A transparent piece defect detection method and system based on double optical path phase deflectometry

Abstract

The application discloses a transparent piece defect detection method and system based on double optical path phase deflection, and belongs to the technical field of optical detection. The system comprises a camera, a reflection light source, a beam splitter, a back light source and a processing unit. The detection method comprises first measurement, product overturning, second measurement, and defect comprehensive analysis and classification. The depth map and modulation map of the product surface are obtained through reflection measurement, the phase difference map and back light modulation map are obtained through transmission zero reference difference measurement, multi-dimensional data are fused, and accurate detection and classification of defects such as concave-convex points, foreign matters and scratches are realized according to preset logic. The application can effectively eliminate inherent errors of the system, realize double-side full detection without omission, has high detection precision, accurate defect classification, and is suitable for defect detection of transparent pieces such as multi-layer transparent glass.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of optical detection technology, in particular to a method and system for detecting internal and external surface and internal defects of ultra-thin glass, multi-layer composite glass and other transparent objects by phase measuring deflectometry. BACKGROUND

[0002] At present, the defect detection of multi-layer transparent glass such as mobile phone cover plate and vehicle-mounted screen mainly relies on manual inspection or traditional machine vision technology.

[0003] Manual inspection: It is heavily dependent on the experience of workers, and has problems such as low efficiency, non-uniform standards, fatigue leading to missed or misdiagnosed defects, and cannot meet the production requirements of large scale and high precision.

[0004] Traditional machine vision: The camera directly captures the image of the product under ordinary illumination, and the detection rate of weak defects such as surface scratches, internal bubbles, inclusions, concave-convex points and surface foreign matters of glass and other transparent objects is very low. Because these defects themselves do not emit light or absorb a large amount of light, only a small disturbance is caused to the propagation path of transmitted or reflected light, and the contrast in ordinary images is extremely low, making it difficult to be found.

[0005] Single-path phase deflectometry: The existing technology uses single reflection or transmission phase deflectometry for detection. The reflection type is mainly sensitive to the front surface topography and defects of the sample; the transmission type obtains the cumulative phase change after the light passes through the entire sample. Both of these two methods are difficult to effectively separate and locate the front surface defects, back surface defects, and material internal inclusions and layer scratches, which can easily cause confusion in defect attribution. In addition, the geometric errors of the optical system (such as camera lens distortion, uneven light source screen) will directly superimpose on the measurement results, affecting the detection accuracy, and a complex and expensive calibration process is needed to eliminate or weaken these system errors.

[0006] Therefore, how to accurately and efficiently detect and accurately classify the internal and external comprehensive defects of transparent objects, and effectively distinguish the defect sources and eliminate the system errors, is a technical problem to be solved in the field. SUMMARY

[0007] The purpose of the present application is to overcome the shortcomings of the prior art, and to provide a detection method and system that can accurately detect and classify various defects such as concave-convex points, surface foreign matters, layer inclusions, surface and layer scratches of multi-layer transparent glass and other transparent objects, and can effectively eliminate the system inherent errors and ensure double-sided full detection without omission.

[0008] The application provides a transparent part defect detection system based on a double optical path phase deflection method, comprising a camera, a reflective light source, a beam splitter, a back transmission light source and a processing unit; the camera is arranged directly above a measured product and is used for collecting a fringe image; the reflective light source is used for projecting multi-frequency phase shift fringe light; the beam splitter is arranged between the camera and the measured product and can reflect the fringe light projected by the reflective light source to the upper surface of the measured product; the back transmission light source is arranged directly below the measured product, is a programmable screen and is used for projecting multi-frequency phase shift fringe light to penetrate the measured product; the processing unit is electrically connected with the camera, the reflective light source and the back transmission light source, is used for controlling the projection timing of the reflective light source and the back transmission light source, processing the image collected by the camera and executing a defect detection algorithm. The application realizes synchronous detection of the surface and internal defects of the transparent part through the double optical path (reflection and transmission) design; the arrangement of the beam splitter ensures that the reflection light path and the imaging light path do not conflict, improving the utilization rate of the light path; the centralized control of the processing unit ensures the accuracy of the measurement timing and the continuity of the data processing, solving the problem that the single optical path cannot distinguish the defect position (surface / interior).

