A desktop high-resolution multi-dimensional optical image intelligent analysis method and device

Abstract

The present application relates to a kind of desktop high-resolution multi-dimensional optical image intelligent analysis method and device, belong to optical image analysis technical field.The method includes: through host control imaging equipment and integrated desktop stage, the sample on stage is synchronously imaged, generates original multi-dimensional image set;Through parallax fusion algorithm, the two-dimensional light image mapping of single frame is generated, the three-dimensional topographic features of sample are simulated by gray and color gradient change;Dynamic light image mapping is output to desktop display in real time, and interactive light source simulation engine is loaded;Real-time simulation light effect on dynamic light image mapping;By automatically configuring light source parameter, the visual saliency of target feature is improved;According to the rendering result after optimization, the topological structure features and optical features of sample are extracted, and quantitative analysis report is generated.The whole process automation and intelligentization of optical image from multi-direction multi-view collection to depth analysis and quantitative report generation are realized.

Description

Technical Field

[0001] This invention belongs to the field of optical image analysis technology, specifically relating to a desktop high-resolution multidimensional optical image intelligent analysis method and device. Background Technology

[0002] The demand for quantification of micron-level surface morphology (scratches, corrosion, texture) has surged in fields such as industrial inspection and cultural relic analysis; traditional single-angle imaging cannot meet the requirements for accurate analysis of complex three-dimensional features (curvature, depressions, reflective areas);

[0003] High-precision optical inspection equipment is generally a large vertical system, requiring an independent support base for the stage and an external light source controller, resulting in a bulky device. Traditional dome light sources have a narrow wavelength range (usually 400-700nm) and LED units are group-controlled, making independent tuning impossible. While technologies such as oblique photography and RTI have achieved multi-angle synchronous acquisition, timing misalignment issues still exist in multi-device collaboration.

[0004] Algorithms such as parallax fusion and lighting rendering have made progress in light field image processing, and GPU-accelerated rendering is gradually becoming more widespread. However, high-resolution data processing still relies on offline computing and cannot support real-time interaction.

[0005] Multidimensional optical imaging technology is at a critical stage of transformation from high-precision offline analysis to compact, real-time, and automated systems, but hardware collaboration, algorithm efficiency, and standardization have not yet met the requirements of desktop integrated systems. Summary of the Invention

[0006] To address the aforementioned problems in the existing technology, this invention provides a desktop-based high-resolution multidimensional optical image intelligent analysis method and device. The objective of this invention can be achieved through the following technical solutions:

[0007] A desktop-based intelligent analysis method for high-resolution multidimensional optical images includes:

[0008] S1: The host controls the imaging equipment and the integrated desktop stage to drive the high-resolution optical lens to perform multi-directional and multi-view synchronous image acquisition of the sample on the stage, generating the original multi-dimensional image set.

[0009] S2: Based on the multidimensional image set, a two-dimensional light and shadow mapping map of a single frame is generated by a parallax fusion algorithm. The two-dimensional light and shadow mapping map simulates the three-dimensional morphological features of the sample through grayscale and color gradient changes.

[0010] S3: Output the dynamic light and shadow mapping map to a desktop display screen in real time and load an interactive light source simulation engine;

[0011] S4: Real-time simulation of light effects of different directions, intensities and spectra on the dynamic light mapping according to the interactive light source simulation engine; based on the preset analysis target, the visual saliency of the target feature is improved by automatically configuring the light source parameters;

[0012] S5: According to the optimized rendering result, the topological structure features and optical features of the sample are extracted to generate a quantitative analysis report.

[0013] Specifically, the method of synchronously collecting images in S1 is: positioning the sample by controlling the rotating stage, and collecting RTI images under different incident angles of the dome light source.

[0014] Specifically, the desktop stage in S1 adopts a three-dimensional size optimized desktop mounting structure, which is directly integrated into the main machine operation plane without the need for an independent support base; the light source control and sample positioning are integrated into a single stage body through modular design.

