System and method for gemstone identification

Abstract

This invention provides a system and computer implementation method for matching gemstones with records in a gemstone database. In one embodiment, the computer implementation method includes the following steps: a) acquiring a source image of a gemstone; b) receiving input for determining a corresponding gemstone from the gemstone database; c) retrieving a set of reference data; d) extracting gemstone features from the source image; e) aligning the gemstone features with the set of reference data and mapping the gemstone features to a standard plane through homography transformation, thereby obtaining a transformed source image of the source gemstone; f) obtaining a transformed reference image based on the set of reference data; g) calculating a similarity value by matching the transformed source image with the transformed reference image; and h) determining whether the gemstone to be matched matches the corresponding gemstone based on the similarity value.

Description

Technical Field

[0001] This invention relates to systems and methods for gem identification. Background Technology

[0002] Gemstones are frequently targeted by fraud and counterfeiting. Successful counterfeits can cause significant economic losses to consumers and businesses. Therefore, developing a reliable gemstone identification system is of great importance for commercial applications.

[0003] Currently, gem identification typically involves trained gemologists using specialized equipment such as microscopes to perform detailed examinations. However, this process is often time-consuming, costly, and not readily available, especially in retail environments. Therefore, there is a need for a portable and efficient gem identification method that can be easily used by non-professionals and deployed in retail settings where a continuous internet connection may not be available.

[0004] In the context of this invention, an edge device refers to a device that possesses both local data processing capabilities and cloud connectivity. Edge devices enable more efficient data processing and provide real-time processing capabilities even when cloud connectivity is limited. The edge capabilities of the gem identification system of this invention enable it to process video stream data locally, thereby reducing its reliance on high bandwidth and cloud computing resources. This makes the system suitable for retail stores and other scenarios where cloud connectivity may not be readily available.

[0005] Cut gemstones possess a unique combination of physical features that can be used for identification. Traditional methods involve examination by gemologists to identify variations in size, table facets, angles, inclusions, and other physical features such as artificial markings, typically created by laser engraving.

[0006] In jewelry, the most easily observed area of ​​a gemstone is typically its crown, or simply the top. Gemstones are usually set with the tip pointing downwards in a setting or prong setting to present the most visually appealing perspective—the crown view. Other areas of the gemstone, such as the girdle and pavilion, are usually obscured by the setting or prong setting. The crown is often chosen as the display area in jewelry because it contains the table, the largest facet of the gemstone, through which most light enters and exits. Therefore, the crown view is the most readily available and provides a wealth of physical characteristics for observation.

[0007] Given that each gemstone possesses unique physical characteristics, and combined with recent advancements in computer vision and machine learning technologies, gemstone identification through image matching has become possible. Image matching typically involves comparing unique gemstone features captured in an image with a database of known real gemstones to determine their authenticity. Image matching can also be extended to matching video streams with images, as is commonly used in facial recognition systems.

[0008] US Patent Publication No. 11,037,282 B2 discloses a method for identifying clarity characteristics and inclusions using machine learning algorithms. Of particular note is the ability of machine learning algorithms to extract useful information, such as edge features, in the form of edges. This invention draws upon the ability of machine learning algorithms to extract edge features and their robustness and performance under varying data conditions, applying them to the field of gem identification.

[0009] U.S. Patent Publication No. 11,232,553 B2 further illustrates a jewelry security analysis system, method, and computer program product that allows a user to authenticate jewelry by scanning at least the girdle portion of a gemstone. This invention expands upon the concept of identifying unique features of a gemstone by capturing specific areas, requiring only the capture of distinctive markings such as certificate numbers and brand labels, along with their surrounding area. Summary of the Invention

[0010] This invention provides a computer-implemented method for matching gemstones with records in a gemstone database. In one embodiment, the computer-implemented method includes the following steps: a) acquiring a source image of a gemstone to be matched; b) receiving input for determining a corresponding gemstone from the gemstone database; c) retrieving a set of reference data related to the corresponding gemstone; d) extracting gemstone features from the source image, the gemstone features including inclusion masks, edge features, and marker points; e) aligning the gemstone features with the set of reference data and mapping the gemstone features to a standard plane through homography transformation, thereby obtaining a transformed source image; f) obtaining a transformed reference image based on the set of reference data; g) calculating a similarity value by matching the transformed source image with the transformed reference image; and h) determining whether the gemstone to be matched matches the corresponding gemstone based on the similarity value.

