Semantic information data dimensionality reduction methods, data dimensionality reduction systems and controllers

Abstract

This application discloses a data dimensionality reduction method, system, and controller for semantic information. The method includes: acquiring image data sent by an image acquisition device; processing the image data to obtain a semantic image; extracting semantic information from the semantic image; determining point clusters in a vehicle coordinate system based on the semantic information; determining the factor loading matrix corresponding to the point clusters in the vehicle coordinate system using factor analysis; and using the factor loading matrix corresponding to the point clusters in the vehicle coordinate system as the descriptor of the point clusters in the vehicle coordinate system to complete data dimensionality reduction. This application uses the factor loading matrix corresponding to the point clusters in the vehicle coordinate system as the descriptor of the point clusters in the vehicle coordinate system, which can reduce storage space consumption while ensuring usability.

Description

Technical Field

[0001] This application relates to the field of autonomous driving technology, specifically to a data dimensionality reduction method, data dimensionality reduction system, and controller for semantic information. Background Technology

[0002] Localization technology is a crucial component of autonomous driving solutions. Multi-sensor fusion localization schemes are increasingly becoming mainstream; for example, in scenarios like highways and ramps, a fusion of vision and high-precision maps is typically used. Currently, existing technologies either employ feature point methods for localization or utilize semantic maps generated from dense semantic point clouds. Feature point methods require extracting locations with drastic grayscale changes in an image as corner points, using identical feature points from consecutive images to estimate the vehicle's position. However, feature point methods are significantly affected by lighting conditions, exhibiting substantial errors in varying scenarios such as daytime, nighttime, or rainy weather, limiting their applicability. Furthermore, localization using semantic maps generated from dense semantic point clouds suffers from excessive storage consumption. Therefore, existing technologies either have limited applicability or inefficient storage requirements. Summary of the Invention

[0003] The purpose of this application is to provide a data dimensionality reduction method, data dimensionality reduction system, and controller for semantic information, in order to solve the problems of limited application scope or excessive storage consumption of existing positioning technologies.

[0004] To achieve the above objectives, the first aspect of this application provides a method for dimensionality reduction of semantic information, applied to a controller, the controller communicating with an image acquisition device, comprising:

[0005] Acquire image data sent by the image acquisition device;

[0006] Image data is processed to obtain semantic images;

[0007] Extract semantic information from semantic images;

[0008] Determine the point cluster in the vehicle coordinate system based on semantic information;

[0009] The factor loading matrix corresponding to the point cluster in the vehicle coordinate system is determined by factor analysis.

[0010] The factor loading matrix corresponding to the point cluster in the vehicle coordinate system is used as the descriptor of the point cluster in the vehicle coordinate system to complete the data dimensionality reduction.

[0011] In this embodiment of the application, extracting semantic information from a semantic image includes:

[0012] Traverse the pixels in the semantic image and classify the pixels into multiple categories based on their pixel values;

[0013] Connectivity analysis is performed on pixels in each category to extract semantic information.

[0014] In this embodiment of the application, the factor loading matrix corresponding to the point cluster in the vehicle coordinate system is determined by factor analysis, including:

[0015] Based on the point clusters in the vehicle coordinate system, the factor load matrix corresponding to the point clusters in the vehicle coordinate system is determined by combining the factor matrix and the special factor matrix.

[0016] In this embodiment, the factor loading matrix satisfies formula (1):

[0017] X = AF + E; (1)

[0018] Where X is the point cluster in the vehicle coordinate system, A is the factor loading matrix, F is the factor matrix, and E is the special factor matrix.

[0019] In this embodiment of the application, determining the point cluster in the vehicle coordinate system based on semantic information includes:

[0020] The semantic information is projected onto coordinates to determine the point clusters in the vehicle coordinate system.

[0021] In this embodiment of the application, the coordinate projection of semantic information to determine the point cluster in the vehicle coordinate system includes:

[0022] Determine the camera intrinsic and extrinsic parameters matrix;

[0023] Based on the camera intrinsic parameter matrix, semantic information is transformed into a cluster of pixels in the camera coordinate system;

[0024] The pixel clusters in the camera coordinate system are transformed into point clusters in the vehicle coordinate system based on the camera extrinsic matrix.

[0025] In this embodiment of the application, image data is processed to obtain a semantic image, including:

[0026] Semantic images are obtained by processing image data using a neural network model.