[0015] Further, the light source control is realized by a miniaturized control module integrated in the upper connecting structure of the dome light source; the dome light source is a low-profile, detachable multi-angle LED array dome suspended above the stage, covering the sample area, for providing multi-angle incident light from above the sample; the stage is a flat platform located directly below the dome light source for carrying the sample.

[0016] Preferably, the miniaturized control module includes a distributed drive circuit, a dome integrated heat dissipation structure, a high-speed communication interface, and a synchronous trigger unit.

[0017] The distributed drive circuit uses multiple micro high-integration LED drive chips, which are directly embedded in the structure layer of the dome light source, independently control the LED units in different areas of the dome, and realize precise adjustment of incident angle and brightness.

[0018] The dome integrated heat dissipation structure independently designs a heat conduction path and a micro heat dissipation fan inside the dome light source shell, uses the space of the dome itself for heat dissipation, and ensures stable operation of the light source for a long time under the limitation of the desktop space.

[0019] The high-speed communication interface is directly connected to the desktop host through the built-in high-speed serial bus from the dome light source, receives light source parameter instructions and feeds back state information, and the communication cable is led out through the dome support structure.

[0020] The synchronous trigger unit integrates a hardware trigger signal generator in the dome control module, maintains precise timing synchronization with the imaging device and stage motion controller, and ensures that image acquisition is completed at the moment of lighting different angle light sources.

[0021] Specifically, the LED array of the dome light source in S1 adopts a full-spectrum adjustable unit with a wavelength range of 380-780 nm, and the color temperature and brightness of each LED unit are independently adjusted by a distributed driving circuit of a miniaturized control module to realize multi-spectral layered imaging of the sample surface texture; a synchronous trigger unit generates a hardware-level trigger pulse during image acquisition to ensure that the exposure period of the imaging device is synchronized with the switching of the LED illumination.

[0022] Specifically, the parallax fusion algorithm is realized by the imaging software of the host, which performs multi-scale decomposition processing on the multi-dimensional image set, separates each frame of image into low-frequency contour information and high-frequency texture information; the low-frequency contour information is subjected to parallax correction and three-dimensional topographic contrast fusion, the small-scale texture and medium-scale edge features of the high-frequency texture information are extracted, the image is reconstructed by inverse multi-scale transformation, the fused low-frequency contour information and high-frequency texture information are superimposed to generate a single two-dimensional light and shadow mapping image, and the topological structure consistency of the reconstructed image and the original multi-dimensional image set is evaluated by using the bottleneck distance quantization.

[0023] Specifically, the interactive light source simulation engine is realized based on a GPU-accelerated ray tracing architecture, and specifically includes:

[0024] A light source abstract model library is used to preset the basic light models of parallel light, point light, spotlight and ambient light and their mixed variants;

[0025] A real-time rendering pipeline is used to adopt a hybrid radiosity / shadow volume rendering technology to support dynamic adjustment of light source direction, intensity and spectral parameters;

[0026] A parameter optimization module is used to dynamically adjust the multi-light source superposition weight by analyzing the gray gradient distribution of the target feature area to enhance the visual saliency of the texture topography;

[0027] A physical light verification module is used to internally store a lighting standard parameter library to verify the compliance of the lighting parameters in real time.

[0028] Specifically, the quantitative analysis report includes:

[0029] A three-dimensional topological quantization matrix is used to extract topological structure features, generate a three-dimensional grid density distribution map of the sample surface, and label the curvature radius, height gradient and deformation tolerance threshold of the key area;

[0030] A multi-spectral feature vector set is used to analyze the optical features into a wavelength-reflectivity matrix, which contains the characteristic wavelength reflectivity peak, half-wave width data and deviation coefficient from the standard spectrum library corresponding to the preset analysis target;

[0031] A dynamic visualization report engine is used to generate interactive 3D models through GPU parallel computing. Clicking on a feature area displays local morphological parameters in real time and overlays optical parameter comparison curves under different light source simulation conditions.

