WO2020019348A1

WO2020019348A1 - Device and method for detecting inner wall of microcapillary on basis of coherent light

Info

Publication number: WO2020019348A1
Application number: PCT/CN2018/097662
Authority: WO
Inventors: 王星泽; 舒远
Original assignee: 合刃科技（深圳）有限公司
Priority date: 2018-07-27
Filing date: 2018-07-27
Publication date: 2020-01-30
Also published as: CN111213053A

Abstract

A device (100) and method for detecting the inner wall of a microcapillary (101) on the basis of coherent light, the device (100) comprising: a coherent light emitter (108), used for generating coherent light and transmitting the same to a detection probe (102) by way of a first lens (105), a half mirror (104) and an optical fiber (103); the detection probe (102), used for projecting the coherent light onto the inner wall of the microcapillary (101) and transmitting the reflected coherent light to a coherent light receiver (107) by way of the optical fiber (103), the half mirror (104) and a second lens (106); the coherent light receiver (107), used for obtaining a speckle image of the inner wall of the microcapillary (101) according to the reflected coherent light and transmitting the speckle image of the inner wall of the microcapillary (101) to a image processor (109); and the image processor (109), used for determining a defect type of the inner wall of the microcapillary (101) according to the speckle image of the inner wall of the microcapillary (101). Employing the described device solves the problem in existing technology of it being impossible to detect a defect of the inner wall of a microcapillary.

Description

Device and method for detecting inner wall of microtube based on coherent light

Technical field

The present application relates to the field of pipeline detection, and in particular, to a device and method for detecting the inner wall of a microtube based on coherent light.

Background technique

Micro-tubes and small-sized characteristic inner holes have been widely used in the fields of machinery manufacturing, chemical engineering, and medical instruments. Once these critical inner hole structures are exploded, leaked, etc., the whole machine will not work, and it will even pose a serious threat to human life and property. Defect detection on the inner wall of microtubes is of great significance for manufacturing, quality control and safety assurance.

For conventional pipeline inspection, non-optical non-destructive inspection methods are mostly used in the industry, mainly by using the electromagnetic characteristics and ultrasonic of pipeline materials, including magnetic flux leakage, eddy current, and ultrasonic methods. The other is the traditional method of contact measurement, which is mainly a measurement method to obtain measurement information through the direct contact between the sensing element of the measurement device and the measured surface. The disadvantage is that frequent contact between the probe and the tube wall will cause the probe to wear. In order to maintain accuracy, it needs to be calibrated frequently. Point-by-point measurement, the measurement speed is slow, the detection efficiency is low, and it is impossible to detect micro pipes or parts with a diameter smaller than the probe diameter. Inner hole, etc.

The detection requirements for small-sized micro-pipes and inner holes of parts are increasing, the internal space of micro-pipes is small, and the system requires high detection accuracy. The above conventional detection methods cannot meet the detection requirements of such pipes, and effective detection methods are basically The above are based on optical visual inspection, wired laser 3D scanning method, and through the use of a charge-coupled device (CCD) camera, the surrounding small lights are directly taken into the pipe to take pictures of the inner wall, and get The image was analyzed to identify defect areas and defect sizes. These two methods are suitable for larger-diameter pipes and can achieve high-precision measurement, but the imaging device required is relatively large, making the relatively small structure of the pipe inoperable. In addition, the image of the inner wall of the tube will be affected by uneven illumination, inaccurate imaging focus, and image distortion, which will reduce the detection effect. In summary, based on the traditional lighting methods and imaging technology, the inner wall of the microtube cannot be detected.

Summary of the Invention

The embodiments of the present application provide a device and a method for detecting the inner wall of a microtube based on coherent light, which solves the problem that the inner wall of the microtube cannot be accurately detected in the prior art.

In a first aspect, an embodiment of the present application provides a microtube inner wall detection device, including:

A detection probe, a transflective mirror connected to the detection probe through an optical fiber, a first lens, a second lens, a coherent light transmitter, a coherent light receiver, and an image processor; the front end of the detection probe is provided with a cone Reflector;

The coherent light emitter is configured to generate incident coherent light, and transmit the coherent light to the detection probe through the first lens, the half mirror, and an optical fiber;

The conical mirror of the detection probe is configured to project the incident coherent light onto the inner wall of the microtube and reflect the coherent light on the inner wall of the microtube to pass through the optical fiber, the semi-transparent mirror, and The second lens is transmitted to the coherent light receiver;

The coherent light receiver is configured to obtain a speckle image of the inner wall of the micro tube according to the reflected coherent light; and transmit the speckle image of the inner wall of the micro tube to the image processor;

The image processor is configured to determine a defect type of the inner wall of the microtube based on a speckle image of the inner wall of the microtube.