[0011] The present invention further provides a non-transitory computer-readable storage medium including pre-stored instructions, characterized in that: when executed by one or more processors, the one or more processors perform operations including the method described in the present invention.

[0012] The present invention also provides a system for matching gemstones with records in a gemstone database using the computer-implemented method described herein. In one embodiment, the system includes: a) one or more devices for acquiring source images, each of the one or more devices including: i) a first processor; ii) an interface for receiving the input in step (b); iii) a first communication module; iv) a memory unit storing executable instructions, which, when executed, cause the first processor to transmit the input in step (b) through the first communication module to further execute step (c) of the computer-implemented method; b) a server including a second processor, the gemstone database, a second communication module, and a memory unit storing executable instructions, which, when executed, cause the second processor to execute steps (c) to (h) of the computer-implemented method; the input is transmitted from the first communication module to the second communication module to execute step (c); the result of step (h) is transmitted from the second communication module to the first communication module. Attached Figure Description

[0013] The embodiments of the present invention are described in conjunction with system schematic diagrams and module drawings. Further embodiments are described using user operation flowcharts and user interface wireframe diagrams.

[0014] Figure 1 This is a flowchart of the identification method, which begins when a user initiates the identification process. The flowchart illustrates the sequence of steps in the gem identification process, including image acquisition, feature extraction, database matching, and the final identification result.

[0015] Figure 2 This is a user operation flowchart that details the sequence of operations a user performs within the system. The flowchart includes multiple decision nodes, such as activating authentication mode, starting the device, updating to the latest version, processing images in the video, and matching the gemstone ID code with the cloud database to complete the authentication.

[0016] Figure 3 This is a schematic diagram of the modules inside the hardware casing of the system, showing the power connection between each module and the rechargeable battery device.

[0017] Figure 4 illustrates the process of matching features extracted from a reference image and an image of the gemstone to be examined using an artificial intelligence (AI) edge detection algorithm. The figure shows a comparison between the features extracted from the image to be examined and the reference image, demonstrating the alignment achieved through the AI ​​matching algorithm and the successful identification between the two images.

[0018] Figure 5 highlights the ability of the algorithm of this invention to extract the edges of gemstones, where minute differences in facet edges due to polishing can be observed. This figure further demonstrates the algorithm's ability to extract the edges of inclusions.

[0019] Figure 6 illustrates the detection of mismatches between the image of the gemstone to be examined and the reference image in the database. This figure demonstrates the process of extracting features from the reference and examination images using an artificial intelligence (AI) edge detection algorithm, showing a mismatch between the features extracted from the two images.

[0020] Figure 7 illustrates the performance of the method of the present invention on reference images of different clarity grades. The table lists the accuracy of the method for each clarity grade (including VVS1, VVS2, VS1, VS2, SI1, and SI2). Examples of corresponding reference images are shown next to the accuracy values, demonstrating the system's ability to maintain high accuracy across different gemstone clarity grades. Detailed Implementation

[0021] This invention provides a computer-implemented method for identifying a specific gemstone using inscribed or natural markings. In one embodiment, the method includes the following steps: (a) acquiring a gemstone identification code and verifying its existence in a database; (b) detecting the presence of the gemstone in a newly acquired image; (c) extracting key features from the image, including edge features and inclusion markings, using inscribed, polished, or natural markings as reference points; (d) retrieving one or more reference images of potential matching gemstones from the gemstone database based on the extracted features; (e) calculating a similarity value by comparing the features of the source image and the reference images; and (f) identifying the specific gemstone from the reference images based on the calculated similarity value.

[0022] The present invention also provides a non-transitory computer-readable storage medium including pre-stored instructions, characterized in that: when executed by one or more processors, the one or more processors perform operations including the computer implementation method of the present invention.