[0009] Further, the camera is a 50-80 million pixel industrial camera, preferably a 65 million pixel industrial camera.

[0010] Further, the back transmission light source is an LCD screen or an OLED screen, which is used as a programmable light source and can flexibly adjust the frequency, phase and brightness of the multi-frequency phase shift fringe, adapting to the detection requirements of transparent parts (such as glass and acrylic) with different thicknesses and different materials.

[0011] In the application, the reflective light source, the beam splitter and the camera work cooperatively to form a reflective measurement light path. The reflective light source provides specific fringe light, the beam splitter changes the light path to make the light irradiate to the measured surface, and the camera is responsible for collecting the reflected fringe image, and the three cooperate to realize the reflective phase deflection measurement of the surface (A surface and B surface) of the product, providing a data basis for obtaining the depth map and the modulation map of the surface.

[0012] The back transmission light source and the camera cooperatively form a transmission measurement light path, the fringe light projected by the back transmission light source is collected by the camera after penetrating the measured product, and through comparison with the reference phase when there is no sample, a phase difference map reflecting the internal and double surface comprehensive conditions of the product and a back light modulation map are obtained, which are complementary to the data obtained by the reflective measurement.

[0013] The processing unit as the core cooperates with the camera and the light source to control the projection timing of the light source, ensures the accuracy and synchronization of image acquisition, processes and analyzes the images collected by the camera, comprehensively measures the multidimensional data obtained by reflection and transmission, and realizes accurate identification and classification of defects by using defect classification logic. The product turning step cooperates with the front and rear measurements, so that the system can detect two surfaces of the product respectively, and realizes double-sided full detection without omission.

[0014] The application also provides a transparent part defect detection method based on a double-light-path phase deflection method.

[0015] First measurement: start the reflection light source, perform reflection phase deflection measurement on the A surface of the measured product, the camera collects the reflection fringe image, and the processing unit calculates the depth map reflecting the three-dimensional topography of the A surface and the reflection modulation map reflecting the fringe contrast according to the collected reflection fringe image; at the same time, start the back transmission light source, the back transmission light source projects multi-frequency phase shift fringe light to penetrate the measured product, the camera collects the transmission fringe image, and performs zero reference difference measurement, that is, first collect the reference phase image without a sample, and then collect the measurement phase image with a sample to obtain the phase difference image, and the processing unit calculates the back light modulation map according to the transmission fringe image;

[0016] Product turning: after the first measurement is completed, the measured product is turned over so that the B surface faces the camera;

[0017] Second measurement: repeat the reflection phase deflection measurement step on the A surface in the first measurement step to perform reflection phase deflection measurement on the B surface of the product, the camera collects the reflection fringe image, and the processing unit obtains the depth map and the reflection modulation map of the B surface according to the collected reflection fringe image;

[0018] Defect comprehensive analysis and classification: the processing unit fuses all the measurement data, and identifies and classifies the defects according to the preset logic. The application realizes comprehensive coverage detection of the A surface, the B surface and the internal defects of the transparent part through the combination of double-surface reflection measurement and transmission difference measurement; the zero reference difference measurement eliminates the inherent error of the system; the product turning step ensures that the double-surface detection conditions are consistent, and avoids the omission problem of single-surface detection; the multi-data fusion analysis solves the misjudgment problem of defect types (such as surface foreign matter / interlayer foreign matter).

[0019] Further, in the reflection phase deflection measurement of the A surface and the B surface, the N-step phase shift method is used to extract the wrapped phase and the modulation, and the gray value Ii(x, y) of the i-th fringe image collected by the camera is represented as: I i (x, y) = A(x, y) + B(x, y)cos[Φ(x, y) + δ iwhere (x, y) is pixel coordinate, A(x, y) is background light intensity map, B(x, y) is fringe amplitude map, Φ(x, y) is the phase to be solved, δ i =2πi / N is the phase shift of the i-th step, and N is the total number of steps. The N-step phase shift method effectively suppresses the noise interference of a single image through the phase superposition of multiple images.

