Device and method for acquiring front view image of underground in-situ perforation

Abstract

The invention discloses a device and a method for acquiring an underground in-situ perforation front-view image. The device comprises a front-view video acquisition unit, a side-view video acquisition unit and a control unit, the foresight video acquisition unit is arranged at the front end of the main body and comprises an ultra-wide-angle camera for acquiring a front panoramic image; the side-view video acquisition unit is arranged in the middle of the main body, and an installed side-view camera acquires perforation front-view images of the inner wall of a shaft; the control unit is arranged at the tail part of the main body to realize positioning of the side-view camera and picture recognition and storage; a three-camera scheme of the front foresight ultra-wide-angle lens and the two rear rotary side-view cameras is adopted, the perforation orientation is determined through the front camera, the rotary side-view cameras are aligned with the perforation orientation, front-view observation of a perforation is achieved, the point-shaped light source is used for assisting, and the perforation orientation can be observed through the point-shaped light source. And the distance between the perforation and the side view camera is measured by adopting a triangulation principle, so that the subsequent perforation accurate measurement work is realized.

Description

Technical Field

[0001] The present invention relates to the technical field of evaluation of fracturing effects of unconventional oil and gas horizontal wells in petroleum and natural gas development, and in particular to a device and method for obtaining downhole in-situ perforation elevation images. Technical Background

[0002] The U.S. shale oil revolution has achieved large-scale and efficient development of shale oil and gas. The use of ultra-long horizontal well drilling and large-scale hydraulic fracturing technology has greatly increased single-well production and reduced development costs.

[0003] my country has a high degree of dependence on foreign crude oil, but rich shale oil and gas reserves. Vigorously promoting the technological revolution in shale oil and gas development and realizing the scale and efficiency development of shale oil and gas in my country are important guarantees for my country to improve its resource self-sufficiency rate and hold its energy security in its own hands.

[0004] my country's fracturing equipment and tools are becoming increasingly mature. Currently, the key to optimizing fracture control in fracturing processes lies in monitoring and evaluating fracturing effectiveness. Monitoring the erosion of perforations after fracturing through downhole video, measuring the size of the eroded holes, and analyzing sand injection and fracture extension distribution are cutting-edge technologies for monitoring fracturing effectiveness.

[0005] EV uses an array side-view lens structure, capturing images of perforation wear for analysis and measurement using four cameras spaced evenly around the circumference. This captures images of the boreholes within the wellbore and then performs statistical analysis. Xi'an Zhengyuan Well Image uses a front-mounted ultra-wide-angle lens to capture an overall image of the pipe wall to monitor and analyze the wear and size of the perforations. For example, the downhole rotary zoom visualization detection device with announcement number CN221531559U has a camera mounted on top, capable of 90° pitch and 360° circumferential rotation, and can only capture wellbore information in front. Since the ideal observation condition for visual monitoring of perforation wear is for the camera to be facing the center of the perforation, the current technical solutions of EV and Zhengyuan Well Image are unable to achieve direct observation of all perforations. Perforation distortion cannot accurately correct perforation measurement errors, affecting observation results and measurement accuracy. Summary of the Invention

[0006] In order to overcome the above-mentioned defects of the prior art, the purpose of the present invention is to provide a device and method for obtaining a front-view image of an in-situ perforation in a well. The device adopts a three-camera solution consisting of a front-facing ultra-wide-angle lens and two rear-facing rotating side-view cameras. The front camera is used to determine the perforation orientation, and the rotating side-view camera is aligned with the perforation orientation to achieve front-view observation of the perforation. With the assistance of a point light source, the distance between the perforation and the side-view camera is measured using the triangulation principle to achieve subsequent precise measurement of the perforation.

[0007] In order to achieve the above object, the technical solution adopted by the present invention is:

[0008] A device for obtaining downhole in-situ perforation front-view images, which can realize front-view shooting of perforations in all directions, that is, the center of the camera can be aligned with the center of the perforation for shooting, comprising a front-view video acquisition unit 1, a side-view video acquisition unit 2 and a control unit 3;

[0009] The forward-view video acquisition unit 1 is located at the front end of the device and includes an ultra-wide-angle camera 1-1 to capture a panoramic image of the front;

[0010] The side view video acquisition unit 2 is located in the middle of the device, and collects the front view image of the perforation on the inner wall of the wellbore through the side view camera 2-8 installed on its rotating frame 2-3;

[0011] The control unit 3 is located at the rear of the device and includes a power supply system, a main control module, a motor drive module, an AI intelligent recognition module and a video storage module; the power supply system provides working power, the motor drive module controls the motor in the side view video acquisition unit 2, the AI ​​intelligent recognition module identifies the perforation targets in the images captured by the front view and side view video acquisition units and outputs the perforation target position information; the main control module determines the angle at which the side view video acquisition unit 2 needs to be rotated according to the perforation position identified by the AI ​​intelligent recognition module, and locks the target perforation position of the side view; the video storage module stores the captured front view and side view high-definition video images, and supports real-time reading and storage working modes.