In a feasible embodiment, the speckle image of the inner wall of the microtube includes multiple panoramic subspeckle images of the inner wall of the microtube, and the image processor determines the speckle image based on the speckle image of the inner wall of the microtube. Types of defects on the inner walls of microtubules, including:

The image processor stitches the panoramic sub-speckle images of the inner wall of the plurality of micro-tubes into the panoramic speckle images of the inner wall of the micro-tube;

Calculating, by the image processor, a panoramic speckle image on the inner wall of the microtube according to a defect recognition model to obtain a calculation result;

The image processor obtains a defect type corresponding to the calculation result from a correspondence table between the calculation result and the defect type to determine the defect type of the inner wall of the microtube.

In a feasible embodiment, the image processor is further configured to:

Before determining the defect type of the inner wall of the microtube according to the calculation result, multiple sets of speckle images are obtained, and each speckle image in each set of speckle images of the multiple set of speckle images corresponds to a defect type ;

Performing neural network training according to the multiple sets of speckle images to obtain the defect recognition model;

Inputting the multiple sets of speckle images to the defect recognition model for calculation respectively to obtain multiple sets of calculation results, and each of the multiple sets of calculation results corresponds to a defect type;

According to the plurality of sets of calculation results, a correspondence table between the calculation results and the defect type is obtained, and the correspondence table between the calculation results and the defect type includes a calculation result range and a corresponding defect type, and an upper limit of the calculation result range and The lower limit is the maximum and minimum of a set of calculation results corresponding to the defect type, respectively.

In a feasible embodiment, the image processor is further configured to:

Before determining the defect type of the inner wall of the microtube according to the calculation result, a request message is sent to a third-party server, where the request message is used to request to obtain the defect identification model and calculation result and defect stored by the third-party server. Type correspondence table;

Receiving a response message sent by the third-party server in response to the request message, the response message carrying the defect identification model and a correspondence table between calculation results and defect types.

In a feasible embodiment, the image processor is further configured to:

After each defect type identification is performed N times using the defect identification model and the correspondence table between the calculation result and the defect type, multiple sets of speckle images are obtained again, where N is an integer greater than 1;

Retraining the defect recognition model according to the reacquired multiple sets of speckle images to obtain the retrained defect recognition model;

Inputting the reacquired multiple sets of speckle images to the trained defect recognition model for calculation respectively to obtain multiple sets of calculation results, and each of the multiple sets of calculation results corresponds to a defect type;

According to the plurality of sets of calculation results, a correspondence table between the calculation results and defect types is obtained again.

In a feasible embodiment, the device further includes: a movement mechanism connected to the detection probe;

The movement mechanism is configured to draw the detection to move in the microtube, and the distance of each movement is less than or equal to the length of the microtube detected by the detection probe each time.

In a feasible embodiment, when the detection probe has a curved shape, the detection probe further includes:

A rotation device is used to rotate the detection probe, and the rotation angle is an angle input by a user.

In a feasible embodiment, the coherent light is a laser at any frequency between ultraviolet light and near-infrared light.

In a second aspect, an embodiment of the present application further provides a method for detecting an inner wall of a microtube based on coherent light, including:

Obtaining a panoramic speckle image of the inner wall of the microtube according to the coherent light reflected from the inner wall of the microtube;

In a feasible embodiment, acquiring a panoramic speckle image of the inner wall of the micro tube according to coherent light reflected from the inner wall of the micro tube includes:

Obtaining a plurality of panoramic sub-speckle images of the inner wall of the micro tube according to the coherent light reflected from the inner wall of the micro tube;

The panoramic sub-speckle images on the inner wall of the plurality of microtubes are stitched into the panoramic speckle images on the inner wall of the micro-tube.

In a feasible embodiment, the method further includes:

Before determining the defect type of the inner wall of the microtube according to the calculation result, acquiring multiple sets of speckle images, each speckle image in each set of speckle images corresponding to a defect type;

In a feasible embodiment, the method further includes:

According to a third aspect, an embodiment of the present application further provides a computer storage medium. The computer storage medium stores a computer program, where the computer program includes program instructions, and the program instructions, when executed by the processor, cause the processor Perform all or part of the method as in the second aspect.

