An image acquisition method and system suitable for cylindrical well walls

Abstract

This invention discloses an image acquisition method and system suitable for cylindrical well walls, belonging to the field of image processing technology. It includes: acquiring images of the cylindrical well wall at different depth positions to obtain multiple original images and recording the depth position information at the time of acquisition; for each original image, obtaining multiple cylindrical unfolded images through cylindrical back projection transformation; establishing a mapping relationship between image pixels and actual height based on the geometric parameters of the camera array and the cylindrical well wall; for multiple cylindrical unfolded images at the same depth position, calculating the overlapping area of ​​the cylindrical unfolded images of adjacent cameras and performing image fusion to obtain panoramic unfolded images at different depth positions; determining the overlapping area of ​​adjacent cylindrical unfolded images along the well depth direction based on the mapping relationship and depth position information, and performing image fusion to obtain continuous unfolded images of the cylindrical well wall. This invention can generate distortion-free and seamlessly stitched panoramic unfolded images.

Description

Technical Field

[0001] This invention belongs to the field of image processing technology, specifically relating to an image acquisition method and system suitable for cylindrical well walls. Background Technology

[0002] With the rapid development of underground engineering construction, the demand for high-precision imaging and condition monitoring of the inner walls of cylindrical structures such as shafts, inclined shafts, and pipelines is becoming increasingly prominent. Currently, the industry typically uses multi-camera arrays in conjunction with mobile platforms for image acquisition, and leverages image stitching technology to generate panoramic unfolded images to support engineering monitoring, lithological identification, and structural analysis. Existing image stitching methods are mainly based on planar projection models or spherical projection models, and often rely on feature point matching to achieve image alignment and fusion, which has been widely used in general planar or spherical scenes.

[0003] However, the aforementioned methods have significant drawbacks when applied to scenarios with well-defined geometric structures, such as cylindrical well walls: the images acquired by these techniques suffer from severe nonlinear geometric distortion, manifesting as image center compression, edge stretching, and proportional distortion. This distortion not only directly affects the visual realism of individual images but also causes circumferential and longitudinal misalignments and inconsistencies during subsequent stitching, making it difficult to obtain accurate results after geometric image processing of the inner walls of cylindrical structures. Summary of the Invention

[0004] The purpose of this invention is to provide an image acquisition method and system suitable for cylindrical well walls, which can solve the problems existing in the background art.

[0005] To solve the above-mentioned technical problems, the present invention is implemented as follows: In a first aspect, embodiments of the present invention provide an image acquisition method suitable for cylindrical well walls, comprising: By using a camera array deployed along the circumference at different depth positions, images of the cylindrical well wall are acquired, resulting in multiple raw images, and the depth position information at the time of acquisition is recorded. For each of the original images, the image pixels are mapped from the planar coordinate system to the cylindrical unfolded coordinate system through cylindrical back projection transformation to obtain multiple cylindrical unfolded images; Based on the geometric parameters of the camera array and the cylindrical well wall, a mapping relationship is established between the image pixels of the cylindrical unfolded image and the actual height of the cylindrical well wall; For multiple cylindrical unfolded images at the same depth position, based on the geometric layout parameters of the camera array, the overlapping area of ​​the cylindrical unfolded images of adjacent cameras is calculated, and the overlapping area is fused to obtain panoramic unfolded images at different depth positions. Based on the mapping relationship and the depth location information, the overlapping area of ​​adjacent panoramic unfolded images along the well depth direction is determined, and the overlapping area is image fused to obtain a continuous unfolded image of the cylindrical well wall.

[0006] Optionally, based on the geometric parameters of the camera array and the cylindrical well wall, a mapping relationship is established between the image pixels of the unfolded cylindrical image and the actual height of the cylindrical well wall, including: The geometric parameter information is determined, including: the vertical field of view of a single camera in the camera array, the horizontal field of view of a single camera, the distribution radius of the camera array, the radius of the cylindrical well wall, and the height of the cylindrical unfolded image; In the vertical direction, the actual height of the cylindrical well wall corresponding to the cylindrical unfolded image is determined based on the vertical field of view of the single camera, the radius of the cylindrical well wall, and the distribution radius of the camera array. Based on the height of the cylindrical unfolded image and the actual height of the cylindrical well wall corresponding to the cylindrical unfolded image, a mapping relationship is established between the actual height of the cylindrical well wall corresponding to each pixel in the cylindrical unfolded image.

[0007] Optionally, for multiple cylindrical unfolded images at the same depth position, based on the geometric layout parameters of the camera array, the overlapping region of the cylindrical unfolded images of adjacent cameras is calculated, and the overlapping region is fused to obtain panoramic unfolded images at different depth positions, including: Based on the horizontal field of view of the single camera, the distribution radius of the camera array, and the radius of the cylindrical well wall, calculate the range of the central angle corresponding to the overlapping image area of ​​the images acquired by adjacent cameras in the camera array on the cylindrical well wall; Based on the central angle range, the corresponding overlapping image regions are extracted from two adjacent cylindrical unfolded images at the same depth position; Using a weighted fusion algorithm, the overlapping image regions extracted from two adjacent cylindrical unfolded images at the same depth position are fused to obtain panoramic unfolded images at different depth positions.

[0008] Optionally, based on the horizontal field of view of the single camera, the distribution radius of the camera array, and the radius of the cylindrical well wall, the range of the central angle corresponding to the overlapping image region on the cylindrical well wall of the images acquired by adjacent cameras in the camera array is calculated, including: Based on the horizontal field of view of the single camera, the first central angle corresponding to the horizontal imaging range of the single camera on the cylindrical well wall is determined; Based on the distribution radius of the camera array and the radius of the cylindrical well wall, calculate the second central angle corresponding to the imaging area that does not overlap with the adjacent camera within the horizontal imaging range of the single camera; Based on the first central angle and the second central angle, the range of central angles corresponding to the overlapping image areas of the images captured by adjacent cameras in the camera array on the cylindrical well wall is calculated.

[0009] Optionally, based on the mapping relationship and the depth location information, the overlapping region of adjacent panoramic unfolded images along the well depth direction is determined, and image fusion is performed on the overlapping region to obtain a continuous unfolded image of the cylindrical well wall, including: Based on the depth position information, determine the relative positional relationship of the panoramic unfolded image of each pair of adjacent depth positions in the vertical direction; Based on the mapping relationship and the relative positional relationship of the panoramic unfolded images of each pair of adjacent depth positions in the vertical direction, calculate the overlapping image area of ​​the panoramic unfolded images of each pair of adjacent depth positions in the vertical direction. In the panoramic unfolded images at different depth positions, the actual position of the overlapping image area in the vertical direction of each pair of adjacent depth positions is located, and the overlapping image areas in the vertical direction of each pair of adjacent depth positions are fused and stitched together to obtain the continuous unfolded image of the cylindrical well wall.

[0010] Optionally, locating the actual position of the overlapping image region in the vertical direction of the panoramic unfolded images at different depth positions includes: Based on the depth location information and the mapping relationship, the theoretical overlap height between each pair of adjacent panoramic unfolded images is determined; Within the theoretical overlap height between each pair of adjacent panoramic unfolded images, template matching and phase correlation algorithms are used to locate the actual position of the overlapping image area in the vertical direction of each pair of adjacent depth positions of the panoramic unfolded images.

