A blind hole depth detection method and device based on coaxial lighting dual-camera single lens

Abstract

The present invention discloses a blind hole depth detection method and device based on a coaxially illuminated dual-camera single-lens, belonging to the field of blind hole depth measurement technology. The steps of the present invention are as follows: S100: Debugging the camera clear focus imaging plane spacing so that the camera focuses on the bottom and mouth of the blind hole to be detected and collects the image to be processed; S200: Performing image preprocessing; S300: Performing focus clarity calculation and detection on the image to obtain a clarity evaluation function value for each image; S400: Repeating S200 and S300 for the obtained clear imaging of the bottom and mouth planes, calculating the high-frequency component grayscale mean function value of the clear images of the bottom and mouth planes; S500: Determining the relationship between the high-frequency component grayscale mean function and depth information; S600: Calculating the depth information of the blind hole to be detected based on the obtained relationship. The present invention effectively reduces the time for calculating the depth and ensures the accuracy of the depth measurement.

Description

Technical Field

[0001] The present invention relates to the technical field of blind hole depth detection, and more particularly to a blind hole depth detection method and device based on coaxial lighting dual-camera single-lens. Background Art

[0002] Threaded workpieces are widely used in the connection and fastening of mechanical structures. During the threading process in the machining industry, the threaded holes on some workpieces are sometimes insufficiently deep or excessively large after machining due to various factors, such as improper processing (welding a workpiece with a pre-threaded hole and then machining the end face of the threaded hole), operational errors (calculation errors, shallow bottom hole drilling, programming errors, and incorrect positioning of the drilling depth). Especially in today's precision-manufactured workpieces, the depth of blind holes directly affects the workpiece's usability. Insufficient depth can affect the fit of the workpiece assembly, preventing effective fastening. A blind hole depth exceeding the reserved error range can also affect the tightness of the workpiece.

[0003] Existing depth detection methods usually use detection and measurement tools to penetrate deep into the blind hole for direct contact measurement and detection. For example, a depth micrometer is used to detect depth information, and a thread gauge plug and thread ring gauge are used to detect thread defects of internal threads and external threads respectively.

[0004] In addition, there are also devices and methods for detecting defects on the inner wall of internal threads based on machine vision detection methods. For example, patent publication number CN103673880A discloses a visual inspection system and inspection method for the inner wall of a micro-hole based on a composite reflector. The inspection system of this application includes: a camera, a telecentric lens, a coaxial light source, a rotating stage, and a base platform arranged on a bracket from top to bottom; a composite reflector arranged on the base platform; a through hole in the middle of the rotating stage; a workpiece to be measured mounted above the through hole in the middle of the rotating stage; the composite reflector passes through the through hole in the middle of the rotating stage and penetrates into the aperture of the workpiece to be measured; the composite reflector includes a conical reflector and a plane reflector; the plane reflector is cut downward along the top angle of the conical reflector. This application can provide a high-brightness light source to be incident on the interior of the hole, thereby obtaining a clear image of the inner wall and achieving lossless and clear acquisition of the inner wall image.

[0005] For example, patent publication number CN103575748A discloses an optical inspection system for the inner wall of a workpiece with a small aperture. In this application, parallel light emitted by a coaxial light source enters the workpiece hole vertically downward, and is reflected by a spherical reflector to illuminate the inner wall of the workpiece. The reflected light on the inner wall of the workpiece is then reflected upward by the spherical reflector, passes through the light-transmitting surface of the coaxial light source, and enters the telecentric lens, thereby forming a reflected image in the CCD. A circular image at a certain height on the inner wall of the workpiece can form a corresponding circular image in the CCD after reflection. This application solves the problem of difficulty in obtaining images of the inner wall of a workpiece with a small aperture.

