An automated sorting apparatus and method for artificial white diamond blanks

Abstract

The application discloses a kind of artificial white diamond blank automatic sorting equipment and method, comprising: feeding mechanism, with conveying belt for conveying white diamond blank to imaging acquisition station;Multi-spectral imaging mechanism, with light source ring assembly, the light source ring assembly sets multiple light source rings for switching scanning combination to be examined white diamond blank irradiation;And set image acquisition unit for multi-direction acquisition white diamond blank optical image under the irradiation of light source ring combination;Spectral analysis mechanism, set microspectrometer for receiving the light signal generated when combined light source ring irradiation white diamond blank;And set data processing unit converts light signal into spectral data;Visual analysis module receives and uses convolutional neural network deep learning and analyzes image and spectral data, obtains the grade data of color and purity of white diamond blank, and outputs comprehensive grade determination according to grading standard. Realize full-process automation operation, improve detection sorting efficiency and quality, mass production.

Description

Technical Field

[0001] This invention relates to the field of diamond rough sorting equipment, and in particular to an automated sorting equipment and method for synthetic white diamond rough. Background Technology

[0002] Lab-grown white diamond roughs are essentially laboratory-grown white diamond blanks, meaning they are uncut and unpolished artificial white diamond prototypes. Their chemical composition, hardness, and optical properties are basically the same as natural diamonds, differing only in their growth process. Lab-grown white diamond roughs are mainly used in the jewelry and industrial sectors.

[0003] In the supply chain of synthetic white diamond roughs, grading is a crucial quality control step. Unprocessed white diamond roughs typically exist in a crystalline state, and their internal structure and surface characteristics require professional testing for accurate evaluation. The inspection and grading of white diamond roughs generally covers dimensions such as shape, color, and clarity. Currently, in the field of synthetic white diamond testing and sorting, traditional sorting equipment largely relies on the standard facet geometry of finished white diamonds, which is unsuitable for the unique shape of rough diamonds. It can only test and sort finished white diamonds, and only within a single dimension. Therefore, the grading of synthetic white diamond roughs mainly relies on manual inspection, i.e., trained graders use testing lights, color charts, and color stones for visual observation and judgment.

[0004] However, the existing technology still has the following problems: manual inspection for grading white diamond blanks is inefficient and cannot meet the needs of large-scale mass production. In particular, in the color and clarity inspection and grading of artificial white diamond blanks, manual inspection faces great challenges due to the complex inclusions, irregular crystals, and large scattering interference of white diamond blanks. The quality of inspection cannot be guaranteed due to human factors. In addition, the existing technology cannot automatically identify the crystal characteristics of white diamond blanks and comprehensively analyze and output the best cutting scheme suggestions. Summary of the Invention

[0005] In order to overcome the shortcomings of the prior art, one of the objectives of this invention is to provide an automated sorting device for artificial white diamond blanks.

[0006] One of the objectives of this invention is achieved by the following technical solution: an automated sorting device for artificial white diamond blanks, comprising:

[0007] The feeding mechanism has a conveyor belt for transporting the white diamond blanks to be sorted along the detection and sorting direction to the imaging acquisition station;

[0008] A multispectral imaging mechanism is provided at the imaging acquisition station. The multispectral imaging mechanism has a light source ring assembly, which has several light source rings arranged in a vertical layer for switching scanning combinations to illuminate the white diamond blank to be inspected; and an image acquisition unit is arranged around the imaging acquisition station for acquiring optical images of the white diamond blank under the combined illumination of the light source rings from multiple directions.

[0009] A spectral analysis unit is set up at the imaging acquisition station. The spectral analysis unit has a miniature spectrometer and receives the light signal generated when the combined light source ring illuminates the white diamond blank through the miniature spectrometer; and a data processing unit is set up to convert the light signal into spectral data.

[0010] The visual analysis module is communicatively connected to the image acquisition unit and the data processing unit. It receives and uses convolutional neural network deep learning to analyze the optical images and spectral data of the rough diamond, obtains the color and clarity grade data of the rough white diamond, and outputs a comprehensive grade judgment according to the grading standard.

[0011] Compared to existing technologies, the advantages of the automated sorting equipment for artificial white diamond blanks of this invention are as follows: This automated sorting equipment, through the configuration of modules such as feeding, multispectral imaging, spectral analysis, and visual analysis, achieves fully automated operation from feeding to sorting. The detection and sorting time for a single white diamond blank can be controlled within 15 seconds, with a daily processing capacity exceeding 3000 pieces, solving the problems of low efficiency and inability to meet large-scale batch production requirements of manual inspection. Simultaneously, by combining multispectral imaging and spectral analysis with deep learning algorithms, the accuracy rate of clarity feature recognition reaches over 98%, overcoming the technical challenge of unstable quality in manual inspection caused by the complex inclusions and irregular crystals of white diamond blanks.

[0012] To overcome the shortcomings of the prior art, the second objective of this invention is to provide an automated sorting method for artificial white diamond blanks.

[0013] The second objective of this invention is achieved by the following technical solution: an automated sorting method for artificial white diamond blanks, comprising the following sorting steps:

[0014] Step 1: Vibration feeding. A batch of white diamond blanks are fed into the vibrating hopper, arranged into a single column, and fed into the conveyor belt. After being adjusted to the standard posture, they are fed into the next station.

[0015] Step 2: Multispectral imaging acquisition. The white diamond blank enters the imaging acquisition station. The central control system sends instructions to control the light source components to switch scanning combinations, illuminating the white diamond blank from multiple directions from top and / or side views. At the same time, the image acquisition unit acquires optical images of the blank under the multi-directional scanning combination.

[0016] Step 3: Spectral data acquisition. A miniature spectrometer is set up to receive the light signal when the light source ring assembly described in Step 2 illuminates the white diamond blank, and the data processing unit converts it into spectral data.

[0017] Step 4: Visual analysis. By setting up a visual analysis module, the image and spectral data from Step 2 and Step 3 are received. Convolutional neural network deep learning and analysis are used to obtain the color and precision grade data of the white diamond rough and output a comprehensive grade judgment according to the preset grading standard.

[0018] Step 5: Automatic sorting and collection. White diamond blanks enter the sorting and collection station. By setting up a sorting and collection mechanism and based on the comprehensive grade judgment result output in Step 4, the corresponding white diamond blanks are collected into the corresponding graded collection containers by air propulsion. The receipts are archived and a test report is generated.