[0023] The present invention further provides a system for identifying a specific gemstone using the computer-implemented method described herein. In one embodiment, the system includes: (a) one or more devices for acquiring source images, each of the one or more devices including: (i) a first processor; (ii) a memory unit storing executable instructions, which, when executed, cause the first processor to perform step (a) of the computer-implemented method; (iii) a first communication module; and (b) a server including a second processor, the gemstone database, a second communication module, and a memory unit storing executable instructions, which, when executed, cause the second processor to perform steps (b) to (e) of the computer-implemented method; wherein the inscription or natural mark is detected and transmitted from the first communication module to the second communication module; and the specific gemstone is identified and transmitted from the second communication module to the first communication module.

[0024] This invention provides a system and computer implementation method for matching gemstones with records in a gemstone database. In one embodiment, the computer implementation method includes the following steps: a) acquiring a source image of a gemstone; b) receiving input for determining a corresponding gemstone from the gemstone database; c) retrieving a set of reference data; d) extracting gemstone features from the source image; e) aligning the gemstone features with the set of reference data and mapping the gemstone features to a standard plane using homography transformation, thereby obtaining a transformed source image of the source gemstone; f) obtaining a transformed reference image based on the set of reference data; g) calculating a similarity value by matching the transformed source image with the transformed reference image; and h) determining whether the gemstone to be matched matches the corresponding gemstone based on the similarity value.

[0025] In one embodiment, the homography transformation includes mapping the inclusion mask and the edge features using the marker points as reference points.

[0026] In one embodiment, the reference data includes a reference image.

[0027] In one embodiment, the source image is a still image from a camera or an image obtained from video streaming.

[0028] In one embodiment, one or more of steps (d), (e), or (f) are performed using an artificial intelligence algorithm.

[0029] In one embodiment, the artificial intelligence algorithm is trained using labeled markers and gem images combined with data augmentation.

[0030] In one embodiment, the source image differs from the set of reference data in one or more of the following characteristics: shooting angle, distortion, lighting conditions, image quality, and noise.

[0031] In one embodiment, one or more of steps (d), (e), or (f) include filtering image noise.

[0032] In one embodiment, the method further includes determining whether the source image contains a gem to be matched before performing step (b).

[0033] In one embodiment, the set of reference data includes one or more reference images or transformed reference images.

[0034] In one embodiment, the edge feature includes facets or waist edges.

[0035] In one embodiment, the inclusion mask includes internal inclusions or surface defects.

[0036] In one embodiment, the marker is an inscribed mark or a natural mark.

[0037] In one embodiment, the alignment in step (e) is performed based on the marker points.

[0038] In one embodiment, the source image is a top view of the gemstone to be matched.

[0039] The present invention further provides a non-transitory computer-readable storage medium including pre-stored instructions, characterized in that: when executed by one or more processors, the one or more processors perform operations including the method described in the present invention.

[0040] The present invention also provides a system for matching gemstones with records in a gemstone database using the computer-implemented method described herein. In one embodiment, the system includes: a) one or more devices for acquiring source images, each of the one or more devices including: i) a first processor; ii) an interface for receiving the input in step (b); iii) a first communication module; iv) a memory unit storing executable instructions, which, when executed, cause the first processor to transmit the input in step (b) through the first communication module to further execute step (c) of the computer-implemented method; b) a server including a second processor, the gemstone database, a second communication module, and a memory unit storing executable instructions, which, when executed, cause the second processor to execute steps (c) to (h) of the computer-implemented method; the input is transmitted from the first communication module to the second communication module to execute step (c); the result of step (h) is transmitted from the second communication module to the first communication module.

[0041] The invention can be better understood by referring to the following examples; however, those skilled in the art should understand that the specific examples detailed are for illustrative purposes only and are not intended to limit the invention described herein. The scope of the invention is defined by the following claims. Various references and publications are cited throughout this application. The full disclosure of these references and publications is incorporated herein by reference to provide a more comprehensive description of the prior art in the field to which this invention pertains. It should be noted that the transitional term "comprising" is synonymous with "including," "containing," or "characterized by," and is inclusive or open-ended, not excluding additional, unlisted elements or method steps.