[0032] The compliance verification label is used to automatically mark whether the key parameters in the report meet the testing standards based on the built-in standard library of the physical optics verification module.

[0033] A desktop high-resolution multidimensional optical image intelligent analysis device includes:

[0034] The multi-angle image synchronous acquisition module is used to control the imaging equipment and integrated desktop stage through the host, drive the high-resolution optical lens to perform multi-angle and multi-view synchronous image acquisition of the sample on the stage, and generate the original multi-dimensional image set.

[0035] The pseudo-3D light and shadow mapping synthesis module is used to generate a single-frame two-dimensional light and shadow mapping map based on the multi-dimensional image set through a parallax fusion algorithm. The two-dimensional light and shadow mapping map simulates the three-dimensional morphological features of the sample through grayscale and color gradient changes.

[0036] An interactive light source engine loading module is used to output the dynamic light and shadow mapping map to a desktop display screen in real time and load an interactive light source simulation engine.

[0037] The adaptive light source optimization analysis module is used to simulate the lighting effects of different directions, intensities, and spectra in real time on the dynamic light and shadow map according to the interactive light source simulation engine; based on the preset analysis target, it improves the visual salience of the target features by automatically configuring the light source parameters;

[0038] The feature quantization translation module is used to extract the topological and optical features of the sample based on the optimized rendering results and generate a quantization analysis report.

[0039] The desktop high-resolution multidimensional optical image intelligent analysis method and device provided by this invention has several significant advantages. In terms of space utilization, the optimized three-dimensional design and modular integration of its desktop stage allow for efficient layout within limited desktop space, eliminating the need for an independent support base, thus saving laboratory space and improving operational convenience. Regarding image acquisition, the multi-angle image synchronous acquisition module, combined with a dome light source, can accurately acquire sample images from different orientations and perspectives, generating a rich set of original multidimensional images, providing a sufficient data foundation for subsequent analysis.

[0040] The combination of the parallax fusion algorithm and the pseudo-three-dimensional light and shadow mapping synthesis module ingeniously converts the multi-dimensional image set into a single two-dimensional light and shadow mapping image, vividly simulates the three-dimensional topographic features of the sample through the changes in the gray scale and color gradient, and clearly displays the details of the sample. The interactive light source simulation engine and the related modules provide the user with high autonomy and flexibility in simulating different light effects, can automatically optimize the light source parameters according to the preset analysis target, greatly improves the visual saliency of the target features, and facilitates more accurate observation and analysis of the sample.

[0041] The quantitative analysis report deeply interprets the sample from multiple dimensions such as topological structure and optical characteristics, generates three-dimensional topological quantitative matrix, multi-spectral feature vector set and other contents, and provides comprehensive and accurate data support for researchers and related practitioners. The existence of the compliance verification label ensures the reliability and standardization of the analysis results, making the entire analysis process more scientific and rigorous. BRIEF DESCRIPTION OF DRAWINGS

[0042] In order to facilitate the understanding of those skilled in the art, the present application will be further described below with reference to the accompanying drawings.

[0043] Fig. 1 A flowchart of a desktop high-resolution multi-dimensional optical image intelligent analysis method of the present application;

[0044] Fig. 2 A structural block diagram of a desktop high-resolution multi-dimensional optical image intelligent analysis device in the present application;

[0045] Fig. 3 A structural diagram of a desktop high-resolution multi-dimensional optical image intelligent analysis device in the present application. DETAILED DESCRIPTION

[0046] In order to further illustrate the technical means and effects adopted by the present application to achieve the predetermined invention purpose, the specific embodiments, structures, features and effects of the present application will be described in detail below with reference to the accompanying drawings and preferred embodiments.