[0006] While all of the aforementioned applications can be used to detect defects on the inner wall of threads and achieve good measurement accuracy, they all involve direct contact with the interior of the threads, which can cause damage to the blind hole, particularly to the fine threads within. Therefore, achieving non-destructive testing of blind hole depth is an urgent challenge in the field of precision machining. Summary of the Invention

[0007] 1. Technical problem to be solved by the invention

[0008] In view of the above-mentioned deficiencies in the prior art, the present invention provides a blind hole depth detection method and device based on coaxial illumination dual-camera single lens. The present invention realizes non-contact and non-destructive detection of blind hole depth, effectively avoiding damage to the precision threads in the blind hole caused by human operation. In addition, the use of data algorithm detection has the advantages of avoiding human misjudgment, fast detection speed and efficient operation, and is suitable for use in industrial manufacturing sites.

[0009] 2. Technical solution

[0010] In order to achieve the above object, the technical solution provided by the present invention is:

[0011] The present invention provides a blind hole depth detection method based on coaxial lighting dual-camera single lens, the steps of which are as follows:

[0012] Step S100: Calibrate and debug the camera clear focus imaging plane spacing H so that the first camera and the second camera, which are perpendicular to each other, are focused on the bottom and the opening of the blind hole to be inspected, respectively, and collect images to be processed at the two positions respectively;

[0013] Step S200: performing image acquisition and image preprocessing for blind hole depth measurement;

[0014] Step S300: performing focus clarity calculation and detection on the image obtained by the pre-processing in step S200 to obtain a clarity evaluation function value of each corresponding image;

[0015] Step S400: Repeat steps S200 and S300 for the clear images of the hole bottom plane and the hole mouth plane obtained in step S100, and calculate the grayscale mean function value of the high-frequency component of the clear images of the hole bottom plane and the hole mouth plane;

[0016] Step S500: determining the relationship between the high-frequency component grayscale mean function and depth information;

[0017] Step S600: Measure and calculate the depth information of the blind hole to be detected according to the relationship obtained in step S500.

[0018] The present invention discloses a blind hole depth detection device based on a coaxially illuminated dual-camera single-lens system. The depth detection device is used to obtain an image to be processed in the blind hole depth detection method. The device comprises an image acquisition device, an illumination device, and a mobile platform. The image acquisition device comprises a first camera, a second camera, a beam splitter prism, and a microscope objective lens. The first camera and the second camera are arranged perpendicular to each other and are both fixed to the side of the beam splitter prism. The beam splitter prism is connected to the microscope objective lens via the illumination device, and a mobile platform is arranged below the microscope objective lens.

[0019] Furthermore, the lighting device includes a semi-transparent and semi-reflective mirror and a coaxial lighting point light source, the coaxial lighting point light source is arranged on the side of the semi-transparent and semi-reflective mirror, and the semi-transparent and semi-reflective mirror are respectively connected to the dichroic prism and the microscope objective lens.

[0020] Furthermore, the first camera and the second camera are both connected to a host computer.

[0021] 3. Beneficial effects

[0022] Compared with the existing known technologies, the technical solution provided by the present invention has the following significant effects:

[0023] (1) The blind hole depth detection method based on coaxial illumination dual-camera single lens of the present invention realizes non-contact non-destructive detection of the blind hole depth, effectively avoiding damage to the precision thread in the blind hole caused by human operation, and the use of data algorithm detection has the advantages of avoiding human misjudgment, fast detection speed and efficient operation, and is suitable for use in industrial manufacturing sites.

[0024] (2) The present invention provides a blind hole depth detection method based on a coaxially illuminated dual-camera single-lens system, which uses a dual-camera single-lens structure to achieve one-click depth measurement. Compared with other image-based depth measurement methods, the present invention effectively reduces the depth measurement process and is suitable for real-time and rapid measurement requirements in industrial sites.

[0025] (3) The blind hole depth detection method based on coaxial illumination dual-camera single lens of the present invention uses a curve fitting method to determine the relationship between the image clarity function value in the system and the depth position of the detection object surface, effectively reducing the time of image processing depth calculation, improving the detection efficiency of the system, and also ensuring the accuracy of depth measurement to meet the processing requirements in industrial production. BRIEF DESCRIPTION OF THE DRAWINGS

[0026] Figure 1 Schematic diagram of the structure of the blind hole depth detection device in the present invention;

[0027] Figure 2 This is a flow chart for blind hole depth detection;

[0028] Figure 3 This is a schematic diagram of the blind hole depth calculation principle;

[0029] Figure 4 FIG. 4 is an illustration of step S620;

[0030] Figure 5 (a) is a graph showing the relationship between the grayscale mean function value score of the high-frequency component of the bottom hole image and the depth position d; Figure 5 (b) is a graph showing the relationship between the grayscale mean function value score of the high-frequency component of the orifice image and the depth position d;

[0031] Figure 6 The data graph shows the results of measuring the blind hole depth for the experimental test system.