[0042] This invention aims to address the need for a portable, easy-to-use gem identification system with edge computing capabilities. More specifically, regarding identification capabilities, the system can capture and process video streams of gemstones, automatically detect their physical characteristics and markings (such as inscriptions and certificate numbers), and perform identification and display the results using video and image matching technology. Regarding ease of use, a typical shop assistant can easily move the system without tools, and the identification process is automated. Regarding edge computing capabilities, the system can meet the computational and storage requirements of video and image matching technology.

[0043] Figure 1 provides a flowchart of the identification method; Figure 2 depicts the overall user operation flow of the identification system. Figure 3 shows the edge computing device components used to capture gemstone features for identification purposes.

[0044] This invention provides an automatic identification system and method for moving objects in gemstone identification, designed for real-time analysis under various conditions. Referring to Figure 1, the process begins in step 501 by capturing a video stream of the gemstone to be examined using a high-resolution video capture device and inputting its ID code. The gemstone can be a loose stone or a gemstone set in jewelry. A frame is sampled from the video, referred to as the "image to be examined." In step 502, the system checks the existence of the gemstone identification code, which is manually entered by the operator in machine-readable text. This input allows the system to retrieve a corresponding set of reference images from a gemstone database for identification. If the ID code does not exist, the system confirms in step 503 that it does not match a database record. If the ID code exists, the system continues to acquire new images from the video.

[0045] Referring to Figure 2, the user flowchart outlines the interaction process between the user and the gem identification system. The process begins at step 403 where the user determines whether to activate "Identification Mode." If Identification Mode is not enabled, the system will operate in "Gem Viewer Mode" as shown at step 400. If the user is new or the system requires an update, the process will guide the user to complete "Device Activation" (for new users) at step 401, or "Version Update" at step 402, to ensure the system is up-to-date. After completing these steps, the user can continue the identification process. Once the user enters Identification Mode at step 403, the system prompts them to manually enter the gem ID code at step 404. Upon receiving the gem ID, the system initiates the identification procedure at step 405. The system first attempts to identify the gem based on the entered ID code; if the ID code exists in the database, the process continues; however, if the ID code does not exist, the system will identify it as "The entered gem ID code does not match the database record" at step 406.

[0046] If the ID code is valid, the system processes the captured video into a gemstone image. Referring to Figure 1, after acquiring the necessary video frames, the system will continue to check if the gemstone has been clearly detected. Referring to step 505, the system uses an artificial intelligence recognition algorithm to determine whether the gemstone in the image has been clearly identified. If the gemstone is not clearly detected, the process will check at step 504 whether the detection time limit has been exceeded. If the time limit has been exceeded, the system will terminate the identification attempt due to timeout, as shown in step 407. If the time limit has not been exceeded, the system will continue to analyze the video until the gemstone is clearly detected.

[0047] Once the gemstone is clearly detected, the system processes the video and compares the generated image to be examined with a reference image stored in a cloud database, as shown in 408. Referring to 506a, the system applies an artificial intelligence edge detection algorithm to extract edge features from the gemstone. This process extracts key structural features from the gemstone image, such as facets and girdle. These extracted edge features are crucial for ensuring that the structural details of the gemstone are captured for further comparison.

[0048] In 506b, the system applies an AI-powered inclusion segmentation algorithm to identify and isolate internal or surface inclusions in the gemstone. Inclusion features, combined with edge features extracted from 506a, represent the gemstone's overall structure and natural characteristics. These features are crucial for accurate comparison, as inclusions are often unique to gemstones and provide additional distinguishing features.

[0049] In 506c, the system employs an AI-powered marker detection algorithm to extract key reference points (i.e., markers) from the gemstone. These markers ensure that edge features and inclusion features are correctly aligned with the reference image, enabling accurate comparison. This alignment eliminates distortions caused by different angles or perspectives during image capture, which is essential for subsequent feature transformations.