[0020] Further, N is 4, and the four-step phase shift δ i =0, π / 2, π, 3π / 2. The wrapped phase map Φw(x, y) is simplified as: The four-step phase shift method reduces the number of image acquisitions while ensuring the accuracy of phase calculation.

[0021] Further, the calculation formulas of the fringe amplitude map B(x, y) and the background light intensity map A(x, y) are respectively: The modulation map M(x, y) is defined as the ratio of the fringe amplitude map to the background light intensity map, that is, The modulation map can directly reflect the degree of interference of the fringe by defects (such as fringe blur caused by foreign matter), and cooperate with the depth map to realize accurate classification of defects.

[0022] Further, it also includes a phase unwrapping step, which restores the continuous true phase Φ(x, y) through a phase unwrapping algorithm, that is, Φ(x, y)=Unwrap(Φw(x, y)).

[0023] Further, in the reflection measurement mode, the depth map h(x, y) of the object is proportional to the phase distortion , that is, where K is a proportional coefficient obtained through system calibration, which is related to the geometric structure of the system, and the depth map h(x, y) corresponds to the numerical value of the depth map in the reflection measurement. The proportional coefficient K quantizes the phase distortion into the actual height, so that the physical meaning of the depth map is clear, and it can be directly used for the measurement of defect size (such as protrusion height and recess depth). Depth map h(x, y): refers to the continuous true phase obtained by phase unwrapping in reflection phase deflection measurement, which is converted to generate image data reflecting the three-dimensional topography of the measured product surface in combination with the proportional coefficient K of system calibration, and its numerical value directly corresponds to the protrusion height or recess depth of the measured surface.

[0024] Further, the preset logic of defect recognition and classification includes:

[0025] Concave-convex point determination: concave-convex points appear as local concave or convex on the depth map obtained by reflection measurement, but there is no significant change on the reflection modulation map and the backlight modulation map;

[0026] ​Foreign matter determination: surface foreign matter appears as a protrusion on the depth map of the reflection measurement, and as a significant dark area on its corresponding reflection modulation map; sandwich foreign matter has no abnormality on the depth map of the reflection measurement, but has a significant disturbance on the phase difference map of the back transmission measurement, and presents a clear dark area on the back light modulation map; wherein the back transmission measurement is a process of penetrating the product to be measured by a back transmission light source, projecting multi-frequency phase shift fringe light, collecting a transmission fringe image by a camera, and performing zero reference surface differential measurement; scratch on the depth map or phase difference map;

[0027] Scratch determination: the defect area presents a high aspect ratio strip feature on the depth map or phase difference map. Through multi-dimensional data (depth, phase difference, modulation) feature matching, accurate classification of concave-convex points, surface foreign matter, sandwich foreign matter and scratches is realized; the preset logical quantitative standard avoids ambiguity of artificial discrimination, and ensures consistency of the detection result.

[0028] The nonlinear screen appearance defect evaluation method of the application has the following technical effects:

[0029] High-precision detection: by measuring the phase instead of the light intensity, the application is extremely sensitive to the slight light deflection caused by defects. The 65 million pixel industrial camera can capture subtle image changes, combined with the projection of multi-frequency phase shift fringe light and accurate phase calculation, so that the system can detect slight defects that are difficult to find by traditional methods, greatly improving the detection precision.

[0030] Elimination of system error: the zero reference surface measurement method effectively eliminates the influence of system inherent errors such as camera lens distortion, light source screen unevenness, etc. by subtracting the reference phase map without sample from the measurement phase map with sample, without the need for complex and expensive calibration process, reducing the system cost, and improving the reliability of the measurement result.

[0031] Defect classification accuracy: combined with the depth map, phase difference map and reflection / back light dual modulation map, and supplemented by geometric parameter analysis, a reliable defect classification logic is established. It can accurately distinguish different types of defects such as concave-convex points, surface foreign matter, sandwich foreign matter and scratches, and solve the problem of confusion in defect attribution in the prior art.