It can be seen that, in the solution of the embodiment of the present application, the conical mirror provided at the front end of the detection probe irradiates incident coherent light onto the inner wall of the microtube, and transmits the coherent light reflected on the inner wall of the microtube to the coherent light receiver; coherent light The receiver obtains a speckle image of the inner wall of the microtube according to the reflected coherent light; transmits the speckle image of the inner wall of the microtube to a graphics processor; the image processor determines the type of defect of the inner wall of the microtube based on the speckle image of the inner wall of the microtube. The embodiment of the present application is adopted to solve the problem that the inner wall defect of the microtube cannot be detected in the prior art.

These or other aspects of the present application will be more concise and easy to understand in the description of the following embodiments.

BRIEF DESCRIPTION OF THE DRAWINGS

In order to explain the technical solutions in the embodiments of the present application or the prior art more clearly, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings in the following description are merely These are some embodiments of the present application. For those of ordinary skill in the art, other drawings can be obtained according to these drawings without paying creative labor.

FIG. 1 is a schematic structural diagram of a microtube inner wall detection device according to an embodiment of the present application; FIG.

2 is a schematic structural diagram of another microtube inner wall detection device according to an embodiment of the present application;

3 is a schematic diagram of the principle of coherent light imaging;

FIG. 4 is a schematic diagram of detecting the inner wall of a microtube based on coherent light according to an embodiment of the present application; FIG.

FIG. 5 is another schematic diagram of detecting the inner wall of a microtube based on coherent light according to an embodiment of the present application; FIG.

FIG. 6 is another schematic diagram of detecting the inner wall of a microtube based on coherent light according to an embodiment of the present application; FIG.

FIG. 7 is another schematic diagram of detecting the inner wall of a microtube based on coherent light according to an embodiment of the present application; FIG.

FIG. 8 is another schematic diagram of detecting the inner wall of a microtube based on coherent light according to an embodiment of the present application; FIG.

FIG. 9 is a schematic diagram of a defect recognition model according to an embodiment of the present application; FIG.

FIG. 10 is another schematic diagram of detecting the inner wall of a microtube based on coherent light according to an embodiment of the present application; FIG.

FIG. 11 is a schematic diagram of detecting the inner wall of a microtube based on coherent light according to an embodiment of the present application; FIG.

FIG. 12 is a schematic flowchart of a method for detecting an inner wall of a microtube based on coherent light according to an embodiment of the present application.

detailed description

The embodiments of the present application are described below with reference to the drawings.

Referring to FIG. 1, FIG. 1 is a schematic structural diagram of a microtube inner wall detection and repair device according to an embodiment of the present application. As shown in FIG. 1, the micro-tube inner wall detection device 100 includes:

The detection probe 102 is a half mirror 104, a first lens 105, a second lens 106, a coherent light transmitter 108, a coherent light receiver 107, and an image processor 109 connected to the detection probe 102 through an optical fiber 103.

The coherent light emitter 108 generates coherent light and irradiates the first lens 105. The first lens 105 irradiates the coherent light on a semi-transparent mirror 104 located on a focal plane of the first lens 105, and passes through the semi-transparent mirror 105 The optical fiber 103 connected to the half mirror 104 transmits the coherent light to the detection probe 102. For the detection probe 102, the coherent light transmitted from the semi-transparent mirror 104 through the optical fiber 103 may also be referred to as incident coherent light.

As shown in FIG. 2, when using the thin tube inner wall detection device 100 for detection, the detection probe 102 needs to be inserted into the micro tube 101 to be detected. A conical mirror 1021 is provided at the front end of the detection probe 102. 1021 reflects the incident coherent light 10 onto the inner wall of the microtube 101, and reflects the coherent light 11 reflected or scattered by the inner wall of the microtube 101.

The coherent light 11 reflected or scattered on the inner wall of the microtube 101 is transmitted to the coherent light receiver 107 through the half mirror 104 and the second lens 106, and the coherent light receiver 107 converts the received coherent light into speckle image. The coherent light receiver 107 is located on a focusing plane of the second lens 106. This design allows the illumination light path and the imaging light path to be combined into one, realizing the simultaneous completion of lighting and imaging in a narrow micro-tube space.

The incident coherent light is a laser light having any frequency between ultraviolet light and near-infrared light.

Optionally, the above-mentioned coherent light emitter may be a charge-coupled device (CCD) sensor, a complementary metal oxide semiconductor (CMOS) sensor.