[0019] Compared to existing technologies, the advantages of the automated sorting method for artificial white diamond blanks in this invention are as follows: This embodiment provides a complete and streamlined automated sorting method for white diamond blanks, integrating feeding, multispectral imaging, spectral analysis, deep learning evaluation, automated sorting, and data archiving into a highly efficient and collaborative process. This automated sorting method ensures high efficiency, high quality, consistency, and full-process data traceability in the detection and sorting process, completely replacing the inefficient and non-standardized manual processes in existing technologies. It achieves fully automated operation from feeding to sorting; the detection and sorting time for a single white diamond blank can be controlled within 15 seconds, with a daily processing capacity exceeding 3000 pieces, solving the problems of low efficiency and inability to meet the needs of large-scale batch production in manual inspection. Attached Figure Description

[0020] Figure 1 This is a three-dimensional structural diagram of an automated sorting device for artificial white diamond blanks in a preferred embodiment of the present invention;

[0021] Figure 2 for Figure 1 Enlarged view of point A in the middle;

[0022] Figure 3 This is an enlarged schematic diagram of the attitude adjustment mechanism in a preferred embodiment of the present invention;

[0023] Figure 4 This is an enlarged schematic diagram of the laser profilometer in a preferred embodiment of the present invention;

[0024] Figure 5 This is an enlarged schematic diagram of the multispectral imaging mechanism in a preferred embodiment of the present invention;

[0025] Figure 6 This is a block diagram illustrating the system module control principle in a preferred embodiment of the present invention.

[0026] Figure 7 This is a flowchart of an automated sorting method for artificial white diamond blanks in a preferred embodiment of the present invention;

[0027] Figure 8 A front view of a rough synthetic white diamond containing clarity flaws such as dark inclusions, feather-like fissures, cloud-like inclusions, and pinpoints;

[0028] Figure 9 A top view of the finished artificial white diamond after the rough stone has undergone a brilliant cut;

[0029] Figure 10 A top view of the finished artificial white diamond after the rough stone has been cut in an emerald style.

[0030] In the picture:

[0031] 10. Feeding mechanism; 101. Conveyor belt; 102. Vibrating hopper; 1021. Vibrating conveyor trough; 103. Attitude adjustment mechanism; 1031. Miniature rotating platform; 1032. Identification sensor; 1033. Miniature gripper;

[0032] 20. 3D scanning module; 201. Laser profilometer; 202. 3D point cloud processor;

[0033] 30. Multispectral imaging mechanism; 301. Light source ring assembly; 3011. Standard light source ring; 3012. UV light source ring; 3013. Multi-color temperature adjustable light source ring; 3014. Dark field light source ring; 302. Image acquisition unit; 3021. First industrial camera; 3022. Second industrial camera; 303. Multi-layer ring support; 3031. Ring shelf;

[0034] 40. Spectroscopic analysis unit; 401. Miniature spectrometer; 4011. Fiber optic probe; 402. Data processing unit;

[0035] 50. Visual Analysis Module; 501. Image Preprocessing Unit;

[0036] 60. Sorting and collecting mechanism; 601. Miniature air nozzle; 602. Graded collection container; 603. Diversion trough;

[0037] 70. Central control system;

[0038] 80. Loading station; 81. Shape scanning station; 82. Imaging acquisition station; 83. Sorting and collection station;

[0039] Z, White diamond blank. Detailed Implementation

[0040] The present invention will now be further described in conjunction with the accompanying drawings and specific embodiments. It should be noted that, without conflict, the various embodiments or technical features described below can be arbitrarily combined to form new embodiments.

[0041] like Figures 1 to 6 As shown, an automated sorting device for artificial white diamond blanks (Z) is used for the detection and sorting of artificial white diamond blanks (Z) in the artificial white diamond blank (Z) industry chain. Specifically, the automated sorting device for artificial white diamond blanks includes a feeding mechanism 10, a multispectral imaging mechanism 30, a spectral analysis mechanism 40, and a visual analysis module 50. The feeding mechanism 10 is equipped with a conveyor belt 101, which transports the white diamond blanks (Z) to be sorted along a preset detection and sorting direction to the imaging acquisition station 82.

[0042] At the imaging acquisition station 82, a multispectral imaging mechanism 30 is provided. This multispectral imaging mechanism 30 includes a light source ring assembly 301, which consists of multiple light source rings arranged in vertical layers. This light source ring can switch between different scanning combinations to illuminate the white diamond blank Z according to instructions from the central control system 70. Simultaneously, an image acquisition unit 302 is arranged around the imaging acquisition station 82, and this image acquisition unit 302 acquires optical images of the white diamond blank Z under the illumination of the combined light source rings from multiple directions.

[0043] The spectral analysis unit 40 is also located at the imaging acquisition station 82. Its miniature spectrometer 401 receives the light signal generated when the combined light source ring illuminates the white diamond blank Z, and the data processing unit 402 integrated within the miniature spectrometer 401 converts the light signal into spectral data. The miniature spectrometer can be a high-performance miniature modular fiber optic spectrometer such as the Ocean Insight FLAME-S-VIS-NIR-ES. In terms of image data transmission and control protocols, this embodiment can employ existing technologies such as CoaXPress (CXP-12) ×4, PCIe Gen3 ×8, EtherCAT, and Ethernet (GigE Vision / TCP / IP).

[0044] The visual analysis module 50 is communicatively connected to the image acquisition unit 302 and the data processing unit 402. The visual analysis module 50 employs a convolutional neural network deep learning algorithm to comprehensively analyze the rough diamond's optical images and spectral data. In initial practical applications, the visual analysis module 50 calculates and trains on more than 100 rough diamond optical images, and this number increases later as the sample size increases. This allows it to obtain grade data for the white diamond rough Z in dimensions such as color and clarity, and outputs a comprehensive grade judgment based on preset grading standards (such as GIA, HRD, CIBJO, or national standards). Clarity characteristics mentioned in this embodiment, such as inclusions, fissures, cloud-like inclusions, and surface defects, are... Figure 8 As shown.

[0045] The vision analysis module 50 is integrated into the central control system 70, which uses an industrial-grade embedded industrial computer and is equipped with a touch screen human-machine interface. The vision analysis module 50 can be a Cognex In-Sight 6900.