[0032] Double-sided full inspection without omission: through the process of product turning and double-sided measurement, the same high-precision detection of the front and back surfaces of the product is ensured. Reflection measurement is for surface defects, and transmission measurement reflects internal and comprehensive defects. The combination of the two makes the system able to comprehensively detect various defects of transparent parts, avoiding the omission that may occur in single-sided detection.

[0033] In order to more clearly illustrate the structural features and effects of the application, the application will be described in detail below in combination with the drawings and specific embodiments. BRIEF DESCRIPTION OF DRAWINGS

[0034] Figure 1 A hardware function module schematic diagram of a transparent part defect detection system based on double light path phase deflection method of the present application;

[0035] Figure 2 A flow chart of a transparent part defect detection method based on double light path phase deflection method of the present application.

[0036] Figure 3 A defect classification logic schematic diagram of a transparent part defect detection method based on double light path phase deflection method of the present application. DETAILED DESCRIPTION

[0037] The present application will be further described in conjunction with the drawings and relevant knowledge, and will be clearly and completely described. Obviously, the described application is only a part of the embodiments of the present application, but not all the embodiments.

[0038] Referring to Figure 1 As shown in the figure, embodiment 1: the present application provides a transparent part defect detection system based on double light path phase deflection method, which comprises a camera 1, a reflection light source 2, a beam splitter 3, a back transmission light source 4 and a processing unit.

[0039] Camera 1: a 6500 million pixel industrial camera set above the measured product 5, used for collecting fringe image;

[0040] Reflection light source 2: set on the side of the system, used for projecting multi-frequency phase shift fringe light.

[0041] Beam splitter 3: set between the camera and the measured product at a certain angle, which can reflect the fringe light projected by the reflection light source to the upper surface of the measured product.

[0042] Back transmission light source 4: set below the measured product 5, which is a programmable screen (such as LCD or OLED), used for projecting multi-frequency phase shift fringe light to penetrate the measured product.

[0043] Processing unit: electrically connected with the camera, reflection light source and back transmission light source, used for controlling the projection timing of the reflection light source and the back transmission light source, processing the image collected by the camera and executing the defect detection algorithm.

[0044] Embodiment 2, referring to Figure 2 , Figure 3 As shown in the figure, a transparent part defect detection method based on double light path phase deflection method, comprising the following core steps:

[0045] First measurement (A surface 6 detection):

[0046] Specifically, A-face 6 reflective measurement: start the reflective light source, and perform phase deflectometry measurement on the A-face 6 (the upper surface of the product 5) of the product under test. The processing unit calculates the depth map reflecting the three-dimensional topography of the A-face and the modulation map reflecting the contrast of the fringes according to the collected images.

[0047] Overall transmission measurement: start the back light source, and perform zero reference differential measurement. That is, first collect the reference phase map (Φ ref ) without sample, and then collect the measurement phase map (Φ test ) with sample, to obtain the phase difference map . At the same time, the processing unit also calculates the back light modulation map according to the transmission fringe image.

[0048] Product flipping:

[0049] In order to ensure that the defects on the upper and lower surfaces are not missed, after the first measurement is completed, the product under test is flipped by 180 degrees by a mechanical hand or manually, so that the B-face 7 (the lower surface of the product 5) faces the camera.

[0050] Second measurement (B-face 7 detection):

[0051] B-face reflective measurement: repeat the A-face reflective measurement steps, and perform reflective phase deflectometry measurement on the B-face of the product to obtain the depth map and the modulation map of the B-face.

[0052] Defect comprehensive analysis and classification: the processing unit fuses all the measurement data above, and identifies and classifies the defects according to the following logic:

[0053] Bump and pit determination: it is manifested as local concave or convex on the depth map obtained by reflective measurement, but there is usually no significant change on the reflective and back light modulation maps.

[0054] Foreign matter determination:

[0055] Surface foreign matter: it is manifested as convex on the depth map obtained by reflective measurement, and is manifested as a significant dark area (reduced modulation) on the corresponding reflective modulation map.

[0056] Layered foreign matter: there is no abnormality on the depth map obtained by reflective measurement, but there is significant disturbance on the phase difference map obtained by back light measurement, and a significant dark area is presented on the back light modulation map.

[0057] Scratch determination: the defect area presents a high aspect ratio strip feature on the depth map or the phase difference map.