[0032] Figure 1 Description of the labels in:

[0033] 1. First camera; 2. Second camera; 3. Beam splitter; 4. Semi-transparent and semi-reflective mirror; 5. Coaxial point light source; 6. Microscope objective lens; 7. Workpiece to be measured; 8. Mobile platform; 9. Host computer. DETAILED DESCRIPTION

[0034] In order to further understand the content of the present invention, the present invention is described in detail with reference to the accompanying drawings and embodiments.

[0035] Example 1

[0036] As the integration of workpieces increases, blind holes are increasingly used inside precision devices. The depth of the blind hole directly affects the tightness of the workpiece after assembly. If the depth of the blind hole does not meet the workpiece assembly standard, the fixing bolts will not be installed tightly enough when the workpiece is assembled, affecting the airtightness of the workpiece. If the depth of the blind hole exceeds the reserved error range, the workpiece will become loose after assembly. Although the existing direct contact measurement and detection method deep inside the blind hole has good measurement accuracy, direct contact measurement and detection will cause certain damage to the inside of the blind hole, especially to the fine threads inside the blind hole. In addition, contact detection measurement requires high labor costs and is time-consuming. Manual detection has many human factors, which is not conducive to the needs of fast and efficient industrial production.

[0037] In view of this, this embodiment provides a blind hole depth detection device based on coaxial lighting dual-camera single lens. Figure 1 , the blind hole depth detection device comprises:

[0038] 1) Image acquisition device: This device primarily consists of a first camera 1, a second camera 2, a beam splitter prism 3, and a microscope objective lens 6. It is used to capture a planar image of the bottom and opening of the target blind hole and transmit the image information of these two locations to a host computer 9 connected to the image acquisition device. The host computer 9 processes and analyzes the characteristic parameter information of the bottom and opening images.

[0039] 2) Coaxial light source lighting equipment: including a semi-transparent and semi-reflective mirror 4 and a coaxial lighting point light source 5, used to illuminate the field of view area in front of the microscope objective 6 to obtain bright and clear plane images of the bottom and opening of the workpiece 7 to be measured.

[0040] 3) Horizontal two-dimensional mobile detection platform: Figure 1 The movable platform 8 is used to place the workpiece 7 to be measured and adjust the position of the workpiece so that the blind hole to be detected on the workpiece 7 to be measured appears within the field of view of the image acquisition device.

[0041] The process of performing blind hole depth detection using the blind hole depth detection device is as follows:

[0042] 1) System camera focal plane distance calibration:

[0043] Place the workpiece 7 to be measured with a blind hole depth of D on the mobile platform 8. First, adjust the platform height so that the camera farther away from the microscope objective lens 6 can clearly image the hole opening of the standard workpiece. Then fix the camera and the platform height, and adjust the position of the other camera so that the image of the plane where the bottom of the workpiece 7 to be measured is clearly imaged.

[0044] 2) Image preprocessing and target area extraction:

[0045] The images of the two positions collected and retained are preprocessed, and the target image of the hole bottom plane position and the target image of the hole mouth plane position are detected and extracted respectively, and reserved for the next step.

[0046] 3) Detection of high-frequency component image clarity:

[0047] The amount of high-frequency grayscale information in the image to be detected directly affects the clarity of the image. When the depth of the blind hole is too large or too small, the imaging object surface is in a defocused position, and the camera used to image the bottom plane of the hole can only capture a blurred image of the bottom plane of the hole. How to establish the relationship between the depth of the blind hole and the blurred image is crucial. In view of the fact that the high-frequency components in the image usually directly reflect the clarity of the image, this embodiment establishes a corresponding relationship between the image grayscale high-frequency component clarity function in the blurred image and the depth of the plane to be measured. After using a Gaussian high-pass filter to filter out the low-frequency component information in the blurred image to be detected, the grayscale mean function score of the high-frequency component information in the blurred image is calculated to represent the clarity of the image.

[0048] 4) Determine the relationship between the grayscale mean function value of the high-frequency component of the image and the depth information:

[0049] The blind hole is placed within the detection field of view of the system, and the position and depth of the blind hole are changed. The target image is moved from a near-defocused position to a focused position and then to a far-focused position. The clarity of the high-frequency component image is recorded in sequence, and the relationship between the grayscale mean function value score of the high-frequency component of the image and the position information d is calculated (score = T(d).

[0050] 5) Obtain the focused surface images of the two cameras of the system:

[0051] The workpiece to be measured with a blind hole depth of D0 is placed on the mobile platform 8, and the height of the mobile platform 8 is adjusted so that the system camera sequentially obtains the hole bottom plane image and the hole mouth plane image on their respective focusing surfaces.

[0052] 6) Calculate the depth of blind hole:

[0053] According to the relationship between the high-frequency component grayscale mean function score and position information of the bottom plane image and the hole opening plane image of the blind hole to be detected, E h , calculate the blind hole depth.

[0054] Figure 3 In the figure, 1 is the imaging plane of industrial camera 1, 2 is the hole bottom focus plane, 3 is the imaging plane of industrial camera 2, 4 is the aperture focus plane, 5 is the imaging prism, 6 is the aperture plane of the blind hole to be inspected, and 7 is the bottom plane of the blind hole to be inspected. Points P and Q are object points on the aperture plane and hole bottom plane, respectively; P' and Q' are image points on the corresponding focus planes; ΔP" and ΔQ" are blurred image spots on the phase plane of the industrial camera, respectively. H is the distance between the two clearly focused imaging planes; Δd1 is the distance between the image plane at the bottom of the blind hole to be inspected and the focus plane; Δd2 is the distance between the image plane at the aperture of the blind hole to be inspected and the focus plane. The specific formula for calculating the blind hole depth is D = H - Δd2 + Δd1.

[0055] This embodiment realizes non-contact non-destructive testing of the blind hole depth, effectively avoiding damage to the precision threads in the blind hole caused by human operation. In addition, the use of data algorithm detection has the advantages of avoiding human misjudgment, fast detection speed, and efficient operation, and is suitable for use in industrial manufacturing sites.

[0056] Example 2

[0057] Combine Figure 2 The present embodiment provides a blind hole depth detection method based on coaxial illumination dual-camera single-lens, and the specific steps are as follows:

[0058] Step S100: Calibrate and debug the system camera to clearly focus the imaging plane distance H, so that Figure 1 The first camera 1 and the second camera 2 are focused on the bottom and the opening of the blind hole to be inspected respectively, and the positions of the components are fixed unchanged, and the images to be processed at the two positions are collected respectively. and It includes the following sub-steps:

[0059] Step S110 , turning on the coaxial illumination point light source 5 , adjusting the acquisition control parameters of the first camera 1 and the second camera 2 to ensure that the two cameras capture clear images picture 1 and picture 2 ;

[0060] Step S120: Place the blind hole with a depth of D0 on the mobile platform 8, and adjust the positions of the first camera 1 and the second camera 2 in the system so that the two cameras focus on the upper and lower surfaces of the blind hole respectively to obtain a clear image of the bottom position of the focused hole. and focus aperture position image Thus, it is determined that the size of the clear focus imaging plane distance H of the system camera satisfies H=D0, and the specific operation is step S121;

[0061] Step S121, in order to obtain the most accurate focused image and When adjusting the camera, the Tenengrad gradient function is used to detect the image. When the Tenengrad detection value reaches the maximum, the image at that position is the best focus position image.