[0050] In 507, after aligning the gemstone features, the system performs inclusion mask and edge feature transformations using homography transformation via marker point alignment. This transformation maps the inclusion mask and edge features to a standard plane using faceted marker points as reference points. Marker points on the gemstone are defined as key faceted intersections and edge features specific to that gemstone. The system calculates a homography matrix that defines the precise mapping required to align these detected features on the gemstone with the corresponding reference data. This process involves solving transformation parameters to minimize distance differences between corresponding marker points on the gemstone and the reference gemstone. By transforming the features to this normalized plane, the system ensures precise alignment of edge features with the edge features of the reference gemstone. This alignment compensates for scaling, rotation, and perspective variations that may occur due to different viewing angles or imaging conditions. The homography transformation effectively normalizes gemstone features, enabling accurate and direct comparison during the matching process. Referring to Figures 4 and 6, after this transformation is completed, the system displays the combined edge features of the reference and gemstones. These illustrations demonstrate how features are superimposed on a normalized plane, highlighting the alignment effect achieved through homography transformation. This transformation ensures that any differences found between the detected gem image and the corresponding reference image are inherent to the gem itself, rather than human error caused by misalignment or distortion.

[0051] The transformed inclusions and edge features are then matched in 508a. The system uses an artificial intelligence matching algorithm to compare the inclusions and edge features of the gemstone under test with reference data. The matching process analyzes the similarity between the combined features of the gemstone under test and the reference gemstone. As shown in Figure 4, closely aligned features indicate extremely strong similarity; while Figure 6 highlights mismatched features, indicating significant differences between the gemstones.

[0052] Following the matching process, the system calculates a similarity value in 508b. This value quantifies the degree of similarity between the gemstone being examined and the reference image based on the transformed edge and inclusion features. A higher score indicates a high match, while a lower score indicates a mismatch. Figure 4 shows a high similarity value due to good feature alignment, while Figure 6 shows a low similarity value due to poor feature alignment.

[0053] In step 508c, the system checks whether the calculated similarity value reaches a preset threshold. If the score reaches or exceeds the threshold, the gemstone is identified as a match for a database record, as shown in step 409. If the score is below the threshold (as shown in Figure 6), the system determines that the gemstone does not match the database record, resulting in identification failure. Operators can adjust this threshold according to their tolerance for potential forgery to meet specific security requirements.

[0054] The system's performance has been validated through extensive testing. Table 1 shows the performance test results of the method of this invention using real diamond sample pairs. This table compares the results of this method with a version excluding marker extraction and image correction. The comparison includes positive results, negative results, and the final accuracy rate, demonstrating a significant improvement in accuracy when using the complete method of this invention. Referring to Table 1, the method achieved an accuracy rate of 99.45% in real diamond sample pair testing, with 398 positive results and 2 negative results. This demonstrates the system's effectiveness in distinguishing gemstones. In contrast, when key steps such as marker extraction and image correction are omitted, the system accuracy drops to 0%, confirming the crucial role these processes play in ensuring high accuracy.

[0055]

[0056] Table 2 illustrates the performance of the method of this invention in searching for real diamond samples from a reference image dataset. The table shows the number of true positives, true negatives, test positives, and test negatives. This method demonstrates its ability to successfully match the image to be examined with the correct reference image, with no false negatives and only a small number of false positives. Furthermore, as shown in Table 2, a large-scale test involving 8,330 gemstone samples yielded 4 true positives, 2 true negatives, and 0 false negatives, further validating the system's ability to accurately distinguish real diamonds.

[0057]

[0058] Furthermore, the system demonstrated its effectiveness across various clarity grades. As shown in Figure 7, the system consistently maintained high accuracy across a wide range of clarity grades, including VVS1 (98.64%), VVS2 (99.11%), VS1 (99.73%), VS2 (99.65%), SI1 (98.53%), and SI2 (98.78%). These results highlight the system's robustness when handling gemstones with varying degrees of inclusions. The examples in Figure 7 illustrate the system's ability to accurately identify and match diamonds of different clarity grades.