[0011] Secondly, embodiments of the present invention provide an image acquisition system suitable for cylindrical well walls, applied to the steps of performing the method described in the first aspect, including a camera array module and a data processing module; the camera array module includes multiple cameras, which are disposed at the central axis of the cylindrical well wall and are uniformly distributed in a ring; the central axis of the camera array module coincides with the central axis of the cylindrical well wall; The camera array module is used to acquire images of the cylindrical well wall at different depth positions, obtain multiple raw images, and record the depth position information at the time of acquisition. The data processing module is used for: For each of the original images, the image pixels are mapped from the planar coordinate system to the cylindrical unfolded coordinate system through cylindrical back projection transformation to obtain multiple cylindrical unfolded images; Based on the geometric parameters of the camera array and the cylindrical well wall, a mapping relationship is established between the image pixels of the cylindrical unfolded image and the actual height of the cylindrical well wall; For multiple cylindrical unfolded images at the same depth position, based on the geometric layout parameters of the camera array, the overlapping area of ​​the cylindrical unfolded images of adjacent cameras is calculated, and the overlapping area is fused to obtain panoramic unfolded images at different depth positions. Based on the mapping relationship and the depth location information, the overlapping area of ​​adjacent panoramic unfolded images along the well depth direction is determined, and the overlapping area is image fused to obtain a continuous unfolded image of the cylindrical well wall.

[0012] Optionally, the horizontal field of view of the camera, the distribution radius of the camera array module, and the radius of the cylindrical well wall satisfy the following conditions: The distance from the intersection of the center lines of the fields of view of two adjacent cameras on the cylindrical well wall to the central axis of the camera array module is less than the radius of the cylindrical well wall, so that the fields of view of adjacent cameras have overlapping image areas on the cylindrical well wall.

[0013] Thirdly, embodiments of this disclosure provide an electronic device, including: a processor, a memory, and a computer program stored in the memory and capable of running on the processor, wherein the computer program, when executed by the processor, implements the steps of an image acquisition method suitable for cylindrical well walls.

[0014] Fourthly, embodiments of this application provide a computer-readable storage medium storing a computer program that, when executed by a processor, implements the steps of an image acquisition method suitable for cylindrical well walls.

[0015] The technical solutions provided by the embodiments of the present invention bring at least the following beneficial effects: First, this invention employs cylindrical back-projection transformation, fundamentally eliminating the nonlinear distortion caused by the arc-shaped structure, thus significantly improving the geometric fidelity of individual images. The corrected image presents a regular rectangular unfolded shape with equal proportions and angles. Second, by establishing a deterministic mapping relationship between pixels and the actual cylindrical well wall dimensions, the image becomes measurable, making the stitched panoramic image not only visually continuous but also directly usable for precise engineering dimension measurement and analysis. Third, this invention achieves integrated processing of circumferential and longitudinal stitching by fusing the geometric parameters and depth position information of the camera array. Finally, the generated distortion-free and seamlessly stitched panoramic unfolded image provides a reliable data foundation for subsequent intelligent extraction and modeling of structural surfaces based on artificial intelligence. Attached Figure Description

[0016] Figure 1 This is a schematic diagram illustrating the steps of a method for migrating permission information according to an embodiment of the present invention; Figure 2 This is a comparative schematic diagram of the original image before and after correction in one embodiment of the present invention; Figure 3 This is a schematic diagram illustrating the mapping relationship between the image pixels of a vertically unfolded cylindrical image and the actual height of the cylindrical well wall in one embodiment of the present invention; Figure 4 This is a schematic diagram of the imaging principle of a camera array in the horizontal direction in one embodiment of the present invention; Figure 5 This is a schematic diagram of the architecture of an image acquisition system suitable for cylindrical well walls provided by an embodiment of the present invention. Detailed Implementation

[0017] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of the present invention. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0018] The terms "first," "second," etc., used in the specification and claims of this invention are used to distinguish similar objects and are not used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate so that embodiments of the invention can be implemented in orders other than those illustrated or described herein. Furthermore, in the specification and claims, "and / or" indicates at least one of the connected objects, and the character " / " generally indicates that the preceding and following objects are in an "or" relationship.

[0019] The following description, in conjunction with the accompanying drawings, details an image acquisition method and system suitable for cylindrical well walls provided by the present invention through specific embodiments and application scenarios.

[0020] It should be noted that the technical solution and core method of this invention are not limited to cylindrical well walls in the strict sense. The principle of image correction and stitching based on cylindrical projection geometry models is also applicable to other high-precision imaging and panoramic unfolding scenarios with similar or near-cylindrical inner wall structures. Specifically, any tubular or cylindrical internal structure with a constant or near-constant radius, such as various pipelines (water, oil and gas, municipal pipelines), vertical shafts, inclined shafts, silos, the inner walls of large tanks, and certain specific tunnel cross-sections, can be image acquired, distortion corrected, and seamlessly stitched using the method proposed in this invention, as long as its inner wall surface can be approximately modeled as a cylindrical or regular cylindrical surface.

[0021] Figure 1 This is a schematic diagram illustrating the steps of an image acquisition method suitable for cylindrical well walls according to an embodiment of the present invention, including: Step S11: By using a camera array deployed along the circumference at different depth positions, images of the cylindrical well wall are acquired, resulting in multiple original images, and the depth position information at the time of acquisition is recorded.

[0022] To acquire complete, continuous, and comprehensive cylindrical wellbore image data, this invention employs a multi-camera circumferential array arrangement. Optionally, a four-camera circular array is used, with each camera uniformly distributed circumferentially along the central axis of the wellbore. The array's central axis coincides with the central axis of the cylindrical wellbore, ensuring consistent distance between each camera and the wellbore, thus maintaining uniform geometric scale during imaging. The camera array is mounted on a movable platform, such as a winch, lifting device, or traveling mechanism, enabling fixed-point image acquisition at different depths. During acquisition, movement is paused at preset acquisition intervals, triggering all cameras to simultaneously capture multiple raw images of the cylindrical wellbore at that depth. Each raw image is associated with the depth position information at the time of acquisition. This depth position information is recorded and stored in real time by an encoder or positioning system, using the wellhead as the reference zero point, recording the absolute depth or relative displacement value relative to the wellhead. This invention supports multiple acquisition modes, including continuous, segmented, and fixed-point acquisition, adapting to different well depths, diameters, and engineering requirements, achieving complete acquisition of images of the entire well section from shallow to deep layers.

[0023] Step S12: For each of the original images, the image pixels are mapped from the planar coordinate system to the cylindrical unfolded coordinate system through cylindrical back projection transformation to obtain multiple cylindrical unfolded images.