[0046] The central control system 70 is further equipped with a main control unit, a data storage and management platform, and a self-testing and calibration unit. The main control unit coordinates the working sequence of each module, controls the conveyor belt speed, triggers detection and acquisition, receives and analyzes results, and sends sorting instructions. It also supports one-click switching between various grading modes for different applications, such as jewelry-grade and industrial-grade. The main control unit can be an IPS960-511-PoE from Axiomtek. The data storage and management platform establishes a unique inspection file for each blank, storing content including a 3D model, images from various angles, spectral curves, grading results, and cutting planning suggestions. It also supports data export to the MES system. The self-testing and calibration unit has a power-on self-test function and is equipped with standard colorimetric stones and standard geometric blocks to ensure long-term stability of inspection accuracy.

[0047] Therefore, the automated sorting equipment in this embodiment, through the configuration of modules such as feeding, multispectral imaging, spectral analysis, and visual analysis, achieves fully automated operation from feeding to sorting. The detection and sorting time for a single white diamond blank Z can be controlled within 15 seconds, with a daily processing capacity exceeding 3,000 pieces, solving the problems of low efficiency and inability to meet the needs of large-scale batch production caused by manual inspection. At the same time, by combining multispectral imaging and spectral analysis with deep learning algorithms, the accuracy rate of clarity feature recognition reaches over 98%, overcoming the technical challenge of unstable quality in manual inspection caused by the complex inclusions and irregular crystals of the white diamond blank Z.

[0048] This embodiment further describes the combined illumination design of the light source ring assembly 301. Specifically, the light source ring assembly 301 includes a standard light source ring 3011 (D65 standard daylight, 6500K), a UV light source ring 3012 (containing 365nm long wavelength and 254nm short wavelength), a multi-color temperature adjustable light source ring 3013 (3000K~7000K), and a dark field light source ring 3014. The scanning combination types triggered by the geometric shape of the white diamond blank Z based on the light source ring assembly 301 are as follows:

[0049] The independent standard light source ring 3011 is suitable for rapid initial color screening of small-particle rough diamonds ranging from 0.01 to 0.10 carats, such as industrial-grade crushed diamond rough diamonds and mass-produced lab-grown diamonds. The standard light source ring 3011 serves as the basic assessment for color grading.

[0050] The combination of standard light source ring 3011 and UV light source ring 3012 is used to detect the fluorescence response characteristics of white diamond blank Z near the G / H / I color grade boundary. The fluorescence intensity is divided into four levels: none, weak, medium and strong, to evaluate the "neutralization effect" of fluorescence on color.

[0051] The combination of standard light source ring 3011 and dark field light source ring 3014 is used for rough white diamonds of 0.50 carats or more or high value (such as rough diamonds of 1 carat or more jewelry grade, or rough diamonds that are candidates for high value finished diamonds after cutting). Dark field illumination highlights internal inclusions, fissures and other inclusions, achieving high-contrast clarity detection; then the bright field of standard light source ring 3011 is superimposed to confirm the crystal morphology and color.

[0052] Example of a light source ring combination: For example, using the DISORIC BE-D series dark field vision light source, side lighting allows light to be reflected only from the irregular surface (inclusions, edges, cracks, and other defects) of the white diamond blank Z to the industrial camera. Normal transmitted and reflected light does not enter the industrial camera, thus forming a bright defect signal against a dark background, achieving high-contrast imaging. It is widely used in professional industrial inspection scenarios such as metal surface damage detection, transparent object scratch detection, engraving and laser marking reading, and is especially suitable for observing colorless and transparent samples.

[0053] The all-source ring combination is used for comprehensive and accurate evaluation of high-value or unidentified raw materials, including color baseline, fluorescence level, authenticity screening, and metamerism verification.

[0054] Therefore, the light source ring assembly 301 achieves a balance between detection accuracy and efficiency by optimizing the intelligent switching of the light source ring scanning combination for different application scenarios. In particular, the combination of dark field light source and standard light source bright field solves the problem that the inclusions in white diamond blank Z are difficult to image clearly due to crystal irregularity and large scattering interference. Through flexible multispectral irradiation, the fluorescence reaction and metamerism effect are accurately evaluated, providing a comprehensive and high-quality data foundation for subsequent grading and sorting.

[0055] This embodiment further describes the optimization of the mounting structure of the multispectral imaging mechanism 30. Specifically, as... Figure 1 , 5 As shown, the multispectral imaging mechanism 30 is provided with a multi-layer ring support 303, which is erected above the conveyor belt 101 via a gantry frame. Multiple light source rings are installed on the vertically layered ring support 3031. For example, different light source rings are installed on each layer of the ring support, and multiple identical light source rings are installed on each layer at equal intervals.

[0056] The image acquisition unit 302 includes several first industrial cameras 3021, such as four 12-megapixel cameras, arranged at 90-degree equidistant intervals around the imaging acquisition station 82 along the direction of the annular shelf 3031. These cameras are used to simultaneously acquire images from four side views of the white diamond blank Z and convert the images into multi-channel image streams, which are then fed back to the visual analysis module 50. Optionally, a 20- to 120x zoom microscopic industrial camera (for local high-resolution re-inspection) can be added to the first industrial camera 3021. The model of the first industrial camera 3021 can be Hikvision's MV-CH120-60GM.

[0057] In addition, the image acquisition unit 302 can also be optionally equipped with a second industrial camera 3022, such as a 20-megapixel telecentric lens camera, mounted on top of the multi-layer ring bracket 303, for acquiring top-view images of the white diamond blank Z, and similarly converting the images into multi-channel image streams to be fed back to the visual analysis module 50. The second industrial camera 3022 model can be the MTL17-11005X-C from Huarui Technology.

[0058] Using the arrangement of the first industrial camera 3021 and the second industrial camera 3022, and taking the data of a 1.02-carat octahedral white diamond rough Z as an example, four inclusions in the octahedral white diamond rough Z were identified:

[0059] 1. Dark mineral inclusion (0.15 mm in diameter, located at the edge of the pavilion, 0.8 mm from the surface).

[0060] 2. Feather-shaped fissure (0.3 mm long, located in the waist area);

[0061] 3. Cloud inclusions (area 0.5mm², near the table surface);

[0062] 4. Needle-like inclusions (0.02 mm in diameter, 0.5 mm below the table surface). Based on the above identification and analysis, the clarity grade can be determined as SI1.

[0063] To achieve precise synchronization, this solution employs a two-level collaborative timing system involving both hardware and software: the first level utilizes the EtherCAT distributed clock to implement global sequence control; the second level uses the quadrature pulse signal from the conveyor belt encoder, with the FPGA capturing position signals in real time to fine-tune the trigger timing.