[0058] This invention enables: high-precision detection, by measuring phase rather than light intensity, making it extremely sensitive to minute light deflections caused by defects; elimination of systematic errors, effectively eliminating inherent system errors through a zero-reference plane measurement method; accurate defect classification, establishing a reliable defect classification logic by combining depth maps, phase difference maps, and dual-modulation maps of reflection / backlight, supplemented by geometric parameter analysis; and complete double-sided inspection without omissions: through product flipping and double-sided measurement processes, equally high-precision inspection of both the front and back surfaces of the product is ensured.

[0059] Example 2: A method for detecting defects in transparent components based on dual-path phase deflection.

[0060] System setup: such as Figure 1 As shown, the system hardware includes a 65-megapixel industrial camera, a beam splitter, a sample stage, a back-transparent LCD screen, and a side-reflective light source. All components are connected to an industrial computer (processing unit).

[0061] The core of the transparent component defect detection method of the present invention lies in accurately extracting information such as phase and modulation from multiple acquired stripe images.

[0062] The N-step phase-shifting method extracts the grayscale value I of the i-th striped image acquired by the camera, which includes the wrapping phase and modulation. i (x,y) can be represented as:

[0063] I i (x,y)=A(x,y)+B(x,y)cos[Φ(x,y)+δ i ];

[0064] Where (x,y) are pixel coordinates, A(x,y) is the background light intensity, B(x,y) is the stripe amplitude, Φ(x,y) is the phase to be determined, and δ i =2πi / N is the phase shift amount at step i, and N is the total number of steps (N is usually taken as 4 in this embodiment).

[0065] By solving the above system of equations, we can obtain the wrapped phase diagram Φw(x,y) (whose range is (-π,π):

[0066] ;

[0067] For the commonly used four-step phase shift (δ) i =0,π / 2,π,3π / 2), the above equation simplifies to:

[0068] ;

[0069] Simultaneously, the amplitude diagram B(x,y) of the stripes and the background light intensity diagram A(x,y) can be calculated:

[0070] ;

[0071] ;

[0072] Modulation map M(x,y) is defined as the ratio of the amplitude to the background light intensity, reflecting the clarity of the fringe: ;

[0073] Foreign matter or partial surface coating defects will cause local light absorption or scattering, resulting in a sharp drop in B(x,y), thus appearing as a dark area in the modulation map M(x,y), which is the key basis for determining such defects.

[0074] The phase unwrapping Φw(x,y) obtained by the above method is wrapped in the interval (-π, π), and needs to be recovered to its continuous true phase Φ(x,y) through a phase unwrapping algorithm (such as a multi-frequency heterodyne method, a Gray code auxiliary method).

[0075] Φ(x,y)=Unwrap(Φw(x,y));

[0076] The zero reference difference method is the core of the present application in transmission measurement. By subtracting twice, the system error is eliminated.

[0077] ; wherein Φ ref is the system phase measured when empty, and Φ test is the phase measured after loading the sample. Accurately reflects the phase distortion caused only by the sample (including its internal and external defects).

[0078] Depth map reconstruction In the reflection measurement mode, the depth map h(x,y) of the object is directly proportional to the phase distortion (relative to the ideal plane) introduced by it.

[0079] ; wherein K is a proportional coefficient obtained by system calibration, which is related to the geometric structure of the system (such as the angle between the camera and the light source, the working distance, etc.). This depth map h(x,y) is directly used to identify three-dimensional topographic defects such as dents (Dent).

[0080] Detection process embodiment:

[0081] First stage: A surface detection, A surface reflection measurement: computer control reflection light source to project four-step phase shift fringe, camera collects I1, I2, I3, I4 four images. The processing unit calculates the depth map h A (x,y) and modulation map M A (x,y) of the A surface according to the above formula. Overall transmission measurement: computer control back transmission light source, first remove the glass, collect the reference fringe and calculate the continuous reference phase map Φref (x,y). Then put back the glass, collect sample fringe to calculate the measured phase map Φ test (x,y). Finally, the phase difference map ΔΦ(x,y) is obtained by subtraction and the backlight modulation map M BG (x,y).