[0012] The ultra-wide-angle camera 1-1 has a field of view greater than 180°, and light sources 1-2 are evenly distributed around the ultra-wide-angle camera 1-1 to provide wellbore lighting.

[0013] The specific structure of the side-view video acquisition unit 2 is as follows: a fixed axis 2-1 is provided at the center of the device body, the fixed axis 2-1 is connected to the front and rear ends of the device body, a through hole for passing a wire is provided at the center of the fixed axis 2-1, two sets of bearings 2-2 and 2-14 are provided outside the fixed axis 2-1, a rotating skeleton 2-3 is provided outside the two sets of bearings 2-2 and 2-14, a large gear 2-6 is provided at one end of the rotating skeleton 2-3, the large gear 2-6 is driven to rotate by the small gear 2-4, the small gear 2-4 is connected to the rotating shaft of the motor 2-5, a slip ring 2-7 is provided at the other end of the rotating skeleton 2-3, the slip ring 2-7 is used to pass through the internal lead of the rotating skeleton 2-3, a side-view camera 2-8 is installed on the outer surface of the rotating skeleton 2-3, and lighting lamp beads 2-10 are installed on the outer shells 2-11 on both sides of the side-view camera 2-8. An annular transparent window 2-13 is installed outside the camera 2-8, and a circular transparent window 2-12 is installed outside the lamp beads 2-10.

[0014] The number of the side-view cameras 2-8 installed is 1 to 4.

[0015] A small bearing 2-15 is installed between the rotating shaft of the motor 2-5 and the housing 2-11.

[0016] The AI ​​intelligent recognition module integrates a perforation recognition model. The perforation recognition model is trained using a single-stage neural network with a large number of perforation data sets. It can identify perforation targets in images captured by the front-view and side-view video acquisition units, and calculate and output the perforation target phase information in the front-view image based on the forward-view perforation recognition target, that is, the perforation phase angle. The calculation method is as follows:

[0017] Perforations are identified in the front view image. Perforation contour data is obtained through image processing, contour edge detection, and contour screening. The minimum circumscribed rectangle of the contour is calculated, and the center of the rectangle is used as the center B of the perforation shape. The image center point is A. The angle θ between the line connecting the image center A and the perforation center point B and the perpendicular line passing through the image center point A is calculated. θ is the perforation phase angle.

[0018] The main control module calculates the angle that the side view video acquisition unit needs to rotate according to the perforation target phase of the front view acquisition unit identified by the AI ​​intelligent recognition module. When the side view camera is located at the phase angle α, the angle β that needs to be rotated is β=a-θ, where θ is the perforation phase angle; and outputs a motor control signal through the motor drive module to control the side view camera 2-8 of the side view video acquisition unit 2 to rotate to the target perforation orientation and lock the target perforation position.

[0019] A method for obtaining downhole in-situ perforation elevation images based on the above device includes the following steps:

[0020] Step 1: Calibrate side view cameras 2-8 in side view video acquisition unit 2

[0021] Use the side view video acquisition unit 2 to shoot more than 10 images of the checkerboard calibration plate from different angles and distances. Use the calibration software to automatically identify the corner points of the calibration plate, obtain the image coordinates, establish the relationship between the world coordinates and the image coordinates, and solve the internal parameter matrix and distortion coefficient of the side view camera 2-8;

[0022] Step 2: Lower the device for acquiring downhole in-situ perforation front-view images into the perforation section. Step 3: Acquire forward-view panoramic images of the wellbore.