[0064] When generating multi-frame multispectral image sequences (such as a four-frame multispectral image sequence of D65 / UV / dark field / multi-color temperature), the IPC master controller issues commands via EtherCAT, and the FPGA hardware implements nanosecond-level alignment between the light source switching and the industrial camera's exposure window, with a switching delay of no more than 10μs. Image transmission uses the CoaXPress interface to meet high bandwidth requirements, and synchronization triggering uses an EtherCAT link. Its technical principle is to convert the optical signals acquired by the camera into digital multi-channel (such as red, green, blue, and near-infrared bands) image data streams.

[0065] Therefore, the multispectral imaging mechanism 30, by setting up a multi-layered annular shelf 3031, optimizes the arrangement of the first industrial camera 3021 and / or the second industrial camera 3022 of the image acquisition unit 302 for simultaneous multi-angle, multispectral image acquisition. Combined with a high-precision synchronous triggering mechanism, it obtains high-fidelity optical image information of the white diamond blank from all directions, avoiding feature omissions caused by a single viewpoint. The second industrial camera 3022 adopts a telecentric lens camera, eliminating perspective errors and providing the deep learning model with dimensionally accurate and detail-rich input data, significantly improving the subsequent analysis capability of crystal morphology and the spatial position of inclusions.

[0066] like Figure 5 As shown, the miniature spectrometer 401 provided in this embodiment employs an optical fiber probe 4011, and receives the light signal transmitted or reflected after the light source ring assembly 301 illuminates the white diamond blank Z. The data processing unit 402 receives and converts this light signal into an electrical signal, forming spectral data R(λ). The data processing unit 402 further converts the spectral data into L under D65 standard light source and 2° illumination conditions. a b and L c h color space data. This data is then matched against the built-in standard colorimetric stone spectral library to automatically determine the 23 color grades of the white diamond rough, from D to Z.

[0067] The Miniature Spectrometer 401 is fiber-coupled and covers wavelengths from 350nm to 1000nm (ultraviolet to near-infrared), with a spectral resolution better than 1nm.

[0068] Simultaneously, by analyzing characteristic absorption peaks in the spectral data, such as the absorption peak at 415 nm in natural Type Ia diamonds or specific luminescence peaks in CVD-grown diamonds, this embodiment provides an automated sorting device that can also automatically identify the type of rough diamond (such as Type Ia, IIa, IIb, etc.). Taking the data of a 1.02-carat octahedral white diamond rough Z as an example, for instance, its color grade is F (CIELAB coordinates L...). =87.2, a =0.5, b =4.8, color difference ΔE=0.6 compared with F grade standard color stone), judged as diamond type Ia.

[0069] Therefore, the miniature spectrometer 401, by incorporating a fiber optic probe 4011 and combining data comparison analysis, introduces spectral analysis technology into the Z-grading of rough white diamonds. By matching objective chromaticity data with spectral "fingerprints," it completely replaces subjective comparison by the human eye, solving the problem of misjudgment of color grades caused by visual fatigue and environmental differences in manual inspection. Simultaneously, it enables automatic identification of diamond types, providing direct evidence for authenticity screening and value assessment.

[0070] This embodiment further describes the visual analysis processing. Specifically, as... Figure 6 As shown, the visual analysis module 50 includes an image preprocessing unit 501, which is communicatively connected to the image acquisition unit 302 and the data processing unit 402 to receive the original blank optical image and spectral data. The image preprocessing unit 501 uses an FPGA+GPU dual-acceleration engine to perform preprocessing operations such as noise reduction, white balance correction, geometric correction, and multi-view image stitching in real time, generating a standardized analysis image.

[0071] In this embodiment, the visual analysis module 50 uses an existing convolutional neural network (CNN) as its basic network architecture, analyzing and standardizing images through multiple rounds of "convolution-activation-pooling".

[0072] For color grading, image features and spectral chromaticity parameters under the D65 standard light source ring 3011 are integrated to generate color scores, which are then mapped to color levels from D to Z using an ordered regression algorithm. These scores are used to output the color level and confidence score. For example, for color level F, the next best candidate is color E, with a confidence score of 93%.

[0073] For clarity grading, the bright inclusion signal is segmented in the dark field image (the flaw appears as a bright signal on the dark background), its size and area are measured, and the two-dimensional image coordinates are back-projected onto the three-dimensional model for precise positioning. Finally, the clarity grade (such as VS1, SI2) is output according to the GIA standard, which integrates the number, size, and location of inclusions.

[0074] The visual analysis module 50 also has a built-in rule engine, which performs cross-dimensional constraint verification (e.g., restrictions on the highest color grade for large inclusions) and compares the obtained color and clarity grade data according to a unified grading standard to output the final comprehensive grade judgment.

[0075] In this embodiment, the visual analysis module 50 has a built-in rule engine and performs cross-dimensional constraint verification, for example:

[0076] Rule 1: If the clarity is ≥ SI2 and the maximum inclusion size is ≥ 0.3mm, the upper limit of the color grade is limited to G (in this case, the clarity is SI1 and the maximum inclusion size is 0.15mm, so this limit is not triggered and F color is reserved).

[0077] Rule 2: If the volume ratio is <85% (severe defects), the shape grade is marked separately as "irregular fragments" and is not applicable to the standard grade system (in this case, 96.8%, which meets the standard for intact crystals).

[0078] No degradation constraints were triggered in this case, and the F-color SI1 complete octahedron was established.

[0079] Therefore, the visual analysis module 50 solves the problem of the inability of manual identification of complex inclusions (such as cloud-like inclusions) by setting image preprocessing and employing a convolutional neural network (CNN) to automatically extract and identify the microscopic inclusion features and color parameters of the white diamond rough Z. The visual analysis module 50 also introduces a rule engine for decision fusion, simulating the comprehensive evaluation logic of expert graders, making the final grade determination more scientific, consistent, and reliable.

[0080] This embodiment further describes the external scanning design of the white diamond blank Z. Specifically, as follows... Figure 1 , 4 As shown in Figure 6, a shape scanning station 81 is added before the imaging acquisition station 82, and a three-dimensional scanning module 20 is installed above this station. The three-dimensional scanning module 20 includes a laser profilometer 201 emitting a 405nm blue-violet laser and a three-dimensional point cloud processor 202. The laser profilometer 201 is positioned directly above the shape scanning station 81 and the conveyor belt 101, and is used to scan the surface of the blank from multiple angles, and to receive reflected shape scanning data through its built-in high-speed CCD camera. The laser profilometer 201 can be a Keyence LJ-V7020.