[0082] Second stage: the measured glass is turned over 180 degrees by the mechanical arm in the product flip detection system, so that the B face is upward.

[0083] Third stage: B face reflection measurement: repeat the A face reflection measurement to perform reflection measurement on the B face to obtain the depth map h B (x,y) and the modulation map M B (x,y).

[0084] Fourth stage: the defect comprehensive analysis processing unit analyzes multi-dimensional data. The gradient of the phase difference map or the depth map is calculated to enhance the defect edge signal: ;

[0085] The initial contour of the defect is determined by threshold segmentation on the gradient map G(x,y). Then, the characteristic values of each data map in the contour are combined to perform classification according to the preset logic.

[0086] It should be noted that in the present application, when there is no measured transparent piece in the optical path, the reference fringe is projected by the back-through light source and collected by the camera to calculate the reference phase map (Φ ref ); the measured transparent piece is placed in the optical path, the measurement fringe is projected by the back-through light source, and the camera is collected to calculate the measured phase map (Φ test ); the measured phase map is subtracted from the reference phase map to obtain the phase difference map ΔΦ(x,y) reflecting the defects of the transparent piece itself, and the mathematical expression is: .

[0087] In the present application, at least one reflection phase deflection measurement is required for the transparent piece to obtain the three-dimensional depth map h(x,y) of its surface, and after the first measurement is completed, the measured transparent piece is turned over. The other surface of the turned-over measured transparent piece is repeatedly measured by reflection phase deflection measurement, and further, in the defect classification step, the classification step includes: judging the dent (Dent) defect according to the local height change in the depth map h(x,y);

[0088] ​In a preferred embodiment of the present application, a defect classification step is further included, the classification step comprising: calculating a reflection modulation map and a backlight modulation map M(x, y) from the collected reflection fringe image and transmission fringe image, respectively; determining surface foreign object defects according to the convex feature in the depth map and in combination with the modulation reduction in the reflection modulation map; determining sandwich foreign object defects according to the phase disturbance in the phase difference map and in combination with the modulation reduction in the backlight modulation map, or performing geometric parameter analysis on the defect area identified in the depth map or the phase difference map, and determining a scratch defect if the length-width ratio of the defect is greater than a preset threshold.

[0089] The present application realizes comprehensive detection of internal and external defects of a transparent piece through the core logic of double optical path measurement, product overturning and multi-dimensional data fusion analysis, through a reflection light path and a transmission light path, in combination with accurate phase calculation and defect classification algorithm. The following is a detailed description of the specific working steps:

[0090] System initialization and parameter configuration, hardware connection and calibration, after the system is started, the processing unit first completes the connection detection of each hardware module, including communication test of 6500 million pixel industrial camera, reflection light source (laterally arranged), beam splitter (45° inclined and placed between the camera and the sample table), back light source (LCD / OLED screen, located directly below the sample table). The processing unit sets the system geometric parameters such as the angle between the camera and the light source, the working distance, etc. through the pre-stored calibration parameters, determines the scale factor K in the depth map calculation, and ensures the conversion accuracy of the phase and the actual height.

[0091] Light source parameter configuration, the processing unit sends control instructions to the reflection light source and the back light source to set the parameters of the multi-frequency phase shift fringe:

[0092] Phase shift step number N=4; the fringe frequency is adjusted according to the thickness of the measured transparent piece.

[0093] First measurement (A surface detection stage):

[0094] Step 1: A surface reflection measurement (detecting upper surface defects);

[0095] Light path starts, the processing unit controls the reflection light source to start, the fringe light is vertically irradiated to the A surface (upper surface) of the measured transparent piece after being reflected by the beam splitter, and the reflected light enters the camera lens after being transmitted by the beam splitter.

[0096] Image acquisition, the camera continuously acquires 4 fringe images according to the phase shift timing, and each image corresponds to a different phase shift amount.