[0023] After entering the perforation section, the device for obtaining downhole in-situ perforation front-view images is placed in the center of the wellbore and moves forward at a constant speed along the central axis of the wellbore at a speed of less than 3m / min. The forward-view video acquisition unit 1 obtains a 360° panoramic image of the wellbore through the ultra-wide-angle camera 1-1;

[0024] Step 4: Image perforation target recognition

[0025] The AI ​​intelligent recognition module identifies the target perforation orientation based on the wellbore image acquired by the forward-view video acquisition unit 1 using its built-in perforation recognition model;

[0026] Step 5: Side view camera locks target perforation position

[0027] Since the relative positions of the side-view cameras 2-8 and the ultra-wide-angle camera 1-1 are known, the side-view cameras 2-8 are rotated according to the target perforation phase in the front view image provided by the AI ​​intelligent recognition module, and the corresponding side-view cameras 2-8 are controlled to rotate to the target perforation position on the side, thereby locking the target perforation position;

[0028] Step 6: The side-view camera collects the perforation front view image

[0029] The center of the side-view camera 2-8 is aligned with the center of the target perforation position. When the side-view camera 2-8 moves to the target perforation position, it realizes front-view observation of the perforation and collects the front-view image of the perforation.

[0030] Step 7: Correct the distortion of the perforation front view image and calculate the perforation size

[0031] The orthographic image is corrected using the distortion coefficient, and the actual perforation size of the target in the corrected undistorted image is calculated in combination with the distance between the target and the camera.

[0032] The specific calculation method of the seventh step is as follows:

[0033] (7.1) First, the pixel coordinates (μ, v) of the orthographic image are converted to normalized image coordinates using formula (1):

[0034]

[0035] Among them, K is the camera intrinsic parameter matrix f x , f y is the focal length, in pixels, c x , c y is the main point coordinate;

[0036] (7.2) Use the distortion coefficient to correct the normalized coordinates and obtain the normalized coordinates without distortion (x c ,y c ), assuming that the ideal undistorted image coordinates are (x, y), and the actual coordinates after distortion are (x′, y′), then the correction formula for radial distortion is:

[0037] x′=x·(1+k1r 2 +k2r 4 +k3r 6 +…) Formula (2)

[0038] y′=y·(1+k1r2 +k2r 4 +k3r 6 +…) Formula (3)

[0039] Among them, r 2 =x 2 +y 2 , the square of the distance from the pixel to the image center; k is the radial distortion coefficient, where k1 is the first order, k2 is the second order), k3 is the third order), and k... are high-order coefficients;

[0040] The correction formula for tangential distortion is:

[0041] x′=x+[2p1xy+p2(r 2 +2x 2 )] Formula (4)

[0042] y′=y+[p1(r 2 +2y 2 )+2p2xy] Formula (5)

[0043] Among them, the tangential distortion coefficients are p1 and p2, which are used to correct the linear deviation caused by plane tilt;

[0044] (7.3) The corrected normalized coordinates are converted back to pixel coordinates (u c , v c ),

[0045]

[0046] Among them, K is the camera intrinsic parameter matrix ( f x , f y is the focal length (pixel unit), c x , c y is the main point coordinate);

[0047] (7.4) Given that the inner diameter of the detection casing is D, the acquisition device for obtaining the perforation image moves along the central axis of the casing. Given that the center distance of the side-view camera is L, calculate the distance Z between the target on the inner surface of the casing and the camera:

[0048]

[0049] The actual size of the target, width W and height H, are calculated using the principle of similar triangles. The relationship between pixel size and actual size is as follows:

[0050]

[0051] Among them, w pix is the width of the target in the image, h pixis the height of the target in the image, fx, fy are the focal lengths in the camera intrinsic parameters, in pixels; Z is the distance from the target to the camera, which is the relationship between the target size and the actual size;

[0052] (7.5) Based on the relationship between the target size and the actual size in the image, the inner circumference and outer circumference of the perforation hole are obtained by calculating the total length of the inner wall contour and the outer wall contour boundary in the image after distortion correction of the perforation front view image. The inner area and outer area are obtained by counting the number of pixels within the inner wall contour and the outer wall contour and the relationship between the pixel size and the actual size. The area value can be used to calculate the average aperture value of the equal area circle using the circle area formula as the average aperture value of the perforation hole, thereby realizing the measurement of the perforation size.

[0053] When the device for obtaining the downhole in-situ perforation front-view image in step 2 is lowered to the perforation section, if it is lowered by a cable device, the device for obtaining the downhole in-situ perforation front-view image is connected to a high-speed cable to achieve real-time observation of downhole video images on the ground;

[0054] If the device is lowered by a cableless device, the power supply system in the control unit 3 of the device for obtaining downhole in-situ perforation elevation images is started, downhole images are collected and stored, and after the collection is completed, the device is lifted to the surface to download the video images.