[0081] The high-speed CCD camera transmits data to the 3D point cloud processor via an industrial-grade high-speed digital interface (such as the Camera Link interface). The 3D point cloud processor 202 receives this data, fuses it using a point cloud registration algorithm to generate a 3D point cloud model, and calculates the 3D geometric data of the blank based on this model. The 3D geometric data specifically includes the maximum length, width, and height, volume, surface area, convex hull volume of the white diamond blank Z, and crystal types such as octahedron and rhombic dodecahedron. The 3D point cloud processor 202 can be a Keyence KEYENCE or XG-X series model.

[0082] The three-dimensional geometric data is fed back to the multispectral imaging mechanism 30 as a basis for triggering the adaptive light source scanning combination; it is also fed back to the visual analysis module 50. The three-dimensional geometric data assists in hierarchical decision-making, which can be considered from the aspects of crystal integrity, size distribution, volume, and weight.

[0083] Taking the data of a 1.02-carat octahedral white diamond blank Z as an example, the octahedral white diamond blank Z was three-dimensionally scanned using a high-speed CCD camera. For example, the three-dimensional dimensions were 6.52×6.48×6.55mm, the volume was 198.5mm³, the bulge volume was 205.1mm³, the crystal face integrity was good, and the volume ratio was 96.8%.

[0084] More specifically, the visual analysis module 50 receives three-dimensional geometric data and integrates bright-field images (image features) from five angles (top + four sides) under the standard light source ring 3011 with the three-dimensional point cloud model obtained from the three-dimensional scanning module 20. This data is used to identify defects such as growth lines, crystal edge features, and pits on the surface of the white diamond blank Z crystal, and to score the crystal integrity. Simultaneously, by combining UV fluorescence images, spectral curves, and color changes under multiple color temperatures, and through three paths—observing fluorescence patterns, spectral fingerprints, and color changes—and a weighted voting mechanism, automatic screening of counterfeit / synthetic products (authenticity screening results) is achieved.

[0085] Finally, the visual analysis module 50 integrates and analyzes the aforementioned color grade, clarity grade, three-dimensional geometric data, and authenticity screening results to output a comprehensive grade judgment; and calls the cutting optimization algorithm to generate cutting scheme suggestions, including recommended cut shape, expected yield, and optimal cutting direction, based on the maximum inscribed geometry fitting and defect space annotation.

[0086] To be precise, authenticity can be verified simultaneously through three routes:

[0087] Route 1: Observe the fluorescence pattern: Diamonds from different sources exhibit different fluorescence patterns under ultraviolet light. Natural diamonds typically fluoresce blue, HPHT-grown diamonds show a cubic octahedral pattern, CVD-grown diamonds show a layered stripe pattern, and moissanite and cubic zirconia have their own specific fluorescence reactions. Use a convolutional neural network to identify these differences.

[0088] Route Two: Examining Spectral "Fingerprints": Different materials exhibit different characteristic absorption peaks. Natural Type Ia diamonds have a distinct absorption peak at 415 nm, CVD-grown diamonds have characteristic emission peaks at 467.6 nm, 503.2 nm, and 524.3 nm, and HPHT-grown diamonds have characteristic peaks at 575 nm and 637 nm for their NV color centers. The spectrometer automatically detects these "fingerprint" peaks and compares them with a standard library.

[0089] Route 3: Observe color changes: Under different color temperature light sources, the color changes of natural diamonds follow a fixed pattern. If the color of a certain "diamond" deviates beyond the normal range, it may be a non-diamond material with metamerism.

[0090] In summary, when a synthetic diamond (Type Ia) exhibits a 98.2% confidence level, weak blue ultraviolet fluorescence, a distinct absorption peak at 415 nm, and a normal metamerism index, it can be determined to be a synthetic product.

[0091] Taking the cutting scheme suggestion for an octahedral diamond rough Z as an example, the following geometric input is provided:

[0092] 1. Alignment of principal directions: Find the principal axis of maximum inertia of the point cloud, rotate the white diamond blank to the standard direction, align the major axis with the coordinate system, and make the principal axis direction coincide with the diagonal of the octahedron;

[0093] 2. Fitting the largest inscribed geometry: Search for the largest regular geometry that can be accommodated inside the point cloud (e.g., an inscribed cylinder corresponds to a circular bright shape, and an inscribed cuboid corresponds to a stepped / emerald shape). The inscribed cylinder has a diameter of 5.8 mm and a height of 4.1 mm; the inscribed cuboid has a size of 5.6 × 3.2 × 4.0 mm.

[0094] 3. Defect spatial annotation: Map the three-dimensional coordinates of the inclusions output by the clarity test to the point cloud model, and annotate the shortest distance of each defect from the fitting body inline surface. The cloud-like inclusions in the platform area are only 0.6mm away from the center of the fitting circular cut platform, and should be avoided in the cutting plan or arranged at the edge of the crown after drilling.

[0095] Then, under the fusion analysis of the visual analysis module 50, the cutting optimization algorithm is invoked to obtain a circular bright cut scheme and an emerald cut scheme (the circular bright cut and the emerald cut scheme are well-known cutting methods in this technical field; the difference is that in this embodiment, the optimal cutting scheme is obtained by fusing the data from the above-mentioned three-dimensional scanning, multispectral imaging, and spectral analysis through the visual analysis module 50). Wherein:

[0096] Circular bright cut scheme, such as Figure 9 As shown, here is a schematic diagram of the finished product after cutting:

[0097] Using a fitted inscribed cylinder (5.8 mm in diameter, 4.1 mm in height);

[0098] Standard proportions for a round, bright pavilion: tabletop diameter ≈ 57% × waistline diameter, crown height ≈ 14.6%, pavilion depth ≈ 43.1%;

[0099] The algorithm searches for the optimal placement of the cylinder, calculates the maximum diameter of the finished product that can be cut, and overlays it with the inclusion distribution map to ensure that none of the four inclusions are within the envelope of the final cylinder (especially avoiding the cloud-like inclusions 0.6mm below the center of the table).

[0100] Emerald-cut schemes, such as Figure 10 As shown, here is a schematic diagram of the finished product after cutting:

[0101] Using a fitted inscribed cuboid (5.6×3.2×4.0mm), the typical aspect ratio is 1.40:1;

[0102] The standard proportions for emerald-cut finished products: the length-to-width ratio is usually between 1.30 and 1.50;

[0103] Estimated weight of finished product = 5.6 × 3.2 × 1.6 × 0.00175 (diamond specific gravity coefficient) ≈ 0.050g = 0.25ct (the yield is low because the usable space in the height direction of the cuboid is limited).