[0047] Please refer to Figs. 1-3 A desktop high-resolution multi-dimensional optical image intelligent analysis method, comprising:

[0048] S1: driving a high-resolution optical lens to perform multi-directional and multi-angle synchronous image acquisition on a sample on a sample stage through a host computer controlling an imaging device and an integrated desktop sample stage, to generate an original multi-dimensional image set;

[0049] S2: generating a single two-dimensional light and shadow mapping image through a parallax fusion algorithm based on the multi-dimensional image set, the two-dimensional light and shadow mapping image simulating the three-dimensional topographic features of the sample through the changes in the gray scale and color gradient;

[0050] S3: outputting the dynamic light-mapping map to a desktop display in real time and loading an interactive light source simulation engine;

[0051] S4: simulating light effects of different directions, intensities and spectrums on the dynamic light-mapping map in real time according to the interactive light source simulation engine; and improving visual saliency of a target feature by automatically configuring light source parameters based on a preset analysis target;

[0052] S5: extracting topological structure features and optical features of the sample according to the optimized rendering result, and generating a quantitative analysis report.

[0053] Specifically, the method for synchronously collecting images in S1 is as follows: a sample is positioned by controlling a rotating sample stage, and RTI images are collected under light sources of different incident angles by a dome light source.

[0054] In this embodiment, the method for synchronously collecting images not only ensures that images of the sample under different angles and light conditions can be accurately captured, but also improves the pertinence and effectiveness of image collection through accurate positioning of the rotating sample stage. In actual operation, an operator can flexibly adjust the position of the rotating sample stage and the incident angle of the dome light source according to the characteristics of the sample and the analysis requirements, so as to obtain the most required original multi-dimensional image set. Moreover, since the LED array of the dome light source adopts full-spectrum adjustable units, it can realize multi-spectral layering of the sample surface texture, which makes the collected images contain richer sample information, and provides a more solid data foundation for subsequent analysis. At the same time, the synchronous triggering unit ensures that the exposure period of the imaging equipment is synchronized with the switching of the LED lighting, further improving the quality and accuracy of image collection.

[0055] Specifically, the desktop sample stage in S1 adopts a three-dimensional size-optimized desktop mounting structure, is directly integrated into the main machine operation plane, and does not need an independent supporting base; and the light source control and sample positioning are integrated into a single sample stage body through modular design.

[0056] In this embodiment, the stage body is made of high-strength lightweight material and is precisely processed and formed. Its three-dimensional size is strictly optimized through human engineering and equipment layout. Through a standardized quick-release interface, it is directly and stably installed on the top of the main machine shell or a specially designed operation plane, completely abandoning the large independent base required by traditional microscopes or imaging systems, significantly saving valuable desktop space, and simplifying the installation process. The stage integrates the driving circuit of the dome light source and the LED array control unit. Through the multifunctional interface (USB-C or Ethernet) on the side of the stage or wireless connection, it communicates with the host computer, receives host instructions, and accurately adjusts the brightness, color temperature, and specific wavelength of hundreds of independently controllable full-spectrum LED units in the dome light source. Users do not need to connect or manage the light source controller.

[0057] The stage core integrates a high-precision electric rotating platform (usually driven by a stepper motor or a servo motor, combined with a high-resolution encoder). The platform has 360° continuous rotation capability and micron-level repeat positioning accuracy. The rotation control instruction is also issued by the host through the interface integrated in the stage. The operator can intuitively set the rotation angle, speed, or perform multi-point positioning collection through the software interface.

[0058] The rotating platform on the top of the stage provides a universal sample clamp interface that can quickly adapt to samples of various sizes and shapes (from small parts to small fragments of cultural relics). Some models also integrate a miniature XYZ fine-tuning platform for fine focusing or small area positioning of samples.

[0059] This deep modular design integrates all key functional components (rotating mechanism, motor drive, light source control circuit, communication module) compactly in a single, solid stage housing. This not only greatly simplifies system wiring (usually only one cable is needed to connect the host for power and communication), reduces failure points, but also ensures strict mechanical alignment and timing coordination between subsystems (rotation, illumination). At the same time, the integrated design also improves the portability and anti-interference ability of the equipment.