It should be noted here that the inner wall of the microtube has a fixed structure or a smooth surface, and the defects are mainly distributed on the inner wall of the component. The size of these defects is generally between 50um and 1mm. For a coherent laser with a wavelength of 650nm, it can The accuracy of the micro-tube inner wall defect detection can reach 1um. Microscopic deformation of the inner wall of the microtube will cause changes in the diffraction spot. When light hits the rough surface, there will be scattered light at each point on the surface. These scattered lights are coherent light, with different amplitudes and phases, and randomly distributed. These scattered lights are superimposed to form a granular structure with obvious contrast. These are speckles. As shown in FIG. 3, after the incident coherent light is irradiated to the inner wall of the microtube, the inner wall of the microtube reflects or scatters the coherent light to the coherent light receiver. The coherent light receiver converts the received light signal into an image, and obtains fine Speckle image of the inner wall of the tube.

Since the detection probe 102 detects the current position of the annulus zone at a fixed measurement position in the microtube 101 each time, it is necessary to draw the detection probe to move along the microtube 101 through a movement mechanism, and the coherent light receiver 107 can obtain The panoramic sub-images of the inner wall of the microtube at different positions are shown in FIG. 4, and the inner-wall sub-images of the current position are circle-shaped shadow parts.

Assuming that the length of the microtube that can be detected by the detection probe 102 is d and the total length of the microtube is L, the detection probe needs to perform L / d detections to complete the detection of the inner wall of the microtube. After each detection by the detection probe 102, a panoramic sub-speckle image of the inner wall is obtained through the coherent light receiver; then, under the traction of the moving structure, the detection probe 102 moves a distance d, and the above detection process is repeated; the above length is After the detection of L's microtubules, L / d inner wall panoramic sub-speckle images can be obtained, as shown in FIG. 5.

After the coherent optical receiver 107 obtains an inner wall panoramic sub-speckle image, it transmits the inner wall panoramic sub-speckle image to the image processor 109. The image processor 109 receives an inner wall panoramic sub-speckle image. Then, the inner wall panoramic sub-speckle image is time stamped, and the time stamp is the current system time. After receiving the L / d inner wall panoramic sub-speckle images, the image processor 109 stitches the L / d inner wall panoramic sub-speckle images into the micro-tube inner wall panoramic speckle images according to the sequence of time tags.

The image stitching technology is briefly described here. The image stitching technology includes the following steps:

1) Image preprocessing: Basic operations of digital image processing such as histogram matching, smoothing filtering, and enhanced transformation on the original image are prepared for the next step of image stitching.

2) Image registration: Image registration is the core of the entire image stitching process, and the accuracy of registration determines the quality of image stitching. The basic idea is: first find the corresponding position of the template or feature point of the image to be registered and the reference image, and then establish a mathematical model of the conversion between the reference image and the image to be registered according to the corresponding relationship, and convert the image to be registered to the reference In the image coordinate system, the overlapping area between the two images is determined. The key to accurate registration is to find a data model that can well describe the transformation relationship between the two images.

3) Image synthesis: After determining the conversion relationship model between the two images, that is, after overlapping regions, the images to be stitched need to be mosaic into a visually feasible panoramic image according to the information of the overlapping regions. Due to different shooting conditions and other factors, the grayscale (or brightness) of the image is different, or there is still a certain registration error in the image registration result. In order to minimize the residual distortion or the brightness (or grayscale) difference between the images, the mosaic result is affected. Impact, you need to choose a suitable image synthesis strategy.

In a feasible embodiment, in order to improve the detection accuracy, during the detection of the inner wall of the microtube 101, the detection probe 102 moves a distance d1 each time under the traction of the moving mechanism, and the d1 is less than d, and passes M In the second inspection, the detection probe 102 completes the inspection of the inner wall of the microtube 101 with the length L, and obtains M sub-speckle panoramic images of the inner wall of the microtube, where M = L / d1; since the detection probe 102 can The length of the detected microtube is d, so there are overlapping detection areas in the two adjacent detection processes, so there are the same parts for the two inner wall panoramic sub-speckle images obtained by the two adjacent detections, such as Figure 6 shows.

After the image processor 109 obtains the panoramic sub-images of the inner walls of the M micro-tubes, the panoramic sub-images of the inner walls of the M micro-tubes are stitched according to an image stitching technique to obtain a panoramic speckle image of the inner walls of the micro-tubes.

In a feasible embodiment, as shown in FIG. 7 (a), in order to adapt to micro tubes of different calibers, the above-mentioned detection probe 102 is provided in a curved shape, and a rotation device 1022 is provided on the detection probe 102. The rotation device The angle input by the user can be received in a wired or wireless manner, and defect detection at different positions and different angles on the inner wall of the microtube 101 can be achieved, and further, directional detection of defects on the inner wall of the microtube 101 can be realized.