[0104] Comparison of yield rates between the two cutting methods:

[0105] Round brilliant crystal: 0.48 ct, 47.1%, high yield, good crystal symmetry, cloud-like inclusions close to the table, requiring precise control of the cutting direction.

[0106] Emerald type: 0.25 ct, 24.5%, suitable crystal aspect ratio, yield is too low, not recommended.

[0107] Finally, the visual analytics module outputs the final recommendation:

[0108] Recommended cut: Round brilliant;

[0109] Expected yield: 47.1%;

[0110] Optimal cutting direction: Cut along the octahedral cleavage plane {111}, avoiding the cloud-like inclusions below the center of the table;

[0111] Based on the above calculation results, the recommended profile, expected yield, optimal cutting direction, and inclusion avoidance coordinates are output for direct use in subsequent cutting processes.

[0112] Therefore, this embodiment integrates multi-dimensional Z-simultaneous detection and comprehensive analysis of white diamond roughs, considering shape, color, clarity, and authenticity, on the same automated sorting equipment. By acquiring three-dimensional geometric data, it not only provides a basis for selecting spectral scanning combinations, but more importantly, it solves the problem that existing technologies cannot automatically identify crystal features and output cutting scheme suggestions. Furthermore, the visual analysis module 50 can automatically calculate the yield, recommend cutting shapes, and accurately avoid inclusions, providing quantifiable optimal solutions for subsequent cutting processes, greatly improving the value conversion efficiency from rough diamonds to finished diamonds.

[0113] This embodiment further describes the vibration feeding and attitude adjustment design of the white diamond blank Z. Specifically, as follows... Figure 1 As shown, a loading station 80 is provided before the imaging acquisition station 82, and the aforementioned loading mechanism 10 is located at this loading station 80. The loading mechanism 10 also includes a vibrating hopper 102 and an attitude adjustment mechanism 103. The vibrating hopper 102 has a spiral vibrating conveying trough 1021 on its inner side. The surface of the trough is covered with a flexible anti-damage layer and has a guide channel, so that the white diamond blanks Z are naturally arranged into a single row during the conveying process. The discharge end of the guide channel is connected to the feed end of the conveyor belt 101. The conveyor belt 101 is a servo motor driven annular belt, on which flexible carrier seats for placing the white diamond blanks Z to be sorted are provided at equal intervals.

[0114] like Figure 1 , 3 As shown, a posture adjustment mechanism 103 is installed above the feeding end of the conveyor belt 101 at the feeding station 80. The posture adjustment mechanism 103 has a micro rotating platform 1031 that can rotate horizontally. Multiple micro grippers 1033 that can rotate 360° along the three axes XYZ are distributed along the edge of the platform (e.g., in a triangular shape).

[0115] A recognition sensor 1032 is located at the center of the bottom of the miniature rotating platform 1031. When the sensor detects that the current Z-position of the white diamond blank does not conform to the optimal placement posture / unqualified posture (such as the main crystal face facing the detection optical path), it feeds back an unqualified signal to the gripper controller, which controls the miniature gripper 1033 to descend, pick up, and follow the rotation of the miniature rotating platform 1031 to correct the placement posture of the white diamond blank Z. The recognition sensor 1032 can be an ARU8010 from Huayi Intelligent Manufacturing.

[0116] The speed of conveyor belt 101 is precisely synchronized with the detection time of each station through "encoder feedback + position trigger sensor + unified clock source". It can adopt "movement-stop-movement" stepping or uniform speed continuous trigger acquisition mode, thereby achieving precise synchronization between conveying and detection acquisition actions.

[0117] Therefore, by integrating a vibration feeding and attitude adjustment structure at the loading station 80 of the equipment, automatic, orderly, and standardized feeding of irregularly shaped white diamond blanks Z is achieved. Furthermore, the flow guidance of the vibrating hopper 102 and the active attitude adjustment of the micro-grippers 1033 ensure that each white diamond blank Z entering the inspection area has a uniform orientation, avoiding inconsistencies in optical inspection results caused by arbitrary placement. This provides a fundamental guarantee for the accuracy and repeatability of subsequent 3D scanning and multispectral imaging.

[0118] This embodiment further describes the Z-classification collection design for the tested white diamond blanks. Specifically, as follows: Figure 1 , 2 As shown, a sorting and collection station 83 is provided after the imaging acquisition station 82, on which a sorting and collection mechanism 60 is installed. This mechanism includes a miniature air nozzle 601 and a graded collection container 602. The miniature air nozzle 601 is installed on one side of the conveyor belt 101, and its diameter (0.5-2mm) is selected according to the Z size of the white diamond blank. Multiple graded collection containers 602 are arranged on the opposite side of the conveyor belt 101 where the miniature air nozzles 601 are installed. Each container has a diversion groove 603 at its inlet to guide the blank.

[0119] The central control system 70 stores a grade mapping table in its memory. When a white diamond blank Z that has completed inspection arrives at the sorting station via conveyor belt 101, the position sensor and encoder signals trigger the central control system 70. The system looks up the table based on the comprehensive grade determination result of the blank to determine the corresponding grade collection container 602, and immediately drives the miniature air nozzle 601 corresponding to the container's position. The miniature air nozzle 601 blows the white diamond blank Z across the conveyor belt 101 with precisely controlled airflow, guides it through the diversion trough 603, and smoothly falls into the grade collection container 602 corresponding to the grade.

[0120] The number of graded collection containers 602 can be flexibly configured according to the number of grades in the grading standard, for example, 8 to 16 collection containers.

[0121] Therefore, by setting up a sorting and collection station 83 and a sorting and collection mechanism 60 at the rear of the equipment, and by using a micro-nozzle 601 to quickly and gently pneumatically sort the corresponding white diamond blanks Z into the corresponding graded collection containers 602 based on airflow injection, the pneumatic method is simple in structure and has a fast response speed, avoiding scratches or contamination to the white diamond blanks Z that may be caused by mechanical contact. Through real-time decision-making and driving by the central control system 70, it is ensured that each blank can be accurately sorted into the designated collection container according to its unique comprehensive grade judgment result, thus completing the closed loop of automated detection and sorting throughout the entire process.