[0024] Figure 2This is a comparative schematic diagram of the original image before and after correction in one embodiment of the present invention. The original image before correction is as follows: Figure 2 As shown above, the corrected multi-cylinder unfolded images are as follows: Figure 2 As shown below, when the camera takes pictures inside the cylinder, there is a geometric inconsistency between its imaging plane and the curved surface of the cylindrical well wall, resulting in arc-shaped distortion of the image with center compression and edge stretching, which directly affects the accuracy of subsequent stitching and measurement. To solve this problem, this invention introduces the idea of ​​cylindrical back projection transformation to establish a coordinate system from the image plane. Cylindrical expansion coordinate system to the cylindrical well wall The mapping relationship. Specifically, firstly, based on the known geometric parameters, including the radius of the cylindrical well wall... Distribution radius of camera array Camera horizontal field of view Image width With height And the azimuth angle of the camera in the array, to construct the imaging geometry model. For each pixel in the original image The coordinates of the corresponding three-dimensional point on the cylindrical surface are calculated by back projection, and then the three-dimensional point is unfolded onto a two-dimensional plane to obtain its coordinates in the unfolded image of the cylinder. During the transformation, the horizontal direction is uniformly expanded according to the relationship between arc length and angle, while the vertical direction maintains a linear correspondence. This results in the cylindrical well wall texture, which originally exhibits arc-shaped distortion in the image, appearing as a rectangular area with uniform horizontal direction and proportional vertical direction in the corrected image. This eliminates visual geometric distortion and establishes a mathematical correspondence between pixel positions and the actual positions of the cylindrical well wall.

[0025] Step S13: Based on the geometric parameter information of the camera array and the cylindrical well wall, establish a mapping relationship between the image pixels of the cylindrical unfolded image and the actual height of the cylindrical well wall.

[0026] This involves establishing a deterministic and high-precision mathematical correspondence between the pixel coordinates of the digital image after cylindrical projection correction and the actual physical dimensions (vertical height or depth) of the cylindrical well wall. The principle is to use the geometric model of camera imaging and known, measurable geometric parameters of the system to derive the actual height of the cylindrical well wall arc segment represented by each row of pixels in the image through trigonometric calculations. In practice, a series of key geometric parameters must first be acquired or calibrated. Then, a linear or nonlinear transformation formula from the pixel row index to the real-world height is established in the vertical direction (i.e., the well depth direction), ultimately forming a lookup table or continuous function. This allows for the lookup or calculation of the height component of any pixel in the corrected image at its corresponding actual three-dimensional position on the cylindrical well wall.

[0027] In an optional implementation, step S13 specifically includes steps S131 to S133: Step S131: Determine the geometric parameter information, which includes: the vertical field of view of a single camera in the camera array, the horizontal field of view of a single camera, the distribution radius of the camera array, the radius of the cylindrical well wall, and the height of the cylindrical unfolded image.

[0028] Geometric parameter information serves as prior information for subsequent image processing. The specific parameters that need to be determined or obtained are as follows: (1) Vertical field of view (θ) of a single camera: Figure 3 This is a schematic diagram illustrating the mapping relationship between the image pixels of the vertically unfolded cylindrical image and the actual height of the cylindrical well wall in one embodiment of the present invention. Please refer to [link / reference]. Figure 3 The vertical field of view of a single camera represents the maximum range of light angles that can be received in the vertical direction of the camera's imaging sensor, which determines the range of objects that the camera can see in the vertical direction at a fixed object distance. (2) The horizontal field of view (α) of a single camera: similar to the vertical field of view, it represents the range of the camera's field of view in the horizontal direction. (3) The distribution radius (r) of the camera array: refers to the horizontal distance from the optical center of each camera in the circular camera array to the axis of rotation of the array. (4) The radius (R) of the cylindrical well wall: refers to the radius of the cylindrical well wall to be imaged. (5) The height of the cylindrical unfolded image: refers to the number of pixel rows in the vertical direction of the single cylindrical unfolded image output after the cylindrical back projection transformation in step S12.

[0029] Step S132: In the vertical direction, determine the actual height of the cylindrical well wall corresponding to the cylindrical unfolded image based on the vertical field of view of the single camera, the radius of the cylindrical well wall, and the distribution radius of the camera array.

[0030] Since the camera is located at the center of the cylindrical well wall (radius r), the observed target surface of the cylindrical well wall is a cylindrical surface with radius R. Therefore, the imaging relationship is not a simple similar triangle, but requires consideration of radial geometry. The specific calculation principle is as follows: Consider the camera as a point; its vertical field of view θ forms a solid pyramid surface in space with the camera as its vertex. This solid pyramid surface intersects with the surface of the cylindrical well wall with radius R, extracting a curved strip-shaped region. In this invention, what needs to be calculated is the vertical height of this strip-shaped region after the cylindrical well wall is unfolded into a plane. For example... Figure 3As shown, through geometric derivation, the actual height H can be calculated by the formula: H = 2 * (R - r) * tan(θ / 2). Here, (R - r) is the radial distance between the cylindrical wellbore and the camera, and tan(θ / 2) reflects the ratio of the opposite side to the adjacent side corresponding to the half field angle. The physical meaning of the above formula is that it correlates the field angle of the camera and the radial distance between the camera and the target, thereby accurately obtaining the real-world height covered by a single image, that is, for converting the pixel scale to the physical scale.

[0031] Step S133, based on the height of the cylindrical unwrapped image and the actual height of the cylindrical wellbore corresponding to the cylindrical unwrapped image, establish the mapping relationship of the actual height of the cylindrical wellbore corresponding to each pixel in the cylindrical unwrapped image.

[0032] In the vertical direction, assume that the height of a cylindrical unwrapped image is H_pixels, and the actual height of the corresponding cylindrical wellbore of this cylindrical unwrapped image is H. Since during the cylindrical projection correction process, the non-linear distortion has been eliminated, making the pixel arrangement in the vertical direction of the cylindrical unwrapped image linearly correspond to the actual height of the cylindrical wellbore. Therefore, a pixel height equivalent Scale_factor = H / H_pixels can be calculated. The pixel height equivalent represents the actual height change amount represented by a pixel row in the vertical direction of the cylindrical unwrapped image. Based on this, for any pixel row coordinate y (0 ≤ y < H_pixels) in the cylindrical unwrapped image, its corresponding actual height coordinate of the cylindrical wellbore (with the top of the image as the reference zero point) can be calculated by the linear formula: Height_real(y) = y * Scale_factor. Conversely, if the actual height on the cylindrical wellbore is known, the approximate pixel row range corresponding to it in the cylindrical unwrapped image can also be calculated inversely. To handle possible minor non-linearity or improve accuracy, the mapping relationship can also be expressed as a lookup table, pre-calculated through actual calibration data, storing the corresponding actual height value for each pixel row index, to obtain the mapping relationship of the actual height of the cylindrical wellbore corresponding to each pixel in the cylindrical unwrapped image.

[0033] Step S14, for multiple cylindrical unwrapped images at the same depth position, based on the geometric layout parameters of the camera array, calculate the overlapping area of the cylindrical unwrapped images of adjacent cameras, and perform image fusion on the overlapping area to obtain panoramic unwrapped images at different depth positions.

[0034] Multiple cylindrical unfolded images acquired by a circular camera array at the same depth (i.e., the same horizontal cross-section) and corrected for cylindrical surface appearance are stitched together to form a complete, continuous panoramic unfolded image. Traditional stitching methods based on feature point matching are unreliable in environments with weak texture, high repeatability, and potential for reflection or dust, such as cylindrical well walls, and are prone to misalignment due to residual geometric distortion. This invention abandons the traditional feature-dependent path and instead fully utilizes the system's inherent geometric prior knowledge, namely geometric layout parameters. Specifically, based on the geometric layout parameters of the camera array, the angular range corresponding to the theoretical overlap area of ​​adjacent camera views on the cylindrical well wall is calculated. Then, the corresponding pixel regions in each corrected cylindrical unfolded image are further located, thus achieving "computational localization" rather than "search matching" of the overlapping image region. Finally, image fusion technology is used to smooth the transition of pixels in the overlapping region, eliminating stitching boundary traces and ensuring consistency of circumferential brightness and texture. By processing all adjacent image pairs in the array sequentially, the local cylindrical unfolded images of all cameras at the same depth can be combined into a panoramic unfolded image covering the entire circumference of the cylindrical well wall.