Specifically, the length of each detection of the inner wall of the microtube 101 by the detection probe 102 is s, as shown in FIG. 7A; the angle of each detection is θ, as shown in FIG. 7B. The 102 coherent light receiver 107 obtains an inner wall panoramic speckle image block after each detection, and transmits the inner wall panoramic speckle image block to the image processor 109, which obtains the inner wall panoramic speckle image. After the block, a first time stamp is added to the inner speckle panoramic image block, and the first time stamp is the current system time; the detection probe 102 is rotated by an angle θ1 according to a preset direction under the control of its rotation device 1022. θ1 = θ; rotate 360 / θ once, the detection probe 102 completes the detection of the inner wall of the microtube at the current position. At this time, the image processor 109 obtains 360 / θ1 panoramic speckle image blocks of the inner wall. The sequence of the first time label of the inner wall panoramic speckle image block, according to the image stitching technique, the above 360 / θ1 inner wall panoramic speckle image block is stitched into an inner wall panoramic sub speckle image, and The inner wall panoramic sub-speckle image is marked with a second time tag, and the second time tag is the current system time.

Optionally, the preset direction may be a clockwise direction or a counterclockwise direction.

After completing the defect detection of the inner wall of the current position of the annulus, the above-mentioned detection probe 102 is moved along the detection direction by a distance s1 and s1 = s under the traction of the moving mechanism; and then the above-mentioned detection probe completes the acquisition of the current position according to the above method. An inner wall panoramic sub-speckle image at the current position, and a second time label is added to the inner wall panoramic sub-speckle image. According to this method, under the traction of the moving structure, the detection probe undergoes the above-mentioned detection process L / s1 to complete the detection of the inner wall of the microtube, and the image processor obtains L / s1 panoramic sub-speckle image of the inner wall.

After the image processor 109 obtains the L / s1 inner wall panoramic sub-speckle images, according to the sequence of the second time label of each inner wall panoramic sub-speckle image, the above L / s1 inner wall panoramic sub-pixels are processed according to the image stitching technology. The speckle image is stitched into a panoramic speckle image on the inner wall.

Optionally, under the control of the rotation device 1022 of the detection probe, when the detection probe 102 rotates according to the preset direction by an angle θ1 <θ, since the detection angle of the detection probe 102 is θ each time, the detection probe 102 The area where the inner wall of the microtube is detected twice in succession will overlap, as shown in FIG. 8. At this time, the two inner wall panoramic speckle image blocks obtained by the above-mentioned coherent light receiver 107 connector have overlapping portions. According to this method, after the above-mentioned detection probe 102 is rotated 360 / θ 1 times under the controller of its rotating device 1022, the defect detection of the inner wall of the current position annulus is completed, and the image processor 109 acquires 360 / θ 1 inner wall panoramic speckles. Image block; the image processor 109 stitches the 360 / θ1 inner wall panoramic speckle image block into an inner wall panoramic sub-speckle image at the current position according to the image stitching technology. Then according to the above method, under the traction of the moving structure, the detection probe 102 undergoes the above-mentioned detection process L / s1 to complete the detection of the inner wall of the microtube, and the image processor 109 obtains L / s1 inner wall panoramic subspeckle image.

In a possible embodiment, after the image processor 109 obtains the panoramic speckle image of the inner wall of the microtube, the image processor 109 inputs the panoramic speckle image of the inner wall of the microtube into a defect recognition model, and the defect The recognition model is a neural network model. The defect recognition model is used to perform a neural network operation on the panoramic speckle image on the inner wall of the microtube to obtain at least one calculation result. Each calculation result corresponds to a defect type. The type of defects existing on the inner wall of the microtube can be determined based on the calculation results. As shown in FIG. 9, the above defect recognition model includes an input layer, an intermediate layer, and an output layer; after the panoramic speckle image of the inner wall of the microtube is input from the input layer and after the intermediate layer operation, the output layer can output four calculation results, Including the first calculation result, the second calculation result, the third calculation result, and the fourth calculation result, respectively, corresponding to defect-free, dirty, cracked and deformed.

Further, the output layer of the defect recognition model may output any one of the four calculation results or output any combination of the second calculation result, the third calculation result, and the fourth calculation result, that is, the output of the defect recognition model. The defect type can be any one of no defect, dirt, cracking, and deformation; or the output defect type is any combination of dirt, cracking, and deformation. This method can directly determine the type of defects existing in the inner wall of the microtube based on the input panoramic speckle image of the inner wall of the microtube.