[0055] Compared with the prior art, the advantages of the present invention are:

[0056] 1. The present invention adopts a front-facing ultra-wide-angle lens and a side-view camera in the middle of the main body, which can obtain a 360° panoramic image from the front lens and a side-view perforation front-view image, enabling a comprehensive understanding of the wellbore condition and realizing perforation front-view photography, providing an image basis for accurate quantitative measurement of perforations.

[0057] 2. A small bearing 2-15 is installed between the rotating shaft of the motor 2-5 of the present invention and the housing 2-11, which can reduce the rotational friction.

[0058] 3. The present invention fixes the axis 2-1, two sets of bearings 2-2 and 2-14, the large gear 2-6 and the small gear 2-4, and the assembly of the rotating gear 2-3, so that the front view video acquisition unit 1 is fixed and the side view video acquisition unit 2 can rotate relative to it, ensuring that the side view relative rotation reference remains unchanged, thereby achieving accurate positioning of the perforating phase of the side view video acquisition unit 2.

[0059] 4. The AI ​​intelligent recognition module of the present invention can automatically identify in-situ perforations underground, enabling the device to perform intelligent autonomous control underground, realizing fully autonomous decision-making and control of the system, and realizing the acquisition of perforation front-view images.

[0060] In summary, the present invention can obtain orthographic images of perforations in all directions, improve the quality of perforation observation images, reduce image distortion, accurately measure the distance between the perforation and the test camera, accurately calibrate the image, and realize the subsequent precise measurement of perforation dimensions. BRIEF DESCRIPTION OF THE DRAWINGS

[0061] Figure 1 It is a structural diagram of the device of the present invention.

[0062] Figure 2 This is the block diagram of the control unit.

[0063] Figure 3 This is the flow chart of perforation recognition model training.

[0064] Figure 4 It is a schematic diagram of the relationship between the side-view camera, casing and perforation positions.

[0065] Figure 5 It is a schematic diagram of the specific structure of the side view video acquisition unit 2.

[0066] Figure 6 It is a schematic diagram of the present invention for transporting to the perforation section.

[0067] Figure 7 This is the principle diagram of forward-looking perforation phase calculation DETAILED DESCRIPTION

[0068] The present invention will be described in further detail below with reference to the accompanying drawings.

[0069] Combine Figure 1 A device for obtaining front-view images of in-situ perforations in a well, which can realize front-view shooting of perforations in all directions, that is, the center of the camera can be aligned with the center of the perforation for shooting, includes a front-view video acquisition unit 1, a side-view video acquisition unit 2 and a control unit 3.

[0070] The forward-looking video acquisition unit 1 collects 360-degree panoramic images of the wellbore, including an ultra-wide-angle camera 1-1 located at the front end of the entire device; the field of view of the ultra-wide-angle camera 1-1 is greater than 180°, and light sources 1-2 are evenly distributed around the ultra-wide-angle camera 1-1 to provide wellbore lighting. The number of light sources is 6 to 9 according to the outer diameter of the device. The camera and the light source are made of transparent pressure-bearing material around them, and a pressure-bearing sealing structure is used between the connecting parts of the outer shell. Its structure and appearance must meet the environmental requirements of the wellbore size, temperature resistance, pressure resistance, etc. of the measurement depth.

[0071] See also Figure 1 and Figure 5The side-view video acquisition unit 2 is located in the middle of the device. A fixed axis 2-1 is provided in the center of the device body. The fixed axis 2-1 is connected to the front and rear ends of the device body. A through hole is provided in the center of the fixed axis 2-1 for passing a wire. Two sets of bearings 2-2 and 2-14 are provided outside the fixed axis 2-1. A rotating skeleton 2-3 is provided outside the bearings 2-2 and 2-14. The rotating skeleton 2-3 can rotate around the fixed axis 2-1, and the outer shell 2-11 does not rotate. A large gear 2-6 is provided at one end of the rotating skeleton 2-3. The large gear 2-6 is driven to rotate by the small gear 2-4. The small gear 2-4 is connected to the rotating shaft of the motor 2-5 and is driven to rotate by the motor 2-5. A small bearing 2-15 is installed between the rotating shaft of the motor 2-5 and the outer shell 2-11 to reduce rotational friction. A slip ring 2-7 is provided at the other end of the rotating skeleton 2-3. The slip ring 2-7 is used to pass the internal lead wires passing through the rotating skeleton 2-3. The rotating skeleton 2-3 will not cause the connecting wires to be twisted off during rotation. The rotating skeleton 2-3 has a circular, hollow interior and one to four flat surfaces on the exterior. Each surface is mounted with a side-view camera 2-8. The housing 2-11 on either side of the side-view camera 2-8 is provided with eight independent lamp mounting holes 2-9. Lamp beads 2-10 are installed in these mounting holes 2-9 to illuminate the shaft. A circular transparent window 2-13 is mounted outside the camera 2-8, and a circular transparent window 2-12 is mounted outside the lamp beads 2-10.