[0059] The invention also relates to a system for implementing the aforementioned identification method. Figure 3 depicts the modules inside the hardware housing of the system. The gem identification system includes a high-resolution digital video capture device 110. The horizontal position of the video capture device is adjustable to find the correct distance to the gem and to achieve better focus for the video stream. The digital video stream can be digitally scaled for easier inspection. The hardware housing 190 contains a processor unit 280, a wireless transceiver module 253, and is powered by a rechargeable battery device 254 also housed within the housing. To provide optimal lighting conditions for capturing gem video, the system integrates horizontal and vertical lighting devices: the horizontal lighting device 100 provides background illumination; the vertical lighting device 101 is supported by a flexible cable fitting, allowing for vertical and rotatable adjustment of the device to provide illumination at different heights and angles. An adjustable worktable is used to place gems and other gems on the jewelry and to capture video. The item to be examined is placed on a soft support on the adjustable worktable, and different points of view (POV) of the gem can be obtained during identification by changing its height and orientation.

[0060] The following section continues the explanation with reference to Figures 2 and 3. When operating the system, the user will start the system via the power switch. The system will start in "Gem Viewer Mode" 401. Subsequently, the user can select to enter "Identification Mode" 403 via the button switch.

[0061] As mentioned earlier, after switching to authentication mode 403, user verification and software version checking will be performed first. For new users, the following procedure is required to activate the system. New users will be directed to "Device Activation" 400. In Device Activation 400, the device activation user interface will display a QR code for the user to scan, and the user's identity will be verified through an external verification system. Once verified, the device will be activated by adding a unique identifier to the database, after which the user can proceed to the next steps. For users who have already been verified and whose systems are already activated, Device Activation 400 will be skipped, and the system will directly proceed to the next step—"Version Update" 402.

[0062] Version update 402 is an automatic process. The wireless transceiver module 253 first transmits the system's software version to the cloud. The cloud server then executes a program to determine if the software is the latest version. If the software is not the latest version, after determining that the system's software is outdated, it will receive the latest version of the software via the wireless transceiver 253. The update will then be automatically installed on the system. If the latest software already exists on the system or there is no network connection, the new software will not be received or installed.

[0063] Continuing from version update 402, the system processor unit 280 (comprising processor module 300 and memory module 301) within the hardware enclosure 190 operates collaboratively, automatically processing the video stream captured by the high-resolution digital video capture device 110. Processing unit 280 executes a real-time workflow loop 404 for detecting the gemstone's physical characteristics and certificate number engraving. The user adjusts the viewing angle of the gemstone and laser engraving by operating the adjustable stage and focus adjustment knob. The user can also adjust the magnification of the video stream on the display 150 as needed.

[0064] Once the physical characteristics of the gemstone and the certificate number engraving are detected, the processed video can be sent either to the cloud or to the local processor unit 280 for video-image matching. If the processed video is to be sent to the cloud, the transmitted data will be encrypted during transmission and decrypted on the cloud server. As described in process 404, an automatic time-limit test will be performed in a loop; if no gemstone and certificate number engraving are detected, the system will enter "timeout" 407, and the display 150 will display the timeout user interface. If the user wishes, they can restart the identification process by pressing the "Start" button displayed on the display 150 interface.

[0065] If the aforementioned artificial intelligence detection and recognition technology successfully detects and identifies the gemstone and its unique mark, the processed video will be sent to the cloud via the wireless transceiver module 253. The cloud server will then perform an automatic video and image matching process 106 according to the method described in this invention, based on the cloud database containing gemstone images.

[0066] To enable the video-to-image matching process 408, a cloud database on a cloud server contains multiple sets of encrypted gemstone images, each set belonging to a specific certified gemstone. Each set of gemstone images can be one or more pictures. This image set includes at least one image used for identification purposes that depicts the gemstone's unique identifying marks or features (or both).