[0062] Step S200: After the step S100 of calibrating the system camera to clearly focus the imaging plane distance H is completed, the overall structural parameters of the system are determined, and then the image acquisition and preprocessing steps of the system depth measurement are carried out. Specifically, it includes the following sub-steps:

[0063] Step S210: During image acquisition, the blind hole is placed within the field of view of the detection system, and two industrial cameras in the system are used to respectively acquire target images of the hole bottom plane and the hole mouth plane;

[0064] Step S220, pre-processing the target images of the hole bottom plane and the hole mouth plane by Gaussian high-pass filtering and image enhancement processing;

[0065] Step S230 , extracting an image picture′1 containing only a clear area of ​​the hole bottom by threshold segmentation on the pre-processed hole bottom image;

[0066] Step S240 , segmenting the pre-processed orifice image to extract an image picture′2 containing only a clear area of ​​the orifice.

[0067] Step S300: Perform image focus clarity calculation and detection on the target images picture'1 and picture'2 obtained by segmentation of the hole bottom and hole mouth, and obtain the clarity evaluation function value of each corresponding image and It includes the following sub-steps:

[0068] In step S310, the two images to be detected, picture'1 and picture'2, are Fourier transformed respectively, and Gaussian high-pass filtered using a Gaussian high-pass filter GHP to retain the high-frequency component information of the images, and then the images are transformed back to real-domain images.

[0069] Step S320: Perform focus detection on the filtered image. The detection algorithm uses the grayscale mean algorithm to obtain the clarity evaluation function value of the hole bottom and hole mouth images corresponding to the two camera focus planes. and

[0070] Step S400: Clearly image the hole bottom plane obtained in step S100 Clear imaging of the aperture plane Repeat steps S210 to S320 to calculate the high-frequency component grayscale mean function value of the clear image of the hole bottom plane and the hole mouth plane and

[0071] Step S500: Determine the relationship between the high-frequency component grayscale mean function and depth information. Step S500 specifically includes the following sub-steps:

[0072] Step S510, adjust the height d of the loading platform i Repeat steps S210 to S230 to obtain the depth d i The corresponding image of the system is picture′ 1-i and picture′ 2-i ;

[0073] Step S520, repeating steps S310 to S320 to sequentially obtain the clarity function value of each height position, where: the bottom plane of the hole is recorded as The hole plane is recorded as In the symbols of the clarity function values, the subscripts 1 and 2 represent the hole bottom position and the hole mouth position respectively, and the superscript d i Indicates the depth position of the corresponding object surface.

[0074] Step S530: normalize the clarity function value to obtain normalized clarity function values ​​score1 and score2, where the normalized clarity function value of the hole bottom plane is recorded as The normalized clarity function value of the aperture plane is denoted as

[0075] Step S540: use the normalized sharpness function value from the near-out-of-focus position to the far-out-focus position obtained in step S530 and the corresponding depth position d i Constructing data pairs and The relationships between the image clarity functions of the high-frequency grayscale components of the hole bottom and hole mouth and the depth information d are fitted respectively, where the hole bottom plane is recorded as score1=T(d) and the hole mouth plane is recorded as score2=T(d).

[0076] Combine Figure 5 , Figure 5 (a) is a graph showing the relationship between the grayscale mean function value score of the high-frequency component of the bottom hole image and the depth position d; Figure 5 (b) is a graph showing the relationship between the grayscale mean function value score of the high-frequency component of the orifice image and the depth position d.

[0077] After the system is calibrated through steps S100 to S500, it is used to measure blind holes on a workpiece. The measurement results are shown in the table. In the table, D represents the measured depth of the blind hole to be inspected; D0 represents the average depth value measured using the depth gauge over multiple times. Positions 1 and 2 in the table represent the depth measurements of the same blind hole at different inspection positions; positions 2, 3, 4, and 5 represent the depth measurements of different blind holes at the same inspection position.