[0122] like Figure 7 As shown, this embodiment provides an automated sorting method for artificial white diamond blanks, including the following steps:

[0123] Step 1: Vibration feeding. The operator puts a batch of white diamond blanks Z into the vibrating hopper 102. They are arranged into a single queue through the spiral vibration groove and enter the conveyor belt 101. Then, they are clamped by the miniature gripper 1033 of the attitude adjustment mechanism 103 and rotated to the standard attitude before being sent to the next station.

[0124] Step 2: Multispectral imaging acquisition. The white diamond blank Z enters the imaging acquisition station 82. The central control system 70 sends a command to control the light source ring assembly 301 to switch scanning combinations, illuminating the blank from multiple directions from top and / or side views. At the same time, the image acquisition unit 302 acquires optical images of the blank under the multi-directional scanning combination.

[0125] Step 3: Spectral data acquisition. The micro spectrometer 401 receives the light signal when the light source ring assembly 301 illuminates the white diamond blank Z in step 2, obtains the spectral curve of the reflected / projected 350-1000nm band, and sets the data processing unit 402 to convert and process the light signal into spectral data.

[0126] Step 4: Visual analysis. The visual analysis module 50 receives the rough optical images and spectral data from Step 2 and Step 3, uses convolutional neural network deep learning and analysis to obtain the color and clarity grade data of the white diamond rough Z, and outputs a comprehensive grade judgment according to the preset grading standard.

[0127] Step 5: Automatic sorting and collection. White diamond blanks Z enter the sorting and collection station 83. Based on the comprehensive grade determination result output in Step 4, the sorting and collection mechanism 60 collects the corresponding white diamond blanks Z into the corresponding grade collection container 602 via pneumatic propulsion. Simultaneously, the system stores all test data (images, spectra, grading results, etc.) in the database and exports it to the enterprise's MES or quality traceability system via a network interface (such as Gigabit Ethernet / TCP / IP), generating standardized PDF / JSON format test reports for quality traceability or production management. The entire process is fully automated; operators only need to periodically replenish raw materials and empty the sorted blank containers.

[0128] Therefore, this embodiment provides a complete and streamlined automated sorting method for white diamond blanks Z, integrating feeding, multispectral imaging, spectral analysis, deep learning evaluation, automated sorting, and data archiving into a highly efficient and collaborative process. This automated sorting method ensures high efficiency, high quality, consistency, and full-process data traceability in the inspection and sorting process, completely replacing the inefficient and non-standardized manual processes in existing technologies. It achieves fully automated operation from feeding to sorting; the inspection and sorting time for a single white diamond blank Z can be controlled within 15 seconds, with a daily processing capacity exceeding 3000 pieces, solving the problems of low efficiency and inability to meet the needs of large-scale batch production in manual inspection.

[0129] Preferably, before step 2 (multispectral imaging acquisition) of the above sorting method, step 1.1, three-dimensional morphological scanning, can be added. Specifically, the conveyor belt 101 first sends the white diamond blank Z into the shape scanning station 81. The laser profilometer 201 at this station emits blue-violet lasers to scan the surface of the blank from multiple angles, and the built-in high-speed CCD camera receives the reflected shape scanning data. The three-dimensional point cloud processor 202 receives this data and generates a three-dimensional point cloud model through a point cloud registration algorithm, thereby calculating the three-dimensional geometric data (length, width, height, volume, crystal type, etc.) of the white diamond blank Z. This three-dimensional geometric data serves as the basis for triggering the scanning combination of the light source components in step 2, and is also fed back to the visual analysis module 50.

[0130] The visual analysis module 50 integrates and analyzes the three-dimensional geometric data with the subsequently obtained color and clarity grade data by introducing a rule engine, outputs a comprehensive grade judgment according to the grading standard, and generates cutting scheme suggestions.

[0131] Furthermore, the method may also include a weight measurement step: for example, a corresponding weighing sensor or precision electronic balance can be installed on the conveyor belt 101 for accurate weighing, and the weight data can be synchronously entered into the central control system 70 as another basis for comprehensive grade determination. Thus, simultaneous detection and comprehensive analysis of multi-dimensional data on appearance, color, clarity, and weight are achieved on the same equipment, ultimately providing a more complete grading test report.

[0132] Therefore, by performing a three-dimensional morphological scan before the automatic sorting process, the subsequent light source selection and core grading processes can be adaptively optimized based on the geometric features of the blank. In particular, by introducing three-dimensional geometric data, the automated sorting method of this embodiment generates corresponding cutting scheme suggestions in the final result, which users can then refer to for cutting and processing white diamond blanks. This solves the technical problem in the prior art where manual sorting or sorting equipment cannot automatically identify crystal features and output cutting scheme suggestions.

[0133] The above embodiments are merely preferred embodiments of the present invention and should not be construed as limiting the scope of protection of the present invention. Any non-substantial changes and substitutions made by those skilled in the art based on the present invention shall fall within the scope of protection claimed by the present invention.

Claims

1. An automated sorting device for artificial white diamond blanks, characterized in that, include: The feeding mechanism has a conveyor belt for transporting the white diamond blanks to be sorted along the detection and sorting direction to the imaging acquisition station; A multispectral imaging mechanism is provided at the imaging acquisition station. The multispectral imaging mechanism has a light source ring assembly, which has several light source rings arranged in a vertical layer for switching scanning combinations to illuminate the white diamond blank to be inspected; and an image acquisition unit is arranged around the imaging acquisition station for acquiring optical images of the white diamond blank under the combined illumination of the light source rings from multiple directions. A spectral analysis unit is set up at the imaging acquisition station. The spectral analysis unit has a miniature spectrometer and receives the light signal generated when the combined light source ring illuminates the white diamond blank through the miniature spectrometer; and a data processing unit is set up to convert the light signal into spectral data. The visual analysis module is communicatively connected to the image acquisition unit and the data processing unit. It receives and uses convolutional neural network deep learning to analyze the optical images and spectral data of the rough diamond, obtains the color and clarity grade data of the rough white diamond, and outputs a comprehensive grade judgment according to the grading standard.

2. The automated sorting equipment for artificial white diamond blanks as described in claim 1, characterized in that, The light source ring assembly includes a standard light source ring, a UV light source ring, a multi-color temperature adjustable light source ring, and a dark field light source ring. The scanning combination types triggered by the light source ring assembly based on the geometric shape of the white diamond blank are: independent standard light source ring, combination of standard light source ring and UV light source ring, combination of standard light source ring and dark field light source ring, and combination of all light source rings.