[0072] See also Figure 2 The control unit 3, located at the rear of the device, includes a power supply system, a main control module, a motor drive module, an AI intelligent recognition module, and a video storage module. The power supply system provides operating power, while the motor drive module controls the motors in the side-view video acquisition unit 2. The AI ​​intelligent recognition module integrates a perforation recognition model trained using a single-stage neural network with a large number of perforation datasets. It can identify perforation targets in images captured by the front-view and side-view video acquisition units and output perforation target location information. The video storage module stores the captured front-view and side-view high-definition video images and supports real-time reading and storage. The main control module calculates the required rotation angle of the side-view video acquisition unit based on the perforation target phase of the front-view acquisition unit identified by the AI ​​intelligent recognition module. If the side-view camera is at phase angle α, the required rotation angle β = ɑ - θ, where θ is the perforation phase angle. The motor drive module then outputs motor control signals to rotate the side-view cameras 2-8 of the side-view video acquisition unit 2 to the target perforation orientation, locking the target perforation location.

[0073] A method for obtaining downhole in-situ perforation elevation images based on the above device includes the following steps:

[0074] Step 1: Calibration of side view cameras 2-8 in side view video acquisition unit 2

[0075] After the side-view video acquisition unit 2 is installed, fix the side-view video acquisition unit 2 and shoot more than 10 images of the checkerboard calibration plate from different angles and distances. Use calibration software (such as OpenCV, Matlab Camera Calibrator Toolbox) to automatically identify the corner points of the calibration plate, obtain image coordinates, establish the relationship between world coordinates and image coordinates through perspective projection transformation, and combine the distortion model to solve the internal parameter matrix K and distortion coefficients k1, k2, k3, p1, p1 of the side-view camera 2-8;

[0076] Step 2: Transport the downhole in-situ perforation shape and size measurement device to the perforation section, refer to Figure 6 The device for measuring the in-situ shape and size of downhole perforations is lowered together with the coiled tubing supporting equipment on the well, the downhole coiled tubing, the straightening device, the cable, etc., and the device is transported to the perforation section in the wellbore;

[0077] When the cable equipment is lowered into the well in step 2, the device is connected to a high-speed communication unit, and the ground can observe the downhole video image in real time;

[0078] When the cable-free device is lowered, the power supply system in the control unit 3 of the device is started, and the downhole images are collected and stored. After the collection is completed, the device is lifted to the surface to download the video images.

[0079] Step 3: Acquisition of panoramic images of the wellbore

[0080] After the device for obtaining downhole in-situ perforation front view images enters the perforation section, refer to Figure 4 , placed in the center of the wellbore, and moving at a constant speed along the central axis of the wellbore at a speed of less than 3m / min. The forward video acquisition unit 1 obtains a 360° panoramic image of the wellbore through the ultra-wide-angle camera 1-1;

[0081] Step 4: Image perforation target recognition

[0082] The AI ​​intelligent recognition module accurately identifies the target perforation location and calculates the perforation direction based on the wellbore image obtained by forward viewing through the built-in perforation recognition model.

[0083] Reference Figure 3 The perforation recognition model is obtained by preparing a downhole in-situ perforation image dataset, training a YOLO model, and then optimizing parameters according to the training results.

[0084] Step 5: Side view camera locks target perforation position

[0085] Since the relative positions of the side-view camera 2-8 and the ultra-wide-angle camera 1-1 are known, the rotation angle of the side-view camera 2-8 is calculated according to the target perforation position provided by the AI ​​intelligent recognition module, and the corresponding side-view camera 2-8 is controlled to rotate to the target perforation position, thereby locking the target perforation position;

[0086] The target perforation orientation is calculated as follows:

[0087] Perforations are identified in the front view image. Perforation contour data is obtained through image processing, contour edge detection, and contour screening. The minimum circumscribed rectangle of the contour is calculated, and the center of the rectangle is used as the center B of the perforation shape. The image center point is A. The angle θ between the line connecting the image center A and the perforation center point B and the perpendicular line passing through the image center point A is calculated. θ is the perforation phase angle.