In a feasible embodiment, after the image processor 109 obtains the panoramic speckle image of the inner wall of the micro tube, the image processor 109 inputs the panoramic speckle image of the inner wall of the micro tube into a defect recognition model. The defect recognition model is used to calculate the panoramic speckle image on the inner wall of the microtube to obtain a calculation result, and then the defect type corresponding to the settlement result is determined according to the correspondence table between the calculation result and the defect type.

The correspondence table between the above calculation results and defect types is shown in Table 1.

计算结果范围Calculation result range	缺陷类型Defect type
(a1,a2](a1, a2)	无缺陷No defects
(a2,a3](a2, a3)	脏污Dirty
(a3,a4](a3, a4)	崩裂crack
(a4,a5)(a4, a5)	形变deformation

Table 1

Specifically, in the table of the correspondence between the calculation results and the defect types, four types of defects are listed, which are no defect, dirt, cracking, and deformation. When the calculation result is greater than a1 and less than or equal to a2, the image processor 109 determines that the inner wall of the microtube is free of defects; when the settlement result is greater than a2 and less than or equal to a3, the image processor 109 determines the inner wall of the microtube. The defect type is dirty; when the settlement result is greater than a3 and less than or equal to a4, the image processor 109 determines that the defect type of the inner wall of the microtube is cracked; when the settlement result is greater than a4 and less than a5, the image processor 109 It is determined that the defect type on the inner wall of the microtube is deformation.

It should be noted that the types of defects on the inner wall of the microtubes include, but are not limited to, four types: non-defective, dirty, cracked and deformed.

Optionally, before the image processor 109 inputs the panoramic speckle image of the inner wall of the microtube into the defect recognition model, before the image processor 109 determines the defect type of the inner wall of the microtube according to the calculation result, Acquiring multiple sets of speckle images, each speckle image in each set of speckle images corresponding to a defect type;

As shown in Table 1, the image processor 109 inputs a group of non-defective speckle images into the above-mentioned defect recognition model, and obtains multiple calculation results. The a1 of the calculation result range corresponding to the non-defect is the number of the multiple calculation results. The minimum value, a2 is the maximum value among the multiple calculation results.

In a feasible embodiment, before the image processor 109 inputs the panoramic speckle image of the inner wall of the microtube into a defect recognition model, and before determining a defect type of the inner wall of the microtube according to the calculation result, Sending a request message to a third-party server, where the request message is used to request to obtain the defect identification model and the correspondence table between the calculation result and the defect type stored by the third-party server;

Further, the image processor 109 is further configured to retrain the defect recognition model and update the correspondence table between the calculation result and the defect type to ensure the accuracy of defect recognition, as follows:

It should be noted that the retraining of the defect recognition model and the update of the correspondence table between the calculation results and the defect types can be performed by the third-party server. Each time the image processor 109 uses the defect recognition model and the calculation results and the defect types, After identifying the defect type N times in the corresponding relationship table, the request message is re-sent to the third-party server for requesting to obtain the correspondence table between the retrained defect identification model and the updated calculation result and defect type.

In a feasible embodiment, the micro-tube inner wall detecting device can perform a defect detection at a fixed position on the micro-tube inner wall.

Specifically, as shown in FIG. 10, the detection probe 102 is provided with a conical reflector; the micro-tube inner wall detection device receives position information to be detected, and the position information to be detected indicates a position between the position to be detected and the micro-tube detection entrance. Distance, the entrance of the capillary tube is the position corresponding to the origin O as shown in FIG. 10. The detection probe 102 moves to the position to be detected indicated by the position information to be detected under the traction of the moving mechanism, and performs defect detection on the annular inner wall of the position to be detected. For the specific process of performing the defect detection by the detection probe 102 on the position to be detected, refer to the related description of the foregoing embodiment, and details are not described herein again.

The coherent light receiver 109 obtains a speckle image of the annular inner wall of the position to be detected according to the reflected or scattered coherent light transmitted by the optical fiber 103, and transmits the speckle image to the image processor 109. The image processor 109 The speckle image is input to the defect recognition model for calculation to obtain a calculation result. The image processor 109 determines a defect type corresponding to the calculation result according to a correspondence relationship table between the calculation result and the defect type, so as to determine whether a defect and a defect type exist in the position to be detected.

As shown in FIG. 11, a rotation device is provided in the detection probe 102, and the detection device for the inner wall of the microtube receives the information of the point to be detected (L1, α), where L1 represents the point between the point to be detected and the detection entrance of the microtube 101. The distance between the two, α represents the included angle between a straight line passing through the position to be detected and the center point of the cross section where the detected position is located, and a straight line L2 passing through the center point. The detection probe 102 is moved to the cross section of the point to be detected by the microtube 101 at the entrance L1 under the traction of the movement mechanism. Then, under the control of the rotating device, the detection probe 102 uses the straight line L2 as a reference line, The detection probe 102 is rotated clockwise, and the rotation angle is α, so that the point to be detected is located within the detection range of the detection probe 102. The detection probe 102 then detects the point to be detected using coherent light. For the specific process of performing the defect detection by the detection probe 102 on the inspection point, refer to the related description of the foregoing embodiment, which will not be described here.