[0035] In an optional implementation, step S14 specifically includes steps S141 to S143: Step S141: Based on the horizontal field of view of the single camera, the distribution radius of the camera array, and the radius of the cylindrical well wall, calculate the range of the central angle corresponding to the overlapping image area of ​​the images acquired by adjacent cameras in the camera array on the cylindrical well wall.

[0036] Figure 4 This is a schematic diagram illustrating the imaging principle of a camera array in the horizontal direction according to one embodiment of the present invention. Please refer to [link / reference]. Figure 4 Based on the horizontal field of view (α) representing the width of the camera's own field of view, the distribution radius (r) describing the radial position of the camera array within the wellbore, and the cylindrical radius (R) defining the size of the cylindrical well wall, a series of trigonometric geometric operations are used to finally output an angle value measured in terms of the central angle, representing the size of the central angle (denoted as ∠θ_overlap) corresponding to the arc-shaped area simultaneously "seen" by two adjacent cameras on the circumference of the cylindrical well wall.

[0037] In an optional implementation, step S141 specifically includes steps S1411 to S1413: Step S1411: Based on the horizontal field of view of the single camera, determine the first central angle corresponding to the horizontal imaging range of the single camera on the cylindrical well wall.

[0038] The horizontal field of view α of a single camera is an inherent optical parameter that defines the maximum horizontal angular range of light that the camera can receive. Ideally, when the camera is mounted on the central axis of the wellbore (i.e., the camera's optical axis passes through the center of the wellbore and the camera's photosensitive surface is perpendicular to the radial direction), the projection of its entire horizontal field of view at infinity corresponds to a sector with angle α. However, in the actual scenario of this invention, the imaging target is not at infinity, but a cylindrical surface with radius R, and the camera is not located at the center (there exists a distribution radius r). Nevertheless, to establish a preliminary correspondence, it is first determined that if the two limiting rays of the camera's field of view are extended in opposite directions to intersect the cylindrical surface of the wellbore wall, the central angle corresponding to the arc segment intercepted by these two rays on the circular cross-section of the wellbore is numerically equal to the camera's horizontal field of view α. This forms the starting point for the calculation; that is, the first central angle is defined as the camera's horizontal field of view α itself, representing the circumferential range of the cylindrical wellbore wall that the camera can cover under an idealized simplified model that ignores camera eccentricity (i.e., assumes the camera is located at the center).

[0039] Step S1412: Based on the distribution radius of the camera array and the radius of the cylindrical well wall, calculate the second central angle corresponding to the imaging area that does not overlap with the adjacent camera within the horizontal imaging range of the single camera.

[0040] Because the camera is located at a distance r from the center, and its optical axis points towards the cylindrical well wall, its left and right field-of-view boundaries are not symmetrically distributed with respect to the well radius. Therefore, the portion of the camera's field of view closer to the adjacent camera overlaps with the adjacent camera's field of view, while the portion farther away is a non-overlapping, exclusive region. This invention calculates half of the central angle corresponding to this exclusive region, denoted as β (the second central angle). A triangle can be constructed using the camera position (distance r from the center), the direction of the field-of-view boundary line (angle α / 2 with the optical axis), and the circumference of the cylindrical well wall (radius R). Applying the sine theorem, the formula for calculating β can be derived: β = arcsin( (r * sin(α / 2)) / R). The physical meaning of this formula is that the ratio (r * sin(α / 2)) / R reflects the degree of offset of the position where the camera's field-of-view boundary line touches the cylindrical well wall relative to the direction of the center due to the camera's eccentricity; its arcsine value is the central angle β corresponding to this offset. β represents the portion of the central angle subtended by the arc from the projection point of the camera's optical axis onto the cylindrical well wall (i.e., the cylindrical well wall point corresponding to the image center) to the intersection point of the field of view boundary on that side onto the cylindrical well wall, that does not intersect with the field of view of the adjacent camera. Therefore, for a single camera, there is a unique central angle region of size β on both its left and right sides.

[0041] Step S1413: Based on the first central angle and the second central angle, calculate the range of central angles corresponding to the overlapping image areas of the images captured by adjacent cameras in the camera array on the cylindrical well wall.

[0042] Based on the first central angle α (total coverage of a single camera) and the second central angle β (exclusive area on one side), the total central angle (∠θ_overlap) corresponding to the overlapping area between two adjacent cameras can be obtained by subtraction. Specifically, the total field of view central angle α of a single camera consists of three parts: the exclusive area on the left (central angle β), the overlapping area shared with the adjacent camera in the middle, and the exclusive area on the right (central angle β). Therefore, the central angle corresponding to the overlapping area in the middle is equal to the total field of view angle minus the exclusive angles on both sides, that is: ∠θ_overlap = α - 2β. Substituting the calculation formula of β in step S1412, we get: ∠θ_overlap = α - 2 * arcsin( (r * sin(α / 2)) / R ).

[0043] The above equation establishes an analytical relationship between the system's geometric parameters (α, r, R) and the size of the overlapping region (∠θ_overlap). The calculated ∠θ_overlap is the range of the central angles corresponding to the overlapping image regions on the cylindrical well wall of the images acquired by adjacent cameras in the camera array.

[0044] Step S142: Based on the central angle range, extract the corresponding overlapping image regions from two adjacent cylindrical unfolded images at the same depth position.

[0045] Since the cylindrical back-projection transformation in step S12 has mapped the image from the original planar coordinate system to a standard cylindrical unfolded coordinate system, in the cylindrical unfolded coordinate system, the horizontal direction (x-axis) of the image has a linear correspondence with the circumferential direction (i.e., the central angle direction) of the cylindrical well wall. Specifically, for a cylindrical unfolded image that completely covers the horizontal field of view α of the camera, its entire image width (W_pixels) corresponds to the central angle α. Therefore, a proportional relationship can be established between the horizontal pixel coordinate x and the central angle θ: the proportionality coefficient k = α / W_pixels. Based on this proportional relationship and the calculated range of the central angle ∠θ_overlap of the overlapping area, the range of pixel columns corresponding to this overlapping area in two adjacent images can be calculated respectively. For the left camera image, the overlapping area is located on the right side of the image, and the starting column coordinates can be calculated as: x_left_start = W_pixels - (∠θ_overlap / k); for the right camera image, the overlapping area is located on the left side of the image, and the starting column coordinates can be calculated as: x_right_start = 0, and the ending column coordinates are x_right_end = ∠θ_overlap / k. Vertically, since they are at the same depth, the overlapping area covers the entire height H_pixels of the image. Based on the calculated column coordinate range, two completely corresponding image blocks representing the same cylindrical well wall region can be cropped from the two adjacent cylindrical unfolded images.

[0046] Step S143: Using a weighted fusion algorithm, the overlapping image regions extracted from two adjacent cylindrical unfolded images at the same depth position are fused to obtain panoramic unfolded images at different depth positions.