[0088] Step 6: The side-view camera collects the perforation front view image

[0089] The center of the side-view camera 2-8 is aligned with the center of the target perforation position. When the side-view camera 2-8 moves to the target perforation position, it realizes front-view observation of the perforation and collects the front-view image of the perforation.

[0090] Step 7: Correct the distortion of the perforation front view image and calculate the perforation size

[0091] Since lens distortion will cause the target pixel size to be distorted, the image must first be corrected using the distortion coefficient to obtain an ideal image without distortion.

[0092] First, the pixel coordinates (μ, v) of the orthographic image are converted to normalized image coordinates using formula (1):

[0093]

[0094] Among them, K is the camera intrinsic parameter matrix f x , f y is the focal length, in pixels, c x , c y The main point coordinates.

[0095] Secondly, the normalized coordinates are corrected using the distortion coefficients to obtain the normalized coordinates without distortion (x c ,y c ), assuming that the ideal undistorted image coordinates are (x, y) (normalized coordinates, i.e., the influence of internal references has been removed), and the actual coordinates after distortion are (x′, y′), then the correction formula for radial distortion is:

[0096] x′=x·(1+k1r 2 +k2r 4 +k3r 6 +…) Formula (2)

[0097] y′=y·(1+k1r 2 +k2r 4 +k3r 6 +…) Formula (3)

[0098] Among them, r 2 =x 2 +y 2 , the square of the distance from the pixel to the image center; radial distortion coefficient: k1 (first order), k2 (second order), k3 (third order), higher order coefficients such as k4, k5, k6... are used for strong distortion scenes such as fisheye lenses;

[0099] The correction formula for tangential distortion is:

[0100] x′=x+[2p1xy+p2(r 2 +2x 2 )] Formula (4)

[0101] y′=y+[p1(r 2 +2y 2 )+2p2xy] Formula (5)

[0102] Among them, the tangential distortion coefficients are p1 and p2, which are used to correct the linear deviation caused by plane tilt.

[0103] Finally, the corrected normalized coordinates are converted back to pixel coordinates (u c , v c ).

[0104]

[0105] Among them, K is the camera intrinsic parameter matrix ( f x , f y is the focal length (pixel unit), c x , c y coordinates of the principal point).

[0106] Finally, the actual size of the target in the perforation orthographic image after correction can be calculated based on the distance Z between the target and the camera.

[0107] Given that the inner diameter of the detection casing is D, the acquisition device for obtaining the perforation image moves along the central axis in the casing, and the center distance of the side-view camera is L (the distance between the camera and the central axis of the device), the distance Z between the target on the inner surface of the casing and the camera can be calculated.

[0108]

[0109] The actual size of the target (width W, height H) can be calculated using the principle of similar triangles. The relationship between pixel size and actual size is as follows:

[0110]

[0111]

[0112] Among them, w pix is the width of the target in the image, h pix is the height of the target in the image, fx, fy are the focal lengths (in pixels) in the camera intrinsics, and Z is the distance from the target to the camera.

[0113] Based on the relationship between the target size and the actual size in the image, the perforation contours on the inner and outer walls of the casing are identified in the distortion-corrected image of the perforation orthogonal view. The inner and outer perimeters of the perforation holes are obtained by calculating the total length of the boundaries of the inner and outer wall contours. The inner and outer areas are obtained by counting the number of pixels within the inner and outer wall contours and using the relationship between the pixel size and the actual size. The area values ​​can be used to calculate the average aperture value of equal-area circles using the circle area formula as the average aperture value of the perforation holes, thereby realizing the measurement of the perforation size.

Claims

1. A device for obtaining downhole in-situ perforation elevation images, characterized in that: It is possible to realize front-view shooting of perforations in all directions, that is, the center of the camera can be aligned with the center of the perforation for shooting, and comprises a front-view video acquisition unit (1), a side-view video acquisition unit (2) and a control unit (3); The forward-view video acquisition unit (1) is located at the front end of the device and includes an ultra-wide-angle camera (1-1) for acquiring a panoramic image of the front; The side-view video acquisition unit (2) is located in the middle of the device and collects front-view images of the perforations on the inner wall of the wellbore through a side-view camera (2-8) installed on its rotating frame (2-3); The control unit (3) is located at the rear of the device and includes a power supply system, a main control module, a motor drive module, an AI intelligent recognition module, and a video storage module; the power supply system provides working power, the motor drive module controls the motor in the side view video acquisition unit (2), and the AI ​​intelligent recognition module identifies the perforation target in the image collected by the front view and side view video acquisition units and outputs the perforation target position information; The main control module determines the angle at which the side view video acquisition unit (2) needs to rotate according to the perforation position identified by the AI ​​intelligent recognition module, and locks the target perforation position of the side view; The video storage module stores the collected front and side high-definition video images and supports real-time reading and storage working modes.