[0067] If the gemstone is determined to be genuine (406) by matching it with gemstone image records in the database, display module 150 will redirect to the "Authentication Passed" page, confirming the gemstone's authenticity. If the gemstone is found to be counterfeit or cannot match any gemstone recorded in the database (405), the display will redirect to "Authentication Failed," indicating that the gemstone does not match the database record, thus completing the automated authentication process. The automated authentication process is crucial to the system's functionality because it improves usability, reduces the need for training personnel without gemological training, further shortens testing time, and reduces human error involved in gemstone authentication. One deployment scenario for this system is a jewelry store operated by sales personnel.

[0068] In addition, the gem identification system includes a battery module 254 for power supply, which enables portability and, together with the wireless transceiver module 253, eliminates the need for wired connections. The compact and portable design of the gem identification system makes it easy to transport and deploy in stores.

[0069] As mentioned above, the system's ability to perform authentication with or without an internet or intranet connection, combined with its automated authentication process and portability, makes it suitable for deployment in store or laboratory environments, and is not limited to these.

Claims

1. A computer-based method for matching gemstones with records in a gemstone database, characterized in that: The method includes the following steps: a. Obtain a source image of the gem to be matched; b. Receive input for determining a corresponding gem from the gem database; c. Retrieve a set of reference data related to the corresponding gemstone; d. Extract gemstone features from the source image, the gemstone features including inclusion masks, edge features, and marker points; e. Align the gemstone features with the set of reference data, and map the gemstone features onto a standard plane through homography transformation to obtain the transformed source image; f. Obtain a transformed reference image based on the aforementioned set of reference data; g. By matching the transformed source image with the transformed reference image, a similarity value is calculated; and h. Determine whether the gemstone to be matched matches the corresponding gemstone based on the similarity value.

2. The method according to claim 1, characterized in that: The homography transformation includes using the marker points as reference points to map the inclusion mask and the edge features.

3. The method according to claim 1, characterized in that: The reference data includes reference images.

4. The method according to claim 1, characterized in that: The source image is a still image from a camera or an image obtained from video stream processing.

5. The method according to claim 1, characterized in that: One or more of steps (d), (e), or (f) are executed using artificial intelligence algorithms.

6. The method according to claim 5, characterized in that: The artificial intelligence algorithm is trained using labeled markers and gem images combined with data augmentation.

7. The method according to claim 1, characterized in that: The source image differs from the set of reference data in one or more of the following features: shooting angle, distortion, lighting conditions, image quality, and noise.

8. The method according to claim 1, characterized in that: One or more of steps (d), (e), or (f) include filtering image noise.

9. The method according to claim 1, characterized in that: The method further includes determining whether the source image contains the gem to be matched before performing step (b).

10. The method according to claim 1, characterized in that: The set of reference data includes one or more reference images or transformed reference images.

11. The method according to claim 1, characterized in that: The edge features include facets or waist edges.

12. The method according to claim 1, characterized in that: The inclusion mask includes internal inclusions or surface defects.

13. The method according to claim 1, characterized in that: The markers are either inscribed marks or natural marks.

14. The method according to claim 1, characterized in that: The alignment in step (e) is performed based on the marker point.

15. The method according to claim 1, characterized in that: The source image is a top view of the gemstone to be matched.

16. A non-transitory computer-readable storage medium including pre-stored instructions, characterized in that: When executed by one or more processors, the one or more processors perform operations including the method of claim 1.

17. A system for matching gemstones with records in a gemstone database using the computer-implemented method of claim 1, comprising: a. One or more devices for acquiring a source image, each of the one or more devices comprising: i. A primary processor; ii. An interface for receiving the input described in step (b); iii. A first communication module; iv. A memory unit containing executable instructions, which, when executed, cause the first processor to transmit the input from step (b) via the first communication module to further execute step (c) of the computer implementation method; b. A server comprising a second processor, the gem database, a second communication module, and a memory unit containing executable instructions, which, when executed, cause the second processor to perform steps (c) to (h) of the computer-implemented method; The characteristic is that the input is transmitted from the first communication module to the second communication module to execute step (c); The result of step (h) is transmitted from the second communication module to the first communication module.

Images