[0047] The overlapping image regions (denoted as Patch_A and Patch_B) extracted from two adjacent cylindrical unfolded images at the same depth position are used as input. A weighted fusion algorithm is adopted, the core idea of ​​which is to assign a weight that varies with time / space to the pixel value from the two adjacent cylindrical unfolded images based on its position in the overlapping image region, and then perform a weighted average. The most commonly used method is linear weighted fusion (also known as fade-in fade-out fusion). In specific implementation, a one-dimensional weight mask is defined for the overlapping region. For a row with an overlapping region width of N pixels, a set of weights W_A(x) that gradually decreases from 1 to 0 is assigned to the pixels from the left image (Patch_A), where x is the column index (from 0 to N-1) in the overlapping region. Correspondingly, a set of weights W_B(x) that gradually decreases from 0 to 1 is assigned to the pixels from the right image (Patch_B), and W_A(x) + W_B(x) = 1. For a pixel at coordinates (i, j) within the overlapping region (where i is the row index and j is the column index), its fused pixel value I_fused(i, j) is calculated as: I_fused(i, j) = W_A(j) * I_A(i, j) + W_B(j) * I_B(i, j), where I_A and I_B are the pixel values ​​of the two images in the overlapping region, respectively. This linear weighting ensures that the side of the overlapping region closer to the left image contributes mainly from the left image; the side closer to the right image contributes mainly from the right image; and in the middle region, the contributions from both are roughly equal, thus achieving a smooth transition. After performing this operation on each row of the overlapping region, a fused overlapping region image block is obtained. Finally, the exclusive portion of the left image, the fused overlapping region, and the exclusive portion of the right image are stitched together horizontally in sequence, thus completing the stitching of the adjacent image pair. By iteratively applying this process to all adjacent camera pairs in the circular array, while paying attention to the connection between the first and last cameras, a complete panoramic unfolded image at that depth position can be finally synthesized. By performing this step S14 independently for each acquisition depth, a series of panoramic unfolded images corresponding to different well depths can be obtained.

[0048] Step S15: Based on the mapping relationship and the depth position information, determine the overlapping area of ​​adjacent panoramic unfolded images along the well depth direction, and perform image fusion on the overlapping area to obtain a continuous unfolded image of the cylindrical well wall.

[0049] To address the issue of "segmented acquisition and overall presentation" in deep well or long-distance wellbore imaging, this invention integrates two different types of information: first, the pixel-height mapping relationship (i.e., the mapping relationship itself); and second, the depth position information bound to each image. By combining these two, the theoretically overlapping area in the depth direction of any two adjacent acquisition positions' panoramic unfolded images can be calculated. However, due to minor fluctuations in equipment movement, cable tension, or errors in the well depth measurement system itself, the theoretically calculated overlap position may deviate from the actual panoramic unfolded image content at the sub-pixel level. Therefore, based on rapid, large-scale positioning based on theoretical calculations, an image registration algorithm is further introduced for refinement. Finally, by smoothly fusing the overlapping areas of all adjacent depth image pairs, seams are eliminated, generating a visually continuous unfolded image.

[0050] In one optional implementation, step S15 specifically includes steps S151 to S153: Step S151: Based on the depth position information, determine the relative positional relationship of the panoramic unfolded images of each pair of adjacent depth positions in the vertical direction.

[0051] Based on a series of panoramic unfolded images arranged in the acquisition order, and the depth position information corresponding to each panoramic unfolded image, the sorted image sequence is traversed, and each panoramic unfolded image is paired with the next panoramic unfolded image to form an adjacent image pair. For each such image pair (denoted as Image_A and Image_B, where Image_A was acquired from a shallower depth Z_A and Image_B was acquired from a deeper depth Z_B), the theoretical relative displacement of the two panoramic unfolded images in the well depth direction can be calculated based on their recorded depth values. This relative displacement consists of two parts: first, the absolute depth difference ΔZ = Z_B - Z_A between the acquisition points of the two panoramic unfolded images; and second, the height H_image of the cylindrical well wall covered by a single panoramic unfolded image in this acquisition mode. Since the acquisition is continuous and the distance between adjacent acquisition points is usually less than the height of a single panoramic unfolded image (i.e., there is overlapping acquisition), the top of Image_B is located ΔZ below the top of Image_A in the well depth coordinate system. At the same time, the bottom of Image_A is located H_image below its top in the well depth coordinate system. This coordinate conversion allows two adjacent panoramic unfolded images to be placed under a unified well depth (vertical) coordinate axis, clarifying their vertical coverage relationship and approximate overlap range.

[0052] Step S152: Based on the mapping relationship and the relative positional relationship of the panoramic unfolded images of each pair of adjacent depth positions in the vertical direction, calculate the overlapping image area in the vertical direction of the panoramic unfolded images of each pair of adjacent depth positions.

[0053] Based on the mapping relationship, the depth difference ΔZ of each pair of adjacent depth positions in the panoramic unfolded images can be converted into the corresponding difference in the number of pixel rows on two adjacent panoramic unfolded images. Specifically, the theoretical overlap height Overlap_Height_meters can be calculated based on the coverage of the two panoramic unfolded images, and can be expressed as: Overlap_Height_meters = H_image - ΔZ. This means that in the depth direction, the top of the next panoramic unfolded image (Image_B) needs to extend upwards (represented downwards within the image) to connect with the bottom of the previous panoramic unfolded image (Image_A), and the length of this extension is the theoretical overlap height. Next, using the scaling factor in the mapping relationship, the number of pixel rows corresponding to the theoretical overlap height is calculated as: Overlap_Pixels = Overlap_Height_meters / Scale_factor.

[0054] For example, if Overlap_Height_meters = 0.3 meters and Scale_factor = 0.0005 meters / pixel, then Overlap_Pixels = 600 pixel rows. This means that, ideally, approximately 600 rows of pixels in the bottom region of Image_A and approximately 600 rows of pixels in the top region of Image_B depict the same cylindrical well wall. Based on the relative positional relationship determined in step S151, the starting row of the theoretical overlapping region (near the bottom of the image) can be determined in Image_A, and the starting row of the theoretical overlapping region (near the top of the image) can be determined in Image_B, along with their common overlap height expressed in pixel rows.

[0055] Step S153: Locate the actual position of the overlapping image area in the vertical direction of each pair of adjacent depth positions in the panoramic unfolded images, and fuse and stitch the overlapping image areas in the vertical direction of each pair of adjacent depth positions to obtain the continuous unfolded image of the cylindrical well wall.

[0056] Although the theoretical overlap area can be calculated based on geometric parameters and depth position information, in actual engineering, factors such as non-uniform motion of the hoisting system (the system used to record depth position information in real time), elastic tension or slight torsion of the cable, pulse counting errors or zero-point drift of the depth encoder, and environmental vibration can cause slight offsets of several to tens of pixels between the actual relative position of two adjacent images in the vertical direction and the theoretically calculated value. If the images are directly stitched together according to the theoretical position, texture misalignment and geometric breaks will inevitably occur at the seams, severely affecting the continuity and measurement accuracy of the panoramic image. Therefore, this invention adopts a strategy of "coarse positioning guidance and fine registration correction": First, the approximate overlap range is quickly locked based on theoretical calculations, significantly narrowing the search area; then, a high-precision digital image registration algorithm is applied within this range to automatically detect the optimal matching position between the contents of the two images, thereby compensating for system errors. Finally, the pixels in the precisely aligned overlap area are smoothly blended to eliminate brightness and color differences, completing the stitching. By performing this localization-fusion operation sequentially on all adjacent image pairs in the acquisition sequence and then connecting the results end to end, a continuous unfolded image covering the entire cylindrical well wall can be constructed segment by segment.