[0027] A second aspect of this application provides a controller, comprising:

[0028] The memory is configured to store instructions; and

[0029] The processor is configured to retrieve instructions from memory and, when executing instructions, to implement the aforementioned semantic information through a data dimensionality reduction method.

[0030] A third aspect of this application provides a data dimensionality reduction system for semantic information, comprising:

[0031] Controller;

[0032] The image acquisition device communicates with the controller and is configured to acquire image data.

[0033] A fourth aspect of this application provides a machine-readable storage medium storing instructions that cause a machine to execute the aforementioned data dimensionality reduction method for semantic information.

[0034] The above technical solution acquires image data sent by an image acquisition device and processes the image data to obtain a semantic image. Then, semantic information is extracted from the semantic image, and point clusters in the vehicle coordinate system are determined based on this semantic information. Next, factor analysis is used to determine the factor loading matrix corresponding to the point clusters in the vehicle coordinate system. Finally, the factor loading matrix corresponding to the point clusters in the vehicle coordinate system is used as the descriptor of the point clusters in the vehicle coordinate system to complete data dimensionality reduction. This application uses the factor loading matrix corresponding to the point clusters in the vehicle coordinate system as the descriptor of the point clusters in the vehicle coordinate system, which can reduce storage space consumption while ensuring usability.

[0035] Other features and advantages of the embodiments of this application will be described in detail in the following detailed description section. Attached Figure Description

[0036] The accompanying drawings are provided to further illustrate the embodiments of this application and form part of the specification. They are used together with the following detailed description to explain the embodiments of this application, but do not constitute a limitation on the embodiments of this application. In the drawings:

[0037] Figure 1 A flowchart illustrating a data dimensionality reduction method for semantic information according to an embodiment of this application is shown schematically.

[0038] Figure 2 A schematic block diagram of a controller according to an embodiment of this application is shown. Detailed Implementation

[0039] To make the objectives, technical solutions, and advantages of the embodiments of this application clearer, the technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. It should be understood that the specific embodiments described herein are only for illustration and explanation of the embodiments of this application and are not intended to limit the embodiments of this application. All other embodiments obtained by those skilled in the art based on the embodiments of this application without creative effort are within the scope of protection of this application.

[0040] It should be noted that if the embodiments of this application involve directional indicators (such as up, down, left, right, front, back, etc.), the directional indicators are only used to explain the relative positional relationship and movement of the components in a certain specific posture (as shown in the figure). If the specific posture changes, the directional indicators will also change accordingly.

[0041] Furthermore, if the embodiments of this application involve descriptions such as "first" or "second," these descriptions are for descriptive purposes only and should not be construed as indicating or implying their relative importance or implicitly specifying the number of technical features indicated. Therefore, features defined with "first" or "second" may explicitly or implicitly include at least one of those features. Additionally, the technical solutions of various embodiments can be combined with each other, but this must be based on the ability of those skilled in the art to implement them. If the combination of technical solutions is contradictory or impossible to implement, it should be considered that such a combination of technical solutions does not exist and is not within the scope of protection claimed in this application.

[0042] Figure 1 A flowchart illustrating a data dimensionality reduction method for semantic information according to an embodiment of this application is shown schematically. Figure 1 As shown in the figure, this application provides a method for dimensionality reduction of semantic information, applied to a controller. The controller communicates with an image acquisition device. The method may include the following steps:

[0043] Step 101: Acquire image data sent by the image acquisition device;

[0044] Step 102: Process the image data to obtain a semantic image;

[0045] Step 103: Extract semantic information from the semantic image;

[0046] Step 104: Determine the point cluster in the vehicle coordinate system based on semantic information;

[0047] Step 105: Determine the factor loading matrix corresponding to the point cluster in the vehicle coordinate system using factor analysis.

[0048] Step 106: Use the factor loading matrix corresponding to the point cluster in the vehicle coordinate system as the descriptor of the point cluster in the vehicle coordinate system to complete the data dimensionality reduction.

[0049] Existing technologies typically employ feature point methods for localization or semantic maps generated from dense semantic point clouds. However, feature point methods are significantly affected by lighting conditions, exhibiting substantial errors in varying scenarios such as daytime, nighttime, or rainy weather, thus limiting their applicability. Furthermore, localization using semantic maps generated from dense semantic point clouds suffers from excessive storage consumption. This application, by using the factor loading matrix corresponding to point clusters in the vehicle coordinate system as the descriptor for these point clusters, ensures the applicability of the data dimensionality reduction method while reducing storage space consumption.