[0078] Position 1 Position 2 Position 3 Position 4 Position 5 D 8.686 8.709 8.768 8.684 8.797 <![CDATA[D0]]> 8.70 8.70 8.75 8.7 8.82 Relative error -0.014 0.009 0.018 -0.016 -0.023 Absolute error 0.0016 0.0010 0.0020 0.0018 0.0026

[0079] Step S600: After the operations from Step S100 to Step S500, the blind hole depth measurement device is calibrated and can be used for blind hole depth measurement. To achieve one-touch operation of the system for depth measurement, optimization is made during actual measurement. The depth information of the blind hole to be detected is measured and calculated according to the relationship obtained by measurement fitting in Step S530. Step S600 specifically includes the following sub-steps:

[0080] Step S610: Place the blind hole to be inspected on the mobile platform 8 and adjust Figure 1 The platform 8 is moved so that the blind hole to be detected is located in the center of the field of view, and steps S210 to S240 are repeated to obtain picture '1 and picture '2 of the measurement position, respectively. The point spread function (PSF) of the system at the corresponding position is used to restore the acquired blurred images to obtain clear images of the bottom plane and the hole mouth plane of the blind hole, which are respectively recorded as and

[0081] Step S620: Perform Gaussian high-frequency filtering on the four images obtained in step S610 to obtain an image containing only high-frequency components, and calculate the grayscale mean function value of the high-frequency components of the image. Figure 4 . Among them: the clarity function value of the fuzzy image at the bottom of the hole Aperture fuzzy image clarity function value The clarity function value of the hole bottom restored focus image and aperture restoration focus image clarity function value

[0082] In step S630 , the clarity function values ​​of the hole bottom and hole mouth plane obtained in step S620 are normalized to obtain normalized function values ​​score1 and score2.

[0083] Denoted as: and

[0084] In step S640 , the normalized clarity function values ​​score1 and score2 calculated in step S630 are substituted into the relationship equations score1=T(d) and score2=T(d) obtained in step S540 to calculate the depth information d1 and d2 of the two positions to be measured.

[0085] Step S650: normalize the function value of the clearly focused imaging plane and Substitute score1 = T(d) and score2 = T(d) to calculate the depth position of the clear focus imaging plane at the bottom and mouth of the hole. and

[0086] Step S660, calculate the depth of the blind hole D = H - Δd2 + Δd1, where Δd represents the depth difference between the detection surface and the focusing surface, that is, and See the schematic diagram of blind hole depth calculation principle Figure 3 .

[0087] From the blind hole depth measurement data table and attached Figure 6 It can be seen that the system's depth measurement results for the same blind hole are close to the actual depth value, with an error stable at around 0.01mm. At the same time, the system's depth measurement results for multiple blind holes also show that the system has good stability in depth measurement and a certain degree of detection accuracy.

[0088] The above is a schematic description of the present invention and its embodiments, which is not restrictive. The drawings show only one embodiment of the present invention, and the actual structure is not limited thereto. Therefore, if a person skilled in the art is inspired by this and, without departing from the purpose of the present invention, designs a structure and embodiment similar to this technical solution without inventiveness, they shall fall within the scope of protection of the present invention.

Claims

1. A blind hole depth detection method based on coaxial lighting dual-camera single lens, characterized in that: The steps are: Step S100: calibrating and debugging the camera to clearly focus the imaging plane spacing H, so that the first camera (1) and the second camera (2) that are perpendicular to each other are focused on the bottom and the hole mouth of the blind hole to be inspected, respectively, and respectively collect the images to be processed at the two positions; Step S200: performing image acquisition and image preprocessing for blind hole depth measurement; Step S300: performing focus clarity calculation and detection on the image obtained by the pre-processing in step S200 to obtain a clarity evaluation function value of each corresponding image; Step S400: Repeat steps S200 and S300 for the clear images of the hole bottom plane and the hole mouth plane obtained in step S100, and calculate the grayscale mean function value of the high-frequency component of the clear images of the hole bottom plane and the hole mouth plane; Step S500: determining the relationship between the high-frequency component grayscale mean function and depth information; Step S600: Measure and calculate the depth information of the blind hole to be detected according to the relationship obtained in step S500.