[0057] In an optional implementation, step S153 specifically includes steps S1531 to S1532: Step S1531: Based on the depth location information and the mapping relationship, determine the theoretical overlap height between each pair of adjacent panoramic unfolded images.

[0058] Based on the depth location information and mapping relationship, the process is performed for each pair of adjacent depth panoramic unfolded images (denoted as Image_A and Image_B, with acquisition depths Z_A and Z_B respectively, and Z_B>Z_A). First, the absolute depth difference between the acquisition points corresponding to the two panoramic unfolded images is calculated: ΔZ = Z_B - Z_A. The absolute depth difference is determined by the stop-shoot interval of the acquisition system. Second, the actual height H_image covered by a single panoramic unfolded image in the well depth direction is known. Based on simple set coverage relationships, since the acquisition is continuous and ΔZ is less than H_image, the top of Image_B must overlap with the bottom of Image_A. The actual physical height of this overlapping area (i.e., the theoretical overlap height) can be calculated using the formula: Overlap_Height_meters = H_image - ΔZ.

[0059] The physical meaning of the above formula is that Image_A covers a depth range of [Z_A, Z_A+H_image], Image_B covers a depth range of [Z_B, Z_B+H_image], and their intersection range is [Z_B, Z_A+H_image], the length of which is the theoretical overlap height. For example, if H_image = 0.8 meters and ΔZ = 0.5 meters, then Overlap_Height_meters = 0.3 meters. Finally, using the scaling factor in the mapping relationship, the theoretical overlap height is converted from physical units to image pixel units. The calculated theoretical overlap height is the number of pixel rows that the two panoramic unfolded images should share in the vertical direction under ideal, error-free conditions.

[0060] Step S1532: Within the theoretical overlap height between each pair of adjacent panoramic unfolded images, the template matching algorithm and the phase correlation algorithm are used to locate the actual position of the overlapping image area in the vertical direction of each pair of adjacent depth positions of the panoramic unfolded images.

[0061] Based on two adjacent panoramic unfolded images, Image_A and Image_B, and the theoretical overlap pixel height Overlap_Pixels calculated in step S1531, a horizontal image strip with a height of Overlap_Pixels and a width equal to the panoramic image is cropped near the bottom of Image_A, according to the theoretical relative position relationship, as a reference area or "template". Simultaneously, a search area with a height slightly larger than Overlap_Pixels is cropped near the top of Image_B. The next goal is to find a sub-region within the search area of ​​Image_B that best matches the template of Image_A.

[0062] To achieve this goal, this method employs two complementary image registration algorithms in synergy: Template matching algorithm: Operations are performed directly in the spatial domain. The template extracted from Image_A is slid across the search area of ​​Image_B, and the similarity metric (such as Normalized Cross-Correlation Coefficient, Squared Difference, and SSD) is calculated pixel-by-pixel. By traversing the entire search area, a position that maximizes the similarity metric (e.g., maximizing NCC) can be found; this is the initially estimated best matching position.

[0063] Phase correlation algorithm: This algorithm operates in the frequency domain. It is based on the translation property of the Fourier transform, where the spatial translation between two images is represented by a linear phase difference in the cross-power spectrum of their Fourier transforms. By calculating the Fourier transforms of candidate blocks within the Image_A template and Image_B search region, and analyzing the phase information of their cross-power spectra, the translation vector between the two images can be directly calculated.

[0064] First, a template matching algorithm is used for fast and robust coarse localization over a large search range, yielding a pixel-level matching position estimate. Since cylindrical well wall images may contain large areas of weak texture (such as smooth concrete surfaces) or repetitive textures, template matching alone may produce mismatches in some cases. Therefore, a phase correlation algorithm is then applied to refine the alignment within a smaller neighborhood, using the optimal position obtained from template matching as the center. The combination of template matching and phase correlation algorithms ensures both efficiency and robustness in a large search range while achieving extremely high alignment accuracy.

[0065] Through the above steps, the final output is the vertical offset, representing the number of pixels that Image_B needs to be adjusted vertically relative to Image_A to ensure complete alignment. Based on the vertical offset, the precise start and end rows of the overlapping area between the two images in actual pixel coordinates can be determined, i.e., the "actual position". Subsequently, a weighted fusion algorithm is applied to this overlapping image area to eliminate stitching seams. This process is iteratively applied to all adjacent image pairs, and the fused image segments are sequentially connected to generate a continuous panoramic image of the cylindrical well wall, completing the digital reconstruction of the entire cylindrical well wall.

[0066] Figure 5 This is a schematic diagram of the architecture of an image acquisition system suitable for cylindrical well walls provided by an embodiment of the present invention, as shown below. Figure 5 As shown, the system application performs the steps of the method described above, including a camera array module and a data processing module; the camera array module includes multiple cameras, which are arranged at the central axis of the cylindrical well wall and are evenly distributed in a ring; the central axis of the camera array module coincides with the central axis of the cylindrical well wall; The camera array module is used to acquire images of the cylindrical well wall at different depths, obtain multiple raw images, and record the depth position information at the time of acquisition.

[0067] To acquire complete, continuous, and comprehensive cylindrical wellbore image data, this invention employs a multi-camera circumferential array arrangement. Optionally, a four-camera circular array is used, with each camera uniformly distributed circumferentially along the central axis of the wellbore. The array's central axis coincides with the central axis of the cylindrical wellbore, ensuring consistent distance between each camera and the wellbore, thus maintaining uniform geometric scale during imaging. The camera array is mounted on a movable platform, such as a winch, lifting device, or traveling mechanism, enabling fixed-point image acquisition at different depths. During acquisition, movement is paused at preset acquisition intervals, triggering all cameras to simultaneously capture multiple raw images of the cylindrical wellbore at that depth. Each raw image is associated with the depth position information at the time of acquisition. This depth position information is recorded and stored in real time by an encoder or positioning system, using the wellhead as the reference zero point, recording the absolute depth or relative displacement value relative to the wellhead. This invention supports multiple acquisition modes, including continuous, segmented, and fixed-point acquisition, adapting to different well depths, diameters, and engineering requirements, achieving complete acquisition of images of the entire well section from shallow to deep layers.

[0068] The data processing module is used for: For each of the original images, the image pixels are mapped from the planar coordinate system to the cylindrical unfolded coordinate system through cylindrical back projection transformation to obtain multiple cylindrical unfolded images.

[0069] When the camera captures images inside the cylinder, a geometric inconsistency exists between its imaging plane and the curved surface of the cylindrical well wall, resulting in arc-shaped distortion of the image, characterized by center compression and edge stretching. This directly affects the accuracy of subsequent stitching and measurement. To solve this problem, this invention introduces the concept of cylindrical back-projection transformation, establishing a coordinate system from the image plane coordinate system. Cylindrical expansion coordinate system to the cylindrical well wall The mapping relationship. Specifically, firstly, based on the known geometric parameters, including the radius of the cylindrical well wall... Distribution radius of camera array Camera horizontal field of view Image width With height And the azimuth angle of the camera in the array, to construct the imaging geometry model. For each pixel in the original image The coordinates of the corresponding three-dimensional point on the cylindrical surface are calculated by back projection, and then the three-dimensional point is unfolded onto a two-dimensional plane to obtain its coordinates in the unfolded image of the cylinder. During the transformation, the horizontal direction is uniformly expanded according to the relationship between arc length and angle, while the vertical direction maintains a linear correspondence. This results in the cylindrical well wall texture, which originally exhibits arc-shaped distortion in the image, appearing as a rectangular area with uniform horizontal direction and proportional vertical direction in the corrected image. This eliminates visual geometric distortion and establishes a mathematical correspondence between pixel positions and the actual positions of the cylindrical well wall.