[0050] In this embodiment, the image acquisition device can acquire image data around the vehicle and send the image data to the controller. The image acquisition device can be a camera or other image acquisition devices. After the controller acquires the image data sent by the image acquisition device, it can process the image data through a pre-trained neural network model, that is, perform semantic segmentation on the image data through the neural network model to obtain a semantic image. The semantic image is a single-channel semantic image. After obtaining the semantic image, the controller can extract the semantic information in the semantic image and perform coordinate projection on the semantic information to determine the point clusters in the vehicle coordinate system. Further, the controller can determine the factor loading matrix corresponding to the point clusters in the vehicle coordinate system through factor analysis. When there are multiple point clusters in multiple vehicle coordinate systems, each point cluster in the vehicle coordinate system has its corresponding factor loading matrix. Finally, the controller can use the factor loading matrix corresponding to the point clusters in the vehicle coordinate system as the descriptor of the point clusters in the vehicle coordinate system. In this way, the data dimensionality reduction of semantic information can be completed, reducing the consumption of storage space.

[0051] The above technical solution acquires image data sent by an image acquisition device and processes the image data to obtain a semantic image. Then, semantic information is extracted from the semantic image, and point clusters in the vehicle coordinate system are determined based on this semantic information. Next, factor analysis is used to determine the factor loading matrix corresponding to the point clusters in the vehicle coordinate system. Finally, the factor loading matrix corresponding to the point clusters in the vehicle coordinate system is used as the descriptor of the point clusters in the vehicle coordinate system to complete data dimensionality reduction. This application uses the factor loading matrix corresponding to the point clusters in the vehicle coordinate system as the descriptor of the point clusters in the vehicle coordinate system, which can reduce storage space consumption while ensuring usability.

[0052] In this embodiment of the application, step 102, processing the image data to obtain a semantic image, may include:

[0053] Semantic images are obtained by processing image data using a neural network model.

[0054] Specifically, the controller can process image data through a pre-trained neural network model, that is, perform semantic segmentation on the image data through the neural network model to obtain semantic images.

[0055] In this embodiment of the application, step 103, extracting semantic information from the semantic image, may include:

[0056] Traverse the pixels in the semantic image and classify the pixels into multiple categories based on their pixel values;

[0057] Connectivity analysis is performed on pixels in each category to extract semantic information.

[0058] Specifically, the controller can iterate through the pixels in the semantic image, determining the pixel value of each pixel. Each time a new pixel value is found, a corresponding blank image is created, and all pixels with the same pixel value are placed in the blank image. It's important to note that the position of a pixel in the blank image is the same as its position in the semantic image. In one example, if a pixel with a value of 8 is detected for the first time while iterating through the semantic image, the controller can create a corresponding blank image and further place all pixels with a value of 8 from the semantic image into this blank image. The steps are the same when other pixel values ​​are detected for the first time. Furthermore, after dividing all pixels in the semantic image, since multiple individuals of the same type may exist in each blank image, connected component analysis needs to be performed on each blank image to segment these individuals. This allows for the extraction of semantic information.

[0059] In this embodiment of the application, step 104, determining the point cluster in the vehicle coordinate system based on semantic information, may include:

[0060] The semantic information is projected onto coordinates to determine the point clusters in the vehicle coordinate system.

[0061] Specifically, the controller can determine point clusters in the vehicle coordinate system based on semantic information. In this embodiment, the semantic information includes pixel clusters on the image plane. The pixel clusters on the image plane are projected onto coordinates to determine the point clusters in the vehicle coordinate system. When the image acquisition device is a camera, due to the characteristics of the camera, the image data will lose depth information of the object. Therefore, it is necessary to assume that the height of all points in the point clusters in the vehicle coordinate system is 0. This completes the process of converting pixel clusters on the image plane into point clusters in the vehicle coordinate system, so as to subsequently determine the factor loading matrix corresponding to the point clusters in the vehicle coordinate system.

[0062] In this embodiment of the application, projecting semantic information onto coordinates to determine point clusters in the vehicle coordinate system may include:

[0063] Determine the camera intrinsic and extrinsic parameters matrix;

[0064] Based on the camera intrinsic parameter matrix, semantic information is transformed into a cluster of pixels in the camera coordinate system;

[0065] The pixel clusters in the camera coordinate system are transformed into point clusters in the vehicle coordinate system based on the camera extrinsic matrix.

[0066] Specifically, semantic information includes clusters of pixels on the image plane. In this embodiment, coordