiation source 1401 can be backscattered from an object 1402 (e.g., a transport container, vehicle, ship, etc.) and projected onto the image sensor 9000. Different internal structures of the object 1402 can backscatter radiation differently. The image sensor 9000 forms an image by detecting the intensity distribution of the backscattered radiation and / or the energy of the backscattered radiation particles.

[0144] Figure 15 Another cargo scanning or non-invasive inspection (NII) system is schematically illustrated, comprising an image sensor 9000 as shown in Figures 1-7. This system can be used for baggage inspection at public transportation stations and airports. The system includes a radiation source 1501. Radiation emitted from the radiation source 1501 penetrates a piece of baggage 1502, is attenuated to varying degrees by the contents of the baggage, and is projected onto the image sensor 9000. The image sensor 9000 forms an image by detecting the intensity distribution of the transmitted radiation. This system can reveal the contents of the baggage and identify prohibited items on public transportation, such as firearms, narcotics, sharp objects, and flammable materials.

[0145] Figure 16 A full-body scanner system is schematically illustrated, comprising an image sensor 9000 as shown in Figures 1-7. The full-body scanner system can detect objects on a human body for security checks without physically removing clothing or making physical contact. The full-body scanner system is capable of detecting non-metallic objects. The full-body scanner system includes a radiation source 1601. Radiation emitted from the radiation source 1601 can be backscattered from the person being inspected 1602 and objects on it, and projected onto the image sensor 9000. Objects and the human body can backscatter radiation differently. The image sensor 9000 forms an image by detecting the intensity distribution of the backscattered radiation. The image sensor 9000 and the radiation source 1601 can be configured to scan a person along a linear or rotational direction.

[0146] Figure 17 A radiation computed tomography (radiation CT) system is schematically illustrated. This radiation CT system uses computer-processed radiation to produce tomographic images (virtual “slices”) of specific areas of a scanned object. The tomographic images can be used for diagnostic and therapeutic purposes in various medical disciplines, or for flaw detection, failure analysis, metrology, assembly analysis, and reverse engineering. The radiation CT system includes an image sensor 9000 and a radiation source 1701, as shown in Figures 1-7. The image sensor 9000 and the radiation source 1701 can be configured to rotate synchronously along one or more circular or helical paths.

[0147] Figure 18 An electron microscope is schematically illustrated. This electron microscope includes an electron source 1801 (also called an electron gun) configured to emit electrons. The electron source 1801 can have various emission mechanisms, such as thermionic, photocathode, cold emission, or plasma source. The emitted electrons pass through an electron optical system 1803, which can be configured to shape, accelerate, or focus the electrons. The electrons then reach a sample 1802, and an image sensor can thereby form an image. The electron microscope may include an image sensor 9000 as shown in Figures 1-7, which is used to perform energy dispersive spectroscopy (EDS). EDS is an analytical technique used for elemental analysis or chemical characterization of a sample. When electrons are incident on a sample, they cause the sample to emit characteristic radiation. The incident electrons excite electrons in the inner shells of atoms in the sample, ejecting them from their shells and creating electron holes where the electrons previously resided. Electrons from outer, higher-energy shells then fill the holes, and the energy difference between the higher and lower energy shells can be released in the form of radiation. The amount and energy of radiation emitted from the sample can be measured by the image sensor 9000.

[0148] The image sensor 9000 can also be used in other applications, such as radiation telescopes, mammography, industrial radiation defect detection, radiation microscopy or radiation photomicrography, radiation casting inspection, radiation nondestructive testing, radiation weld inspection, and digital subtraction angiography. It can be used in place of photographic film, PSP film, radiation image intensifiers, scintillators, or other semiconductor radiation detectors.

[0149] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of this application, and not to limit them; under the concept of this application, the technical features of the above embodiments or different embodiments can also be combined, the steps can be implemented in any order, and there are many other variations of different aspects of this application as described above, which are not provided in detail for the sake of brevity; although this application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the foregoing embodiments, or make equivalent substitutions for some of the technical features; and these modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of this application.

Claims

1. An image calibration method, characterized in that, The method includes: Using at least one radiation detector, multiple local images of a calibration pattern are captured at multiple shooting positions using radiation from a radiation source, wherein each local image of the calibration pattern is unique within the calibration pattern; and Based on each of the plurality of local images, a calibration dataset and a stitching position dataset related to the shooting position of the radiation detector at the time of obtaining the local image are determined for each local image. The calibration dataset related to the shooting position of the radiation detector can be used to calibrate the local image obtained by the radiation detector at the shooting position into a local calibration image. The stitching position dataset related to the shooting position of the radiation detector can be used to determine the relative position of the local calibration image with respect to the one or more other local calibration images calibrated from one or more other local images when stitched together. The calibration pattern is formed by arranging at least two kinds of icons in two two-dimensional space along two mutually orthogonal directions. The arrangement rule of the icons makes the m×n feature icon array at any position in the calibration pattern unique in the calibration pattern, where m and n are positive integers, and the area of ​​each of the multiple local areas of the calibration pattern is greater than the area of ​​the feature icon array. The stitching position dataset represents the position of the local calibration image in the global reference image of the calibration pattern, which is an image generated by simulating the projection of the calibration pattern onto an image plane, wherein the image plane is parallel to the surface of the calibration pattern facing the radiation source.

2. The image calibration method according to claim 1, characterized in that, The method further includes: Using the at least one radiation detector, multiple local scenes are captured at the multiple shooting locations using the radiation, thereby obtaining multiple local scene images; For each of the plurality of local scene images, the calibration dataset and the stitching position dataset related to the shooting position of the radiation detector at the time the local scene image was obtained are respectively used as the calibration dataset and the stitching position dataset related to the local scene image; and Using multiple calibration datasets associated with the multiple local scene images, the multiple local scene images are calibrated into multiple local scene calibration images. Using multiple stitching position datasets associated with the multiple local scene images, the multiple local scene calibration images are stitched together to obtain the stitched calibration image of the scene.

3. The image calibration method according to claim 1, characterized in that, The area of ​​each of the multiple localities of the calibration pattern is not less than the area of ​​the icon array of size (m+1)×(n+1).

4. The image calibration method according to claim 1, characterized in that, The calibration pattern is formed by arranging the at least two icons in a two-dimensional de Bruin sequence.

5. The image calibration method according to claim 4, characterized in that, The calibration pattern consists of four icons arranged in a two-dimensional de Bruin sequence, and the feature icon array is 3×3 in size.

6. The image calibration method according to claim 1, characterized in that, The step of determining, based on each of the plurality of local images, a calibration dataset and a stitching location dataset related to the imaging position of the radiation detector at the time the local image was acquired, for each local image, includes: Identify the original image of the feature icon array contained in each local image, and determine the shape and arrangement of the icons in the feature icon array based on the shape and arrangement of the icon images in the original image of the feature icon array; A calibration reference image of the feature icon array is generated based on the shape and arrangement of the icons in the feature icon array; Based on the original image of the feature icon array and the calibration reference image of the feature icon array, a transformation dataset is determined that can be used to transform the original image of the feature icon array into the calibration reference image of the feature icon array. The transformation dataset is used as the calibration dataset of the radiation detector that obtained the local image, which is related to the shooting position of the radiation detector when the local image was obtained. Using the calibration dataset of the radiation detector related to the shooting location, the local image is calibrated into the local calibration image, wherein the original image of the feature icon array contained in the local image is calibrated into the calibration image of the feature icon array contained in the local calibration image; The global reference image for generating the calibration pattern; and Based on the position of the calibration image of the feature icon array and / or the calibration reference image of the feature icon array in the global reference image of the calibration pattern, and the local calibration image, determine the stitching position dataset of the radiation detector that obtained the local image, which is related to the shooting position of the radiation detector when the local image was obtained.

7. The image calibration method according to claim 6, characterized in that, The calibration reference image of the feature icon array is an image generated by simulating the projection of the feature icon array onto the image plane.

8. The image calibration method according to claim 6, characterized in that, Before determining the stitching position dataset related to the shooting position of the radiation detector at the time of obtaining the local image, based on the position of the calibration image of the feature icon array and / or the calibration reference image of the feature icon array in the global reference image of the calibration pattern and the local calibration image, the stitching position dataset of the radiation detector that obtained the local image is related to the shooting position of the radiation detector when the local image was obtained. The position of the feature icon array within the calibration pattern is determined based on the shape and arrangement of the icons in the feature icon array. Based on the position of the feature icon array in the calibration pattern, determine the position of the calibration image of the feature icon array and / or the calibration reference image of the feature icon array in the global reference image of the calibration pattern.

9. The image calibration method according to claim 6, characterized in that, The stitching position dataset of the radiation detector related to the shooting position includes a first stitching position data subset and a second stitching position data subset. The first stitching position data subset represents the position of the calibration image of the feature icon array and / or the calibration reference image of the feature icon array in the global reference image of the calibration pattern. The second stitching position data subset represents the position of the portion of the local calibration image other than the calibration image of the included feature icon array in the global reference image of the calibration pattern. The step of determining the stitching position dataset, which is related to the shooting position of the radiation detector at the time of obtaining the local image, based on the position of the calibration image of the feature icon array and / or the calibration reference image of the feature icon array in the global reference image of the calibration pattern and the local calibration image, includes: The first splicing position data subset is determined based on the position of the calibration image of the feature icon array and / or the calibration reference image of the feature icon array in the global reference image of the calibration pattern; Based on the position of the calibration image of the feature icon array within the local calibration image and the positions of the other portions of the local calibration image within the local calibration image, determine the relative positional relationship between the other portions of the local calibration image and the calibration image of the feature icon array; and A second splicing position data subset is determined based on the position of the calibration image of the feature icon array and / or the calibration reference image of the feature icon array in the global reference image of the calibration pattern, as well as the relative positional relationship between the other parts of the local calibration image and the calibration image of the feature icon array.

10. The image calibration method according to claim 6, characterized in that, After calibrating the local image to the local calibration image using the calibration dataset, it is determined whether the calibration image of the feature icon array is the same as the calibration reference image of the feature icon array.

11. The image calibration method according to any one of claims 1 to 10, characterized in that, Multiple radiation detectors are used, and each of the multiple radiation detectors remains fixed relative to the calibration pattern.

12. The image calibration method according to any one of claims 1 to 10, characterized in that, Only one radiation detector is used, and the radiation detector can be moved relative to the calibration pattern to the plurality of shooting positions.

13. The image calibration method according to any one of claims 1 to 10, characterized in that, Multiple radiation detectors are used, and at least one of the multiple radiation detectors is capable of moving to the multiple shooting positions relative to the calibration pattern.

14. An image sensor, characterized in that, Includes at least one radiation detector, and the image sensor is configured to perform the steps of the image calibration method as described in any one of claims 1-13.

15. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores a computer program that, when executed by a processor, performs the steps of the image calibration method as described in any one of claims 1-13.

Images