[0070] Based on the geometric parameters of the camera array and the cylindrical well wall, a mapping relationship is established between the image pixels of the cylindrical unfolded image and the actual height of the cylindrical well wall.

[0071] This involves establishing a deterministic and high-precision mathematical correspondence between the pixel coordinates of the digital image after cylindrical projection correction and the actual physical dimensions (vertical height or depth) of the cylindrical well wall. The principle is to use the geometric model of camera imaging and known, measurable geometric parameters of the system to derive the actual height of the cylindrical well wall arc segment represented by each row of pixels in the image through trigonometric calculations. In practice, a series of key geometric parameters must first be acquired or calibrated. Then, a linear or nonlinear transformation formula from the pixel row index to the real-world height is established in the vertical direction (i.e., the well depth direction), ultimately forming a lookup table or continuous function. This allows for the lookup or calculation of the height component of any pixel in the corrected image at its corresponding actual three-dimensional position on the cylindrical well wall.

[0072] For multiple cylindrical unfolded images at the same depth position, based on the geometric layout parameters of the camera array, the overlapping area of ​​the cylindrical unfolded images of adjacent cameras is calculated, and the overlapping area is fused to obtain panoramic unfolded images at different depth positions.

[0073] This invention precisely stitches together multiple images acquired by a circular camera array at the same depth (i.e., the same horizontal cross-section) and corrected for cylindrical surfaces into a complete, continuous panoramic unfolded image. Traditional stitching methods based on feature point matching are unreliable in environments with weak texture, high repeatability, and potential for reflection or dust, such as cylindrical well walls, and are prone to misalignment due to residual geometric distortion. This invention abandons the traditional feature-dependent path and instead fully utilizes the system's inherent geometric prior knowledge, namely geometric layout parameters. Specifically, based on the geometric layout parameters of the camera array, the angular range corresponding to the theoretical overlap area of ​​adjacent camera fields of view on the cylindrical well wall is calculated. Then, the corresponding pixel regions in each corrected cylindrical unfolded image are further located, thus achieving "computational localization" of the overlapping image region rather than "search matching". Finally, image fusion technology is used to smooth the transition of pixels in the overlapping region, eliminating stitching boundary traces and ensuring consistency of circumferential brightness and texture. By processing all adjacent image pairs in the array sequentially, the local cylindrical unfolded images of all cameras at the same depth can be combined into a panoramic unfolded image covering the entire circumference of the cylindrical well wall.

[0074] Based on the mapping relationship and the depth location information, the overlapping area of ​​adjacent panoramic unfolded images along the well depth direction is determined, and the overlapping area is image fused to obtain a continuous unfolded image of the cylindrical well wall.

[0075] To address the issue of "segmented acquisition and overall presentation" in deep well or long-distance wellbore imaging, this invention integrates two different types of information: first, the pixel-height mapping relationship (i.e., the mapping relationship itself); and second, the depth position information bound to each image. By combining these two, the theoretical overlap area that inevitably exists in the depth direction between any two adjacent acquisition locations can be calculated. However, due to minor fluctuations in equipment movement, cable tension, or errors in the well depth measurement system itself, the theoretically calculated overlap position may deviate from the actual image content at the sub-pixel level. Therefore, based on rapid, large-scale positioning based on theoretical calculations, an image registration algorithm is further introduced for refinement. Finally, by smoothly fusing the overlap areas of all adjacent depth image pairs, seams are eliminated, generating a visually continuous panoramic unfolded image.

[0076] In one optional implementation, the horizontal field of view of the camera, the distribution radius of the camera array module, and the radius of the cylindrical well wall satisfy the following conditions: The distance from the intersection of the center lines of the fields of view of two adjacent cameras on the cylindrical well wall to the central axis of the camera array module is less than the radius of the cylindrical well wall, so that the fields of view of adjacent cameras have overlapping image areas on the cylindrical well wall.

[0077] Consider two adjacent cameras in a camera array, whose optical axes (i.e., the center lines of their fields of view) are distributed at a certain angle in the horizontal plane (for N uniformly distributed cameras, this angle is 360° / N). The two optical axes point to the cylindrical well wall and intersect at a single point. This intersection point is not on the surface of the cylindrical well wall, but rather in the space of the extended optical axes. The key geometric relationship is that the distance from the central axis of the camera array (i.e., the center line of the well) to the intersection point of the two optical axes is less than the radius R of the cylindrical well wall. Physically, this means that only when the two optical axes intersect before reaching the surface of the cylindrical well wall can their respective field-of-view cones intersect on the surface of the cylindrical well wall, thus forming an overlapping area observed simultaneously by both cameras. Conversely, if the intersection point is located outside the cylindrical well wall surface (i.e., the distance from the intersection point to the central axis is greater than or equal to R), it means that the fields of view of the two cameras are exactly touching or disjoint on the cylindrical well wall surface. This will result in no overlapping information available during stitching, thus creating gaps in the final panoramic image or relying on unreliable inferences to fill them, disrupting the continuity and geometric integrity of the image. This fundamental condition is equivalent to requiring that the intersection of the fields of view of adjacent cameras must effectively cover the cylindrical well wall. In practice, based on the known cylindrical well wall radius R and the selected camera horizontal field of view angle α, the maximum allowable value of the camera array distribution radius r is determined by calculation, or based on the existing camera array distribution radius r and the horizontal field of view angle α of a single camera, the minimum adaptable cylindrical well wall radius R is verified.

[0078] This invention also provides an electronic device, including a processor, a memory, and a computer program stored in the memory and capable of running on the processor. When the computer program is executed by the processor, it implements the various processes described above in an embodiment of an image acquisition method applicable to cylindrical well walls, and achieves the same technical effect. To avoid repetition, it will not be described again here.

[0079] This invention also provides a computer-readable storage medium storing a computer program. When executed by a processor, the computer program implements the various processes described above in an embodiment of an image acquisition method suitable for cylindrical well walls, achieving the same technical effects. To avoid repetition, these processes will not be repeated here. The various embodiments in this specification are described in a progressive manner, with each embodiment focusing on its differences from other embodiments. Similar or identical parts between embodiments can be referred to mutually.

[0080] Although preferred embodiments of the present invention have been described, those skilled in the art, upon learning the basic inventive concept, can make other changes and modifications to these embodiments. Therefore, the appended claims are intended to be interpreted as including the preferred embodiments as well as all changes and modifications falling within the scope of the embodiments of the present invention.

[0081] Finally, it should be noted that in this document, relational terms such as "first" and "second" are used only to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or terminal device that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or terminal device. Without further limitations, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or terminal device that includes said element.

[0082] The embodiments of the present invention have been described above with reference to the accompanying drawings. However, the present invention is not limited to the specific embodiments described above. The specific embodiments described above are merely illustrative and not restrictive. Those skilled in the art can make many other forms under the guidance of the present invention without departing from the spirit and scope of the claims. All of these forms are within the protection scope of the present invention.