[0097] Data processing, wrapped phase calculation: processing unit calculates wrapped phase map by four-step phase shifting formula; phase unwrapping: uses multi-frequency heterodyne method to eliminate phase jumps and restore continuous true phase; depth map generation: generates A-surface three-dimensional depth map (reflects protrusions / recesses) according to the proportional relationship between depth map and phase distortion; reflection modulation degree map calculation: generates modulation degree map (reflects stripe clarity, foreign matter will cause local dark area) through the ratio of amplitude to background light intensity.

[0098] Step 2: overall transmission measurement (detects internal and double-sided comprehensive defects);

[0099] Reference phase acquisition (zero reference surface calibration), processing unit controls robot to remove the measured piece on the sample table, back light source is turned on and projects four-step phase shift stripes, camera acquires reference stripe image without sample, and system inherent phase (including lens distortion, screen unevenness and other system errors) is calculated.

[0100] Measurement phase acquisition, robot puts the measured piece back on the sample table (A surface upward), back light source projects the same stripes again, light is collected by the camera after penetrating the transparent piece, and the measurement phase containing sample defects is calculated.

[0101] Phase difference and back light modulation degree calculation, phase difference map generation: system errors are eliminated by difference: the map only reflects the light deflection caused by internal (such as interlayer foreign matter) and double-sided defects of the sample.

[0102] Back light modulation degree map calculation: the same method as reflection measurement is used.

[0103] Product turnover (realizes double-sided detection without omission), processing unit controls robot to clamp the measured transparent piece, rotates 180° along the vertical axis, so that the original lower surface (B surface) faces upward, ensures that the B surface and the A surface are in the same optical path position during measurement, and avoids detection errors caused by geometric deviation. After turning over, the sample table is repositioned to ensure that the center of the measured piece is aligned with the optical axis of the camera.

[0104] Second measurement (B surface detection stage), all steps of "A surface reflection measurement" are repeated to detect the turned-over B surface:

[0105] Reflection light source projects stripes, camera acquires four stripe images reflected by the B surface; processing unit calculates B surface depth map and reflection modulation degree map; data is stored in the cache area of the processing unit and is associated with A surface and transmission measurement data.

[0106] Defect comprehensive analysis and classification, processing unit fuses measurement data to realize defect identification through the following logic:

[0107] Concave-convex point determination, h A and h BLocal gradient analysis is performed, if there is a height mutation (such as a depression / protrusion) of continuous pixels, and the M A , M B , M BG No significant change, determined as a concave convex point.

[0108] Surface foreign matter determination, if h A (or h B ) has an isolated protruding area, and the area shows a significant decrease in modulation in M A (or M B ), it is determined as a surface foreign matter (foreign matter shielding causes the stripe definition to decrease) of A face (or B face).

[0109] Laminated foreign matter determination, if h A and h B are normal, but there is a local phase disturbance in h , and the corresponding position M BG modulation decreases, it is determined as an internal laminated foreign matter (foreign matter scattered light causes phase distortion and modulation to decrease).

[0110] Scratch determination, edge detection is performed on the depth map or phase difference map, and the length-width ratio of the defect area is calculated: if the length / width>5 (preset threshold), and it is continuously distributed in the form of a strip, it is determined as a scratch (surface scratches are developed in h A / h B , and internal scratches are developed in h ).

[0111] Result output and feedback, the processing unit generates a detection report (defect type, position coordinates, size parameters) of the detection result, displays it in real time through the display screen, and triggers a sorting signal:

[0112] In the present application, the reflected light path focuses on the surface three-dimensional topography (depth map) and surface shielding (reflection modulation), and the transmitted light path focuses on the internal defect (phase difference map) and internal shielding (backlight modulation). The data of the two are complementary, realizing full defect coverage of the surface and the interior.

[0113] Through the fusion analysis of multi-dimensional data (depth, phase difference, modulation), the limitation of single parameter detection is broken through.

[0114] The technical principles of the present application are described above in combination with specific embodiments, which are only preferred embodiments of the present application. The protection scope of the present application is not limited to the above-mentioned embodiments. Any technical solution falling within the idea of the present application shall fall within the protection scope of the present application. Other specific embodiments of the present application can be conceived by those skilled in the art without creative labor, and these embodiments shall fall within the protection scope of the present application.