2. The device for obtaining downhole in-situ perforation elevation images according to claim 1, characterized in that: The ultra-wide-angle camera (1-1) has a viewing angle greater than 180°, and light sources (1-2) are evenly distributed around the ultra-wide-angle camera (1-1) to provide wellbore lighting.

3. The device for obtaining downhole in-situ perforation elevation images according to claim 1, characterized in that: The specific structure of the side view video acquisition unit (2) is as follows: a fixed axis (2-1) is provided at the center of the device body, the fixed axis (2-1) is connected to the front and rear ends of the device body, a through hole for passing a line is provided at the center of the fixed axis (2-1), two groups of bearings (2-2, 2-14) are provided outside the fixed axis (2-1), a rotating skeleton (2-3) is provided outside the two groups of bearings (2-2, 2-14), a large gear (2-6) is provided at one end of the rotating skeleton (2-3), the large gear (2-6) is driven to rotate by the small gear (2-4), and the small gear (2-4) is driven to rotate by the large gear (2-6). The gear (2-4) is connected to the rotating shaft of the motor (2-5); a slip ring (2-7) is provided at the other end of the rotating skeleton (2-3); the slip ring (2-7) is used to pass through the internal lead of the rotating skeleton (2-3); a side-view camera (2-8) is installed on the outer surface of the rotating skeleton (2-3); lighting lamp beads (2-10) are installed on the outer shells (2-11) on both sides of the side-view camera (2-8); an annular transparent window (2-13) is installed outside the camera (2-8); and a circular transparent window (2-12) is installed outside the lamp bead (2-10).

4. The device for obtaining downhole in-situ perforation elevation images according to claim 3, characterized in that: The number of the side-view cameras (2-8) installed is 1 to 4.

5. The device for obtaining downhole in-situ perforation elevation images according to claim 3, characterized in that: A small bearing (2-15) is installed between the rotating shaft of the motor (2-5) and the housing (2-11).

6. The device for obtaining downhole in-situ perforation elevation images according to claim 1, characterized in that: The AI ​​intelligent recognition module integrates a perforation recognition model. The perforation recognition model is trained using a single-stage neural network with a large number of perforation data sets. It can identify perforation targets in images captured by the front-view and side-view video acquisition units, and calculate and output the perforation target phase information in the front-view image based on the forward-view perforation recognition target, that is, the perforation phase angle. The calculation method is as follows: Perforations are identified in the front view image. Perforation contour data is obtained through image processing, contour edge detection, and contour screening. The minimum circumscribed rectangle of the contour is calculated, and the center of the rectangle is used as the center B of the perforation shape. The image center point is A. The angle θ between the line connecting the image center A and the perforation center point B and the perpendicular line passing through the image center point A is calculated. θ is the perforation phase angle.

7. The device for obtaining downhole in-situ perforation elevation images according to claim 6, characterized in that: The main control module calculates the angle that the side view video acquisition unit needs to rotate according to the perforation target phase of the front view acquisition unit identified by the AI ​​intelligent recognition module. When the side view camera is located at the phase angle α, the angle β that needs to be rotated is β=a-θ, where θ is the perforation phase angle; and outputs a motor control signal through the motor drive module to control the side view camera 2-8 of the side view video acquisition unit 2 to rotate to the target perforation orientation and lock the target perforation position.