2. The blind hole depth detection method based on coaxial illumination dual-camera single lens according to claim 1 is characterized in that: Step S100 specifically includes the following sub-steps: Step S110, adjusting the acquisition control parameters of the first camera (1) and the second camera (2) to ensure that the two cameras acquire clear images picture1 and picture2; Step S120: Place the blind hole with a depth of D0 on the mobile platform, adjust the positions of the first camera (1) and the second camera (2) in the system so that the two cameras focus on the upper and lower surfaces of the blind hole respectively to obtain a clear image of the bottom position of the focused hole. and focus aperture position image 3. The blind hole depth detection method based on coaxial illumination dual-camera single lens according to claim 2, characterized in that: In step S120 , when adjusting the camera, the Tenengrad gradient function is used to detect the image. When the Tenengrad detection value reaches the maximum, the image at that position is the best focus position image.

4. The blind hole depth detection method based on coaxial illumination dual-camera single lens according to claim 2 or 3, characterized in that: Step S200 pre-processes the target images of the hole bottom plane and the hole mouth plane by Gaussian high-pass filtering and image enhancement processing; and extracts the image picture1′ containing only the clear area of ​​the hole bottom and the image picture′2 containing only the clear area of ​​the hole mouth by threshold segmentation from the pre-processed images.

5. The blind hole depth detection method based on coaxial illumination dual-camera single-lens according to claim 4, characterized in that: Step S300 specifically includes the following sub-steps: Step S310: Fourier transform the two images to be detected, picture '1 and picture '2, respectively, and perform Gaussian high-pass filtering on the images to retain the high-frequency component information of the images, and then transform the images back to real-domain images; Step S320: Use the grayscale mean algorithm to perform focus detection on the filtered image to obtain the clarity evaluation function value of the hole bottom and hole mouth images corresponding to the two camera focus planes. and 6. The blind hole depth detection method based on coaxial illumination dual-camera single-lens according to claim 5, characterized in that: Step S500 specifically includes the following sub-steps: Step S510, adjust the height d of the loading platform i , step S200, obtain depth d i The corresponding image of the system is picture′ 1-i and picture′ 2-i ; Step S520, repeat step S300 to obtain the clarity function value of each height position in turn, where: the bottom plane of the hole is recorded as The hole plane is recorded as Step S530, normalize the clarity function value to obtain normalized clarity function values ​​score1 and score2, where: the normalized clarity function value of the hole bottom plane is Normalized clarity function value of aperture plane in, and are the grayscale mean function values ​​of the high-frequency components of the clear images of the hole bottom plane and the hole mouth plane respectively; Step S540: use the normalized sharpness function value from the near-out-of-focus position to the far-out-focus position obtained in step S530 and the corresponding depth position d i Constructing data pairs and The relationship between the image clarity function of the high-frequency grayscale component at the bottom and mouth of the hole and the depth information d was fitted respectively.

7. The blind hole depth detection method based on coaxial illumination dual-camera single-lens according to claim 6, characterized in that: Step S600 includes the following sub-steps: Step S610: Place the blind hole to be inspected on a mobile platform so that the blind hole to be inspected is located in the center of the field of view. Acquire picture '1' and picture '2' in the center of the field of view respectively, and use the point spread function to restore the acquired blurred images to obtain clear images of the bottom plane and the hole mouth plane of the blind hole, which are respectively denoted as and Step S620, performing Gaussian high-frequency filtering on the four images obtained in step S610 to obtain an image containing only high-frequency components, and calculating the grayscale mean function value of the high-frequency components of the image; wherein: the clarity function value of the hole bottom fuzzy image Aperture fuzzy image clarity function value The clarity function value of the hole bottom restored focus image and aperture restoration focus image clarity function value Step S630, normalizing the clarity function values ​​of the hole bottom and hole mouth plane obtained in step S620, and calculating normalized function values ​​score1 and score2; Step S640 , substituting the normalized clarity function values ​​score1 and score2 calculated in step S630 into the relationship obtained in step S540 to calculate depth information d1 and d2 of the two positions to be measured; Step S650: normalize the function value of the clearly focused imaging plane and Substitute the relationship obtained in step S540 to calculate the depth position of the clear focus imaging plane at the bottom of the hole and the hole mouth. and Step S660, calculate the depth of the blind hole D = H - Δd2 + Δd1, where Δd represents the depth difference between the detection surface and the focusing surface, that is, and

Images