Claims

1. A method for detecting defects in a transparent piece based on two-beam phase deflectometry, characterized in that, The method comprises the following steps: Step S1: start the reflection light source, perform reflection phase deflection measurement on the A surface of the measured product, the camera collects the reflection fringe image, and the processing unit calculates the depth map reflecting the three-dimensional topography of the A surface and the reflection modulation map reflecting the fringe contrast according to the collected reflection fringe image; at the same time, start the back transmission light source, the back transmission light source projects multi-frequency phase shift fringe light through the measured product, the camera collects the transmission fringe image, and performs zero reference plane differential measurement, that is, first collect the reference phase map without the sample, and then collect the measurement phase map with the sample to obtain the phase difference map, and the processing unit calculates the back light modulation map according to the transmission fringe image; Step S2: after the first measurement is completed, the measured product is turned over, and the B surface faces the camera; Step S3: repeat the reflection phase deflection measurement step on the A surface in step S1 to perform reflection phase deflection measurement on the B surface of the product, the camera collects the reflection fringe image, and the processing unit obtains the depth map and the reflection modulation map of the B surface according to the collected reflection fringe image; Step S4: the processing unit fuses all the measurement data, and identifies and classifies defects according to a preset logic; The camera is arranged directly above the measured product; the reflection light source is used to project multi-frequency phase shift fringe light; the beam splitter is arranged between the camera and the measured product, and can reflect the fringe light projected by the reflection light source to the upper surface of the measured product; the back transmission light source is arranged directly below the measured product and is a programmable screen; the processing unit is electrically connected with the camera, the reflection light source and the back transmission light source, and is used to control the projection timing of the reflection light source and the back transmission light source, process the images collected by the camera and execute a defect detection algorithm.

2. The method of claim 1, wherein, The camera is a 50-80 million pixel industrial camera.

3. The method of claim 1, wherein, The back transmission light source is an LCD screen or an OLED screen.

4. The method of claim 1, wherein, In the reflective phase deflectometry of the A and B surfaces, the N-step phase shift method is used to extract the wrapped phase and modulation. The gray value I of the i-th fringe image captured by the camera is i The expression of (x, y) is: I i (x,y) = A(x,y) + B(x,y) cos [Φ(x,y) + δ i ], where (x,y) is pixel coordinate, A(x,y) is background intensity map, B(x,y) is fringe amplitude map, Φ(x,y) is the phase to be solved, δ i = 2πi / N is the phase shift of the i-th step, and N is the total number of steps.

5. The method of claim 4, wherein, N is 4, four-step phase shift delta i = 0, π / 2, π, 3π / 2, the wrapped phase map Φw(x, y) simplifies to: .

6. The method of claim 5, wherein, The formulae for calculating the fringe amplitude map B(x, y) and the background light intensity map A(x, y) are respectively: , ; the modulation map M(x, y) is defined as the ratio of the fringe amplitude map to the background light intensity map, i.e. .

7. The method of claim 6, wherein, It also comprises a phase unwrapping step, which recovers the continuous real phase Φ(x, y) through a phase unwrapping algorithm, that is, Φ(x, y) = Unwrap(Φw(x, y)).

8. The method of claim 4, wherein, In the reflection measurement mode, the depth map h(x, y) of an object is proportional to the phase distortion it introduces where K is a proportionality factor obtained by system calibration, and h(x, y) corresponds to the values of the depth map in the reflection measurement.​ 9. The method of claim 4, wherein, The preset logic of defect identification and classification includes that: concave-convex points are locally concave or convex on the depth map obtained by reflection measurement, but have no significant change on the reflection modulation map and the back light modulation map; surface foreign matters are convex on the depth map of reflection measurement, and are significantly dark on the reflection modulation map corresponding thereto; interlayer foreign matters have no abnormality on the depth map of reflection measurement, but have significant disturbance on the phase difference map of back transmission measurement, and present obvious dark area on the back light modulation map; wherein the back transmission measurement refers to that the back transmission light source projects multi-frequency phase shift fringe light through the measured product, the camera collects the transmission fringe image, and performs zero reference plane differential measurement; scratches present high aspect ratio strip features on the depth map or the phase difference map.

Images