[0060] Further, the light source control is realized through a miniaturized control module integrated in the upper connection structure of the dome light source; the dome light source is a low-profile, detachable multi-angle LED array dome suspended above the stage, covering the sample area, for providing multi-angle incident light from above the sample; the stage is a flat platform located directly below the dome light source for carrying the sample.

[0061] Preferably, the miniaturized control module includes a distributed driving circuit, a dome-integrated heat dissipation structure, a high-speed communication interface, and a synchronous trigger unit.

[0062] The distributed driving circuit adopts multiple micro high-integration LED driving chips, which are directly embedded in the structural layer of the dome light source, independently control the LED units in different areas of the dome, and realize precise adjustment of the incident angle and brightness.

[0063] The dome integrated heat dissipation structure independently designs a heat conduction path and a micro heat dissipation fan inside the dome light source shell, dissipates heat by using the space of the dome itself, and ensures that the light source works stably for a long time under the limitation of the desktop space.

[0064] The high-speed communication interface is directly connected to the desktop host from the dome light source through the built-in high-speed serial bus, receives light source parameter instructions and feeds back state information, and the communication cable is led out through the dome support structure.

[0065] The synchronous triggering unit integrates a hardware trigger signal generator in the dome control module, maintains precise timing synchronization with the imaging device and the motion controller of the object table, and ensures that image acquisition is completed at the moment when the light source at different angles is turned on.

[0066] In this embodiment, the distributed driving circuit discards the traditional centralized large light source controller and adopts multiple micro high-integration LED driving chips. These chips are directly embedded in the structural layer inside the dome light source and distributed near the corresponding LED unit group. Each micro driving chip is responsible for controlling the LED unit group in a specific sector or annular area of the dome light source. This distributed architecture greatly shortens the driving signal transmission path, reduces signal attenuation and interference. Receiving instructions from the host (through the high-speed communication interface), each driving chip can independently and in parallel perform high-precision pulse width modulation or constant current driving on the LED units in its jurisdiction area, realizing: independently turning on / dimming the LED group at a specific orientation and height, simulating incident light in any direction; fine control of the brightness of each or each group of LEDs; independently adjusting the intensity of different color temperature or specific wavelength LEDs to achieve complex spectral combination.

[0067] The dome integrated heat dissipation structure adopts an embedded heat conduction path, and the special-shaped copper heat pipe or high thermal conductivity metal substrate is precisely embedded inside the dome shell. These heat conduction elements are in close contact with the heat dissipation pads of the driving chips and the LED substrate, forming an efficient "point (heat source) - line (heat pipe) - surface (shell)" heat conduction network.

[0068] An ultra-thin, low-noise micro fan is integrated at a key position in the non-optical path of the dome structure. The airflow generated by the fan flows through the pre-designed air duct inside the dome shell, accelerating the exhaust of hot air from the pre-designed vent of the dome. The shell heat dissipation fully utilizes the large area of the inner / outer surface of the dome shell as a passive heat sink.

[0069] High-speed communication interface adopts built-in high-speed serial bus protocol (USB 3.x Gen1 / 2, GigabitEthernet,). The communication physical interface (USB-C, RJ45) is directly integrated on the circuit board of the micro control module. The communication cable is led out through the support arm of the dome light source or the special cable management channel and connected to the host. Only a single cable is required to complete power supply and high-speed data communication, minimizing exposed cables and keeping the desktop clean.

[0070] The synchronous trigger unit integrates a dedicated hardware trigger signal generator in the micro control module.

[0071] Receive precise position / angle feedback signals from the host or stage motion controller. Receive exposure trigger signals from the imaging device. According to the preset synchronization logic and extremely short fixed delay, real-time generation of accurate trigger pulse signals directly drives the corresponding distributed LED drive chip, ensuring that the target LED group has reached the set brightness and remains stable within the precise time window of the camera sensor exposure. At the same time, this unit also coordinates the stable state signal after the stage rotation is in place, realizing the hard synchronization of "position-illumination-imaging".

[0072] Specifically, the LED array of the dome light source in S1 uses a full-spectrum adjustable unit with a wavelength range of 380nm-780nm, and the color temperature and brightness of each LED unit are independently adjusted by the distributed drive circuit of the micro control module to realize multi-spectral layered visualization of the sample surface texture; the synchronous trigger unit generates a hardware-level trigger pulse during image acquisition to ensure synchronization of the exposure period of the imaging device and the LED illumination switching.

[0073] Specifically, the parallax fusion algorithm is realized through the imaging software of the host, which performs multi-scale decomposition processing on the multi-dimensional image set, separates each image into low-frequency contour information and high-frequency texture information; the low-frequency contour information is corrected for parallax and fused with three-dimensional topographic contrast, the small-scale texture and medium-scale edge features of the high-frequency texture information are extracted, the image is reconstructed through inverse multi-scale transformation, the fused low-frequency contour information and high-frequency texture information are superimposed to generate a single two-dimensional light and shadow mapping image, and the consistency of the reconstructed image and the original multi-dimensional image set in topological structure is evaluated using bottleneck distance quantization.

[0074] Specifically, the interactive light source simulation engine is realized based on GPU-accelerated ray tracing architecture, specifically including:

[0075] Light source abstract model library for presetting basic lighting models of parallel light, point light, spotlight, and ambient light and their mixed variants;

[0076] A real-time rendering pipeline supports dynamic adjustment of light source direction, intensity and spectral parameters using hybrid radiosity / Shadow Volume rendering technology.

[0077] A parameter optimization module dynamically adjusts the weight of multi-light source superposition by analyzing the gray gradient distribution of the target feature area to enhance the visual saliency of the texture topography.

[0078] A physical light testing module has a built-in lighting standard parameter library to verify the compliance of lighting parameters in real time.

[0079] Specifically, the quantitative analysis report includes:

[0080] A three-dimensional topological quantization matrix is used to extract topological structure features, generate a three-dimensional grid density distribution map of the sample surface, and label the curvature radius, height gradient and deformation tolerance threshold of the key area.

[0081] A multi-spectral feature vector set is used to analyze optical features into a wavelength-reflectivity matrix, which includes the characteristic wavelength reflectivity peak, half-wave width data and deviation coefficient from the standard spectrum library corresponding to the preset analysis target.

[0082] A dynamic visualization report engine is used to generate an interactive three-dimensional model through GPU parallel computing, display local topography parameters in real time by clicking on feature areas, and superimpose optical parameter comparison curves under different light source simulation conditions.

[0083] A compliance verification label is used to automatically label whether the key parameters in the report meet the detection standards according to the built-in standard library of the physical light testing module.

[0084] A desktop high-resolution multi-dimensional optical image intelligent analysis device includes:

[0085] A multi-angle image synchronous acquisition module is used to drive high-resolution optical lenses to perform multi-directional and multi-view synchronous image acquisition of samples on the sample stage through host control of imaging equipment and integrated desktop sample stages, generating a set of original multi-dimensional images.

[0086] A pseudo-three-dimensional light mapping synthesis module is used to generate a two-dimensional light mapping map for a single frame based on the multi-dimensional image set through a parallax fusion algorithm, which simulates the three-dimensional topographic features of the sample through gray and color gradient changes.

[0087] An interactive light source engine loading module is used to output the dynamic light mapping map to a desktop display screen in real time and load an interactive light source simulation engine.

[0088] An adaptive light source optimization analysis module is configured to simulate the lighting effects of different directions, intensities and spectrums in real time on the dynamic light mapping map according to the interactive light source simulation engine; and to improve the visual saliency of the target features by automatically configuring the light source parameters based on the preset analysis target;

[0089] A feature quantification interpretation module is configured to extract the topological structure features and optical features of the samples according to the optimized rendering results, and to generate a quantification analysis report.

[0090] The above description is only the preferred embodiment of the present application, and does not limit the present application in any form. Although the present application has been disclosed as the above preferred embodiment, it is not intended to limit the present application. Any person skilled in the art can make some changes or modifications to the above disclosed technical content to obtain equivalent embodiments with equivalent changes, without departing from the technical solution of the present application. Any simple modification, equivalent change and modification of the above embodiments based on the technical essence of the present application are still within the scope of the technical solution of the present application.

Claims

1. A desktop high-resolution multi-dimensional optical image intelligent analysis method, characterized in that, The application relates to a method for generating a dynamic light and shadow mapping image of a sample. S1: through host control, an imaging device and an integrated desktop stage drive a high-resolution optical lens to collect multi-directional and multi-view synchronous images of a sample on the stage, and generate an original multi-dimensional image set; S2: based on the multi-dimensional image set, a single-frame two-dimensional light and shadow mapping image is generated through a parallax fusion algorithm, the two-dimensional light and shadow mapping image simulates three-dimensional topographic features of the sample through changes in gray scale and color gradient; The parallax fusion algorithm is realized through imaging software of the host, the imaging software performs multi-scale decomposition processing on the multi-dimensional image set, separates each frame of image into low-frequency contour information and high-frequency texture information, performs parallax correction and three-dimensional topographic contrast fusion on the low-frequency contour information, extracts small-scale texture and medium-scale edge features of the high-frequency texture information, reconstructs the image through inverse multi-scale transformation, superimposes the fused low-frequency contour information and high-frequency texture information to generate a single-frame two-dimensional light and shadow mapping image, and adopts a bottleneck distance quantization to evaluate the topological structure consistency of the reconstructed image and the original multi-dimensional image set; S3: the dynamic light and shadow mapping image is output to a desktop display screen in real time, and an interactive light source simulation engine is loaded; S4: different directions, intensities and spectra of light illumination effects are simulated on the dynamic light and shadow mapping image in real time according to the interactive light source simulation engine; and the visual saliency of a target feature is improved through automatic configuration of light source parameters based on a preset analysis target; S5: according to the optimized rendering result, topological structure features and optical features of the sample are extracted to generate a quantitative analysis report.

2. The method of claim 1, wherein, The method for synchronous image collection in S1 is that sample positioning is performed through control of a rotating stage, and RTI images are collected under different incident angles of a dome light source.

3. The method of claim 1, wherein, The desktop stage adopts a three-dimensionally optimized desktop mounting structure and is directly integrated on a host operation plane without a separate supporting base; light source control and sample positioning are integrated in a single stage body through modular design.

4. The method of claim 3, wherein, The light source control is realized through a miniaturized control module integrated in an upper connecting structure of the dome light source; the dome light source is a low-profile, detachable multi-angle LED array dome suspended above the stage and covering a sample area, and is used for providing multi-angle incident light from above the sample; and the stage is a plane platform located directly below the dome light source and used for carrying the sample.

5. The method of claim 4, wherein, The miniaturized control module comprises a distributed driving circuit, a dome integrated heat dissipation structure, a high-speed communication interface and a synchronous trigger unit; The distributed driving circuit adopts a plurality of miniaturized high-integration LED driving chips, is directly embedded in a structure layer of the dome light source, independently controls LED units in different regions in the dome, and realizes accurate adjustment of incident angles and brightness; The dome integrated heat dissipation structure independently designs a heat conduction path and a miniaturized heat dissipation fan inside a dome light source shell, dissipates heat by using the space of the dome itself, and ensures long-time stable work of the light source under the limitation of the desktop space; The high-speed communication interface is directly connected to the desktop host through a built-in high-speed serial bus from the dome light source, receives light source parameter instructions and feeds back state information, and a communication cable is led out through a dome supporting structure. The synchronous trigger unit integrates a hardware trigger signal generator in the dome control module, which keeps precise timing synchronization with the imaging device and the stage motion controller, ensuring that the image acquisition is completed at the moment when the light source at different angles is turned on.

6. The method of claim 5, wherein, The LED array of the dome light source in S1 uses a full-spectrum adjustable unit with a wavelength range of 380-780 nm, and the color temperature and brightness of each LED unit are independently adjusted through the distributed drive circuit of the miniaturized control module to realize multi-spectral layered imaging of the sample surface texture. The synchronous trigger unit generates a hardware-level trigger pulse during image acquisition to ensure synchronization between the exposure period of the imaging device and the switching of the LED illumination.

7. The method of claim 1, wherein, The interactive light source simulation engine is implemented based on a GPU-accelerated ray tracing architecture, specifically including: A light source abstraction model library is used to pre-set the basic light models of parallel light, point light, spotlight, and ambient light and their mixed variants; A real-time rendering pipeline uses hybrid radiosity / shadow volume rendering technology to support dynamic adjustment of light source direction, intensity, and spectral parameters; A parameter optimization module dynamically adjusts the weight of multiple light sources to enhance the visual saliency of texture features by analyzing the gray gradient distribution of the target feature area; A physical light verification module has an internal lighting standard parameter library to verify the compliance of lighting parameters in real time.

8. The method of claim 1, wherein, The quantitative analysis report includes: A three-dimensional topological quantization matrix is used to extract topological structure features, generate a three-dimensional grid density distribution map of the sample surface, and label the curvature radius, height gradient, and deformation tolerance threshold of key areas; A multi-spectral feature vector set is used to analyze optical features into a wavelength-reflectivity matrix, which contains the characteristic wavelength reflectivity peak, half-wave width data, and deviation coefficient from the standard spectrum library corresponding to the preset analysis target; A dynamic visualization report engine is used to generate an interactive three-dimensional model through GPU parallel computing, display local topography parameters in real time by clicking on feature areas, and superimpose optical parameter comparison curves under different light source simulation conditions; A compliance verification label is used to automatically label whether the key parameters in the report meet the detection standards according to the built-in standard library of the physical light verification module.

9. A desktop high-resolution multi-dimensional optical image intelligent analysis device for performing the method according to any one of claims 1 to 8, characterized in that, It includes: A multi-angle image synchronous acquisition module is used to drive high-resolution optical lenses to perform multi-directional and multi-view synchronous image acquisition of samples on the stage through host control of imaging devices and integrated desktop stages, generating a set of original multi-dimensional images; A pseudo-three-dimensional light and shadow mapping synthesis module is used to generate a two-dimensional light and shadow mapping image based on the multi-dimensional image set through a parallax fusion algorithm, which simulates the three-dimensional topography features of the sample through gray and color gradient changes; The parallax fusion algorithm is realized by imaging software of the host, the imaging software performs multi-scale decomposition processing on the multi-dimensional image set, separates each frame of image into low-frequency contour information and high-frequency texture information, performs parallax correction and three-dimensional topographic contrast fusion on the low-frequency contour information, extracts small-scale texture and medium-scale edge features of the high-frequency texture information, reconstructs the image through inverse multi-scale transformation, superimposes the fused low-frequency contour information and high-frequency texture information to generate a single two-dimensional light and shadow mapping diagram, and adopts a bottleneck distance to quantitatively evaluate the topological structure consistency of the reconstructed image and the original multi-dimensional image set; The interactive light source engine loading module is used for real-time output of the dynamic light and shadow mapping diagram to a desktop display screen and loading of an interactive light source simulation engine; The adaptive light source optimization analysis module is used for real-time simulation of illumination effects of different directions, intensities and spectra on the dynamic light and shadow mapping diagram according to the interactive light source simulation engine; and based on a preset analysis target, visual saliency of a target feature is improved by automatically configuring light source parameters. The feature quantitative interpretation module is used for extraction of topological structure features and optical features of a sample according to an optimized rendering result and generation of a quantitative analysis report.

Images