The coherent light receiver 107 obtains a speckle image of the annular inner wall of the point to be detected according to the reflected or scattered coherent light transmitted by the optical fiber 103, and transmits the speckle image to the image processor, and the image processor 109 transmits the speckle image The speckle image is input to the above defect recognition model and calculated to obtain the calculation result. The image processor 109 determines a defect type corresponding to the calculation result according to a correspondence table between the calculation result and the defect type, so as to determine whether a defect and a defect type exist at the point to be detected.

It can be seen that, in the scheme of the embodiment of the present application, a speckle image of the inner wall of the microtube is obtained by using the speckle image to determine whether there is a defect in the inner wall of the microtube; when it is determined that the defect exists, the type of the defect is determined; The type of defect fixes it. The embodiments of the present application not only solve the problem that the defects of the inner wall of the microtube cannot be accurately detected in the prior art, but also repair the defects of the inner wall of the microtube.

Referring to FIG. 12, FIG. 12 is a schematic flowchart of a method for detecting an inner wall of a microtube according to an embodiment of the present application. As shown in Figure 12, the method includes:

S1201. The micro-tube inner wall detecting device acquires a panoramic speckle image of the inner wall of the micro-tube according to coherent light reflected from the inner wall of the micro-tube.

S1202. The micro-tube inner wall detecting device calculates a panoramic speckle image of the inner wall of the micro-tube according to a defect recognition model to obtain a calculation result.

S1203. The micro-tube inner wall detecting device obtains a defect type corresponding to the calculation result from a correspondence table between the calculation result and the defect type, so as to determine the defect type of the micro-tube inner wall.

In a feasible embodiment, the method further includes:

It should be noted that, for specific implementation manners of the foregoing steps S1201 to S1203, reference may be made to related descriptions of the embodiments shown in FIGS. 1 to 11, and details are not described herein again.

An embodiment of the present application further provides a computer storage medium. The computer storage medium stores a computer program, where the computer program includes program instructions, and when the program instructions are executed by a processor, the computer executes the programs as shown in FIG. 12. Shows all or part of the method of the embodiment.

The embodiments of the present application have been described in detail above. Specific examples have been used in this document to explain the principles and implementation of the present application. The descriptions of the above embodiments are only used to help understand the methods and core ideas of the present application. Persons of ordinary skill in the art may change the specific implementation and application scope according to the idea of the present application. In summary, the content of this description should not be construed as a limitation on the present application.

Claims

A microtube inner wall detection device based on coherent light is characterized in that it includes:

A detection probe, a transflective mirror connected to the detection probe through an optical fiber, a first lens, a second lens, a coherent light transmitter, a coherent light receiver, and an image processor; the front end of the detection probe is provided with a cone Reflector;

The coherent light emitter is configured to generate incident coherent light, and transmit the coherent light to the detection probe through the first lens, the half mirror, and an optical fiber;

The conical mirror of the detection probe is used for projecting the incident coherent light onto the inner wall of the microtube, and reflecting the coherent light reflected on the inner wall of the microtube, passing through the optical fiber and the semi-transparent mirror And the second lens is transmitted to the coherent light receiver;

The coherent light receiver is configured to obtain a speckle image of the inner wall of the micro tube according to the reflected coherent light; and transmit the speckle image of the inner wall of the micro tube to the image processor;

The image processor is configured to determine a defect type of the inner wall of the microtube based on a speckle image of the inner wall of the microtube.
The device according to claim 1, wherein the speckle image of the inner wall of the microtube includes multiple panoramic subspeckle images of the inner wall of the microtube, and the image processor is based on the speckle image of the inner wall of the microtube. To determine the type of defect on the inner wall of the microtube, including:

The image processor stitches the panoramic sub-speckle images of the inner wall of the plurality of micro-tubes into the panoramic speckle images of the inner wall of the micro-tube;

Calculating, by the image processor, a panoramic speckle image on the inner wall of the microtube according to a defect recognition model to obtain a calculation result;

The image processor obtains a defect type corresponding to the calculation result from a correspondence table between the calculation result and the defect type to determine the defect type of the inner wall of the microtube.
The apparatus according to claim 2, wherein the image processor is further configured to:

Before determining the defect type of the inner wall of the microtube according to the calculation result, multiple sets of speckle images are obtained, and each speckle image in each set of speckle images of the multiple set of speckle images corresponds to a defect type ;

Performing neural network training according to the multiple sets of speckle images to obtain the defect recognition model;

Inputting the multiple sets of speckle images to the defect recognition model for calculation respectively to obtain multiple sets of calculation results, and each of the multiple sets of calculation results corresponds to a defect type;

According to the plurality of sets of calculation results, a correspondence table between the calculation results and the defect type is obtained, and the correspondence table between the calculation results and the defect type includes a calculation result range and a corresponding defect type, and an upper limit of the calculation result range and The lower limit is the maximum and minimum of a set of calculation results corresponding to the defect type, respectively.
The apparatus according to claim 2, wherein the image processor is further configured to:

Before determining the defect type of the inner wall of the microtube according to the calculation result, a request message is sent to a third-party server, where the request message is used to request to obtain the defect identification model and calculation result and defect stored by the third-party server. Type correspondence table;

Receiving a response message sent by the third-party server in response to the request message, the response message carrying the defect identification model and a correspondence table between calculation results and defect types.
The apparatus according to claim 3 or 4, wherein the image processor is further configured to:

After each defect type identification is performed N times using the defect identification model and the correspondence table between the calculation result and the defect type, multiple sets of speckle images are obtained again, where N is an integer greater than 1;

Retraining the defect recognition model according to the reacquired multiple sets of speckle images to obtain the retrained defect recognition model;

Inputting the reacquired multiple sets of speckle images to the trained defect recognition model for calculation respectively to obtain multiple sets of calculation results, and each of the multiple sets of calculation results corresponds to a defect type;

According to the plurality of sets of calculation results, a correspondence table between the calculation results and defect types is obtained again.
The device according to any one of claims 1-5, further comprising: a movement mechanism connected to the detection probe;

The movement mechanism is used for pulling the detection to move in the microtube, and the distance of each movement is less than or equal to the length of the microtube detected by the detection probe each time;

The detection probe further includes a rotation device for rotating the detection probe, and the rotation angle is an angle input by a user.
A method for detecting the inner wall of a microtube based on coherent light, comprising:

Obtaining a plurality of panoramic sub-speckle images of the inner wall of the micro tube according to the coherent light reflected from the inner wall of the micro-tube; and stitching the panoramic sub-speckle images of the inner wall of the plurality of micro-tubes into a panoramic speckle of the inner wall of the micro-tube image;

Calculating the panoramic speckle image on the inner wall of the microtube according to the defect recognition model to obtain a calculation result;

A defect type corresponding to the calculation result is obtained from a correspondence table between the calculation result and the defect type to determine the defect type of the inner wall of the microtube.
The method according to claim 7, further comprising:

Before determining the defect type of the inner wall of the microtube according to the calculation result, multiple sets of speckle images are obtained, and each speckle image in each of the multiple sets of speckle images corresponds to a defect Types of;

Performing neural network training according to the multiple sets of speckle images to obtain the defect recognition model;

Inputting the multiple sets of speckle images to the defect recognition model for calculation respectively to obtain multiple sets of calculation results, and each of the multiple sets of calculation results corresponds to a defect type;

According to the plurality of sets of calculation results, a correspondence table between the calculation results and the defect type is obtained, and the correspondence table between the calculation results and the defect type includes a calculation result range and a corresponding defect type, and an upper limit of the calculation result range and The lower limit is the maximum and minimum of a set of calculation results corresponding to the defect type, respectively.
The method according to claim 7, further comprising:

Before determining the defect type of the inner wall of the microtube according to the calculation result, a request message is sent to a third-party server, where the request message is used to request to obtain the defect identification model and calculation result and defect stored by the third-party server. Type correspondence table;

Receiving a response message sent by the third-party server in response to the request message, the response message carrying the defect identification model and a correspondence table between calculation results and defect types.
The method according to any one of claims 7-9, wherein the method further comprises:

After each defect type identification is performed N times using the defect identification model and the correspondence table between the calculation result and the defect type, multiple sets of speckle images are obtained again, where N is an integer greater than 1;

Retraining the defect recognition model according to the reacquired multiple sets of speckle images to obtain the retrained defect recognition model;

Inputting the reacquired multiple sets of speckle images to the trained defect recognition model for calculation respectively to obtain multiple sets of calculation results, and each of the multiple sets of calculation results corresponds to a defect type;

According to the plurality of sets of calculation results, a correspondence table between the calculation results and defect types is obtained again.