Claims

1. An image acquisition method suitable for cylindrical well walls, characterized in that, include: By using a camera array deployed along the circumference at different depth positions, images of the cylindrical well wall are acquired, resulting in multiple raw images, and the depth position information at the time of acquisition is recorded. For each of the original images, the image pixels are mapped from the planar coordinate system to the cylindrical unfolded coordinate system through cylindrical back projection transformation to obtain multiple cylindrical unfolded images; Based on the geometric parameters of the camera array and the cylindrical well wall, a mapping relationship is established between the image pixels of the cylindrical unfolded image and the actual height of the cylindrical well wall; For multiple cylindrical unfolded images at the same depth position, based on the geometric layout parameters of the camera array, the overlapping area of ​​the cylindrical unfolded images of adjacent cameras is calculated, and the overlapping area is fused to obtain panoramic unfolded images at different depth positions. Based on the mapping relationship and the depth location information, the overlapping area of ​​adjacent panoramic unfolded images along the well depth direction is determined, and the overlapping area is image fused to obtain a continuous unfolded image of the cylindrical well wall.

2. The method according to claim 1, characterized in that, Based on the geometric parameters of the camera array and the cylindrical well wall, a mapping relationship is established between the image pixels of the unfolded cylindrical image and the actual height of the cylindrical well wall, including: The geometric parameter information is determined, including: the vertical field of view of a single camera in the camera array, the horizontal field of view of a single camera, the distribution radius of the camera array, the radius of the cylindrical well wall, and the height of the cylindrical unfolded image; In the vertical direction, the actual height of the cylindrical well wall corresponding to the cylindrical unfolded image is determined based on the vertical field of view of the single camera, the radius of the cylindrical well wall, and the distribution radius of the camera array. Based on the height of the cylindrical unfolded image and the actual height of the cylindrical well wall corresponding to the cylindrical unfolded image, a mapping relationship is established between the actual height of the cylindrical well wall corresponding to each pixel in the cylindrical unfolded image.

3. The method according to claim 2, characterized in that, For multiple cylindrical unfolded images at the same depth position, based on the geometric layout parameters of the camera array, the overlapping region of the cylindrical unfolded images of adjacent cameras is calculated, and the overlapping region is fused to obtain panoramic unfolded images at different depth positions, including: Based on the horizontal field of view of the single camera, the distribution radius of the camera array, and the radius of the cylindrical well wall, calculate the range of the central angle corresponding to the overlapping image area of ​​the images acquired by adjacent cameras in the camera array on the cylindrical well wall; Based on the central angle range, the corresponding overlapping image regions are extracted from two adjacent cylindrical unfolded images at the same depth position; Using a weighted fusion algorithm, the overlapping image regions extracted from two adjacent cylindrical unfolded images at the same depth position are fused to obtain panoramic unfolded images at different depth positions.

4. The method according to claim 3, characterized in that, Based on the horizontal field of view of the single camera, the distribution radius of the camera array, and the radius of the cylindrical well wall, the range of central angles corresponding to the overlapping image regions on the cylindrical well wall captured by adjacent cameras in the camera array is calculated, including: Based on the horizontal field of view of the single camera, the first central angle corresponding to the horizontal imaging range of the single camera on the cylindrical well wall is determined; Based on the distribution radius of the camera array and the radius of the cylindrical well wall, calculate the second central angle corresponding to the imaging area that does not overlap with the adjacent camera within the horizontal imaging range of the single camera; Based on the first central angle and the second central angle, the range of central angles corresponding to the overlapping image areas of the images captured by adjacent cameras in the camera array on the cylindrical well wall is calculated.

5. The method according to claim 2, characterized in that, Based on the mapping relationship and the depth location information, the overlapping area of ​​adjacent panoramic unfolded images along the well depth direction is determined, and image fusion is performed on the overlapping area to obtain a continuous unfolded image of the cylindrical well wall, including: Based on the depth position information, determine the relative positional relationship of the panoramic unfolded image of each pair of adjacent depth positions in the vertical direction; Based on the mapping relationship and the relative positional relationship of the panoramic unfolded images of each pair of adjacent depth positions in the vertical direction, calculate the overlapping image area of ​​the panoramic unfolded images of each pair of adjacent depth positions in the vertical direction. In the panoramic unfolded images at different depth positions, the actual position of the overlapping image area in the vertical direction of each pair of adjacent depth positions is located, and the overlapping image areas in the vertical direction of each pair of adjacent depth positions are fused and stitched together to obtain the continuous unfolded image of the cylindrical well wall.

6. The method according to claim 5, characterized in that, Locating the actual position of the overlapping image region in the vertical direction of the panoramic unfolded images at each pair of adjacent depth positions in the panoramic unfolded images includes: Based on the depth location information and the mapping relationship, the theoretical overlap height between each pair of adjacent panoramic unfolded images is determined; Within the theoretical overlap height between each pair of adjacent panoramic unfolded images, template matching and phase correlation algorithms are used to locate the actual position of the overlapping image area in the vertical direction of each pair of adjacent depth positions of the panoramic unfolded images.

7. An image acquisition system suitable for cylindrical well walls, characterized in that, The method is applied to the steps of performing the method as described in any one of claims 1-6; it includes a camera array module and a data processing module; the camera array module includes a plurality of cameras, which are disposed at the central axis of the cylindrical well wall and are uniformly distributed in a ring; the central axis of the camera array module coincides with the central axis of the cylindrical well wall; The camera array module is used to acquire images of the cylindrical well wall at different depth positions, obtain multiple raw images, and record the depth position information at the time of acquisition. The data processing module is used for: For each of the original images, the image pixels are mapped from the planar coordinate system to the cylindrical unfolded coordinate system through cylindrical back projection transformation to obtain multiple cylindrical unfolded images; Based on the geometric parameters of the camera array and the cylindrical well wall, a mapping relationship is established between the image pixels of the cylindrical unfolded image and the actual height of the cylindrical well wall; For multiple cylindrical unfolded images at the same depth position, based on the geometric layout parameters of the camera array, the overlapping area of ​​the cylindrical unfolded images of adjacent cameras is calculated, and the overlapping area is fused to obtain panoramic unfolded images at different depth positions. Based on the mapping relationship and the depth location information, the overlapping area of ​​adjacent panoramic unfolded images along the well depth direction is determined, and the overlapping area is image fused to obtain a continuous unfolded image of the cylindrical well wall.

8. The system according to claim 7, characterized in that, The horizontal field of view of the camera, the distribution radius of the camera array module, and the radius of the cylindrical well wall satisfy the following conditions: The distance from the intersection of the center lines of the fields of view of two adjacent cameras on the cylindrical well wall to the central axis of the camera array module is less than the radius of the cylindrical well wall, so that the fields of view of adjacent cameras have overlapping image areas on the cylindrical well wall.

9. An electronic device, characterized in that, include: A processor, a memory, and a computer program stored in the memory and capable of running on the processor, wherein the computer program, when executed by the processor, implements the steps of the method as described in any one of claims 1-6.

10. A computer-readable storage medium, characterized in that, A computer program is stored on the computer-readable storage medium, which, when executed by a processor, implements the steps of the method as described in any one of claims 1-6.

Images