3. The automated sorting equipment for artificial white diamond blanks as described in claim 1, characterized in that, The multispectral imaging mechanism also has a multi-layer ring support, which is arranged above the conveyor belt, and several light source rings are respectively arranged on the vertically layered ring frame of the multi-layer ring support. The image acquisition unit has several first industrial cameras, which are arranged around the imaging acquisition station at 90° equidistant intervals along the direction of the annular shelf. They are used to simultaneously acquire optical images of the white diamond blank from four side views. The first industrial cameras are connected in communication and convert the acquired optical images of the blank into multi-channel image streams before feeding them back to the vision analysis module. And / or, the image acquisition unit also has a second industrial camera, which is set on top of the multi-layer ring bracket, for acquiring optical images of the white diamond blank from a top-down view; the second industrial camera is communicatively connected and converts the acquired optical images of the blank into a multi-channel image stream and feeds them back to the vision analysis module.

4. The automated sorting equipment for artificial white diamond blanks as described in claim 1, characterized in that, The miniature spectrometer is equipped with an optical fiber probe to receive the light signal transmitted or reflected when the light source ring assembly illuminates the white diamond blank; the data processing unit receives and converts the light signal into spectral data, which is then matched and compared with a standard library to determine the color grade and type of the white diamond blank.

5. The automated sorting equipment for artificial white diamond blanks as described in claim 1, characterized in that, The visual analysis module has an image preprocessing unit, which is communicatively connected to the image acquisition unit and the data processing unit, and receives and processes the blank optical image and spectral data respectively to generate a standardized analysis image. The visual analysis module uses convolutional neural network deep learning and standardized image analysis to obtain color and clarity grade data for rough white diamonds. The visual analysis module also has a built-in rule engine, which executes cross-dimensional constraints to compare the obtained grade data with the grading standards and output a comprehensive grade judgment for the rough white diamonds.

6. The automated sorting equipment for artificial white diamond blanks as described in claim 5, characterized in that, A shape scanning station is set up before the imaging acquisition station. A three-dimensional scanning module is set up above the conveyor belt at the shape scanning station. The three-dimensional scanning module has a laser profilometer and a three-dimensional point cloud processor. The laser profilometer is set directly above the shape scanning station and emits blue-violet laser for scanning the surface of the white diamond blank from multiple angles. The laser profilometer also has a built-in high-speed CCD camera for receiving the shape scanning data reflected when scanning the white diamond blank. The three-dimensional point cloud processor receives and fuses the shape scanning data into a three-dimensional point cloud model through a point cloud registration algorithm, which is used to calculate the three-dimensional geometric data of the white diamond blank. The three-dimensional point cloud processor communicates with and feeds back the three-dimensional geometric data to the multispectral imaging mechanism to trigger the light source ring component to switch phase-adaptive scanning combinations. The three-dimensional point cloud processor communicates with and feeds back the three-dimensional geometric data to the visual analysis module. After data preprocessing and crystal feature recognition, the visual analysis module performs fusion analysis on the color and clarity grade data of the white diamond rough and outputs a comprehensive grade judgment and generates cutting scheme suggestions according to the grading standard.

7. The automated sorting equipment for artificial white diamond blanks as described in claim 1, characterized in that, A loading station is provided before the imaging acquisition station, and the loading mechanism is located at the loading station. The feeding mechanism also has a vibrating hopper and a posture adjustment mechanism. The inner side of the vibrating hopper is provided with a spiral vibrating conveyor trough for arranging and conveying white diamond blanks. The discharge end of the vibrating conveyor trough is correspondingly provided with the feed end of the conveyor belt. The loading station is equipped with an attitude adjustment mechanism located above the feed end of the conveyor belt. The attitude adjustment mechanism has a micro rotating platform. The micro rotating platform is equipped with an identification sensor and a micro gripper. After the micro gripper receives the identification sensor's detection of the current unqualified attitude signal of the white diamond blank, it descends to grip and rotates to correct the placement attitude of the white diamond blank.

8. The automated sorting equipment for artificial white diamond blanks as described in claim 1, characterized in that, Following the imaging acquisition station, a sorting and collection station is also provided. The sorting and collection station is equipped with a sorting and collection mechanism, which has a micro air nozzle and multiple graded collection containers. The micro air nozzle is located on one side of the conveyor belt, and the multiple graded collection containers are arranged on the opposite side of the conveyor belt where the micro air nozzle is located. The sorting driver drives the corresponding micro air nozzle to push the current white diamond blank to the graded collection container of the corresponding level.

9. An automated sorting method for artificial white diamond blanks, characterized in that, The following sorting steps are included: Step 1: Vibration feeding. A batch of white diamond blanks are fed into the vibrating hopper, arranged into a single column, and fed into the conveyor belt. After being adjusted to the standard posture, they are fed into the next station. Step 2: Multispectral imaging acquisition. The white diamond blank enters the imaging acquisition station. The central control system sends instructions to control the light source components to switch scanning combinations, illuminating the white diamond blank from multiple directions from top and / or side views. At the same time, the image acquisition unit acquires optical images of the blank under the multi-directional scanning combination. Step 3: Spectral data acquisition. A miniature spectrometer is set up to receive the light signal when the light source ring assembly described in Step 2 illuminates the white diamond blank, and the data processing unit converts it into spectral data. Step 4: Visual analysis. By setting up a visual analysis module, the image and spectral data from Step 2 and Step 3 are received. Convolutional neural network deep learning and analysis are used to obtain the color and precision grade data of the white diamond rough and output a comprehensive grade judgment according to the preset grading standard. Step 5: Automatic sorting and collection. White diamond blanks enter the sorting and collection station. By setting up a sorting and collection mechanism and based on the comprehensive grade judgment result output in Step 4, the corresponding white diamond blanks are collected into the corresponding graded collection containers by air propulsion. And to archive receipts and generate test reports.

10. The automated sorting method for artificial white diamond blanks as described in claim 9, characterized in that, Before step 2, the process may include step 1.1, three-dimensional morphological scanning. A conveyor belt delivers the white diamond blank to the shape scanning station, where a blue-violet laser emitted by a laser profilometer scans the surface of the white diamond blank from multiple angles. Then, a built-in high-speed CCD camera receives the reflected shape scanning data, and a three-dimensional point cloud processor receives this data and generates a three-dimensional point cloud model to calculate the three-dimensional geometric data of the white diamond blank, which serves as the trigger for the scanning combination of the light source components in step 2. This three-dimensional geometric data is then fed back to the visual analysis module, which performs a fusion analysis on the color and clarity grade data of the white diamond blank and outputs a comprehensive grade judgment and cutting scheme suggestions according to the grading standards.

Images