8. A method for obtaining downhole in-situ perforation elevation images based on the device according to any one of the preceding claims, characterized in that: The following steps are involved: Step 1: Calibrate the side-view camera (2-8) in the side-view video acquisition unit (2). Use the side-view video acquisition unit (2) to shoot more than 10 images of a checkerboard calibration plate from different angles and distances. Use the calibration software to automatically identify the corner points of the calibration plate, obtain the image coordinates, establish the relationship between the world coordinates and the image coordinates, and solve the internal parameter matrix and distortion coefficient of the side-view camera (2-8); Step 2: Lower the device for obtaining downhole in-situ perforation front view images into the perforation section. Step 3: Acquisition of panoramic images of the wellbore After entering the perforation section, the device for obtaining the downhole in-situ perforation front view image is placed in the center of the wellbore and moves forward at a uniform speed along the central axis of the wellbore at a speed of less than 3 m / min. The forward video acquisition unit (1) obtains a 360° panoramic image of the wellbore through an ultra-wide-angle camera (1-1); Step 4: Image perforation target recognition The AI ​​intelligent recognition module identifies the target perforation orientation through its built-in perforation recognition model based on the wellbore image obtained by the forward-view video acquisition unit (1). Step 5: Side view camera locks target perforation position Since the relative positions of the side-view camera (2-8) and the ultra-wide-angle camera (1-1) are known, the side-view camera (2-8) is rotated according to the target perforation position provided by the AI ​​intelligent recognition module, and the corresponding side-view camera (2-8) is controlled to rotate to the target perforation position on the side, thereby locking the target perforation position; Step 6: The side-view camera collects the perforation front view image The center of the side-view camera (2-8) is aligned with the center of the target perforation position. When the side-view camera (2-8) moves to the target perforation position, it realizes front-view observation of the perforation and collects the front-view image of the perforation; Step 7: Correct the distortion of the perforation front view image and calculate the perforation size The orthographic image is corrected using the distortion coefficient, and the actual perforation size of the target in the corrected undistorted image is calculated in combination with the distance between the target and the camera.

9. The method for obtaining downhole in-situ perforation elevation images according to claim 8, characterized in that: When the device for obtaining the downhole in-situ perforation front-view image in step 2 is lowered to the perforation section, if it is lowered by a cable device, the device for obtaining the downhole in-situ perforation front-view image is connected to a high-speed cable to achieve real-time observation of downhole video images on the ground; If a cable-free device is used, the power supply system in the device for obtaining the downhole in-situ perforation elevation image is activated, and the downhole image is collected and stored. After the collection is completed, the device is lifted to the surface to download the video image.

10. The method for obtaining downhole in-situ perforation elevation images according to claim 8, characterized in that: The specific calculation method of the seventh step is as follows: (7.1) First, the pixel coordinates (μ, v) of the orthographic image are converted to normalized image coordinates using formula (1): Among them, K is the camera intrinsic parameter matrix f x , f y is the focal length, in pixels, c x , c y is the main point coordinate; (7.2) Use the distortion coefficient to correct the normalized coordinates and obtain the normalized coordinates without distortion (x c ,y c ), assuming that the ideal undistorted image coordinates are (x, y), and the actual coordinates after distortion are (x′, y′), then the correction formula for radial distortion is: x′=x·(1+k1r 2 +k2r 4 +k3r 6 +…) Formula (2) y′=y·(1+k1r 2 +k2r 4 +k3r 6 +…) Formula(3) Among them, r 2 =x 2 +y 2 , the square of the distance from the pixel to the image center; k is the radial distortion coefficient, where k1 is the first order, k2 is the second order), k3 is the third order), and k... are high-order coefficients; The correction formula for tangential distortion is: x′=x+[2p1xy+p2(r 2 +2x 2 )] Formula (4) y′=y+[p1(r 2 +2y 2 )+2p2xy] Formula (5) Among them, the tangential distortion coefficients are p1 and p2, which are used to correct the linear deviation caused by plane tilt; (7.3) The corrected normalized coordinates are converted back to pixel coordinates (u c , v c ), Among them, K is the camera intrinsic parameter matrix ( f x , f y is the focal length (pixel unit), c x , c y is the main point coordinate); (7.4) Given that the inner diameter of the detection casing is D, the acquisition device for obtaining the perforation image moves along the central axis of the casing. Given that the center distance of the side-view camera is L, calculate the distance Z between the target on the inner surface of the casing and the camera: The actual size of the target, width W and height H, are calculated using the principle of similar triangles. The relationship between pixel size and actual size is as follows: Among them, w pix is the width of the target in the image, h pix is the height of the target in the image, fx, fy are the focal lengths in the camera intrinsic parameters, in pixels; Z is the distance from the target to the camera, which is the relationship between the target size and the actual size; (7.5) Based on the relationship between the target size and the actual size in the image, the inner circumference and outer circumference of the perforation hole are obtained by calculating the total length of the inner wall contour and the outer wall contour boundary in the image after distortion correction of the perforation front view image. The inner area and outer area are obtained by counting the number of pixels within the inner wall contour and the outer wall contour and the relationship between the pixel size and the actual size. The area value can be used to calculate the average aperture value of the equal area circle using the circle area formula as the average aperture value of the perforation hole, thereby realizing the measurement of the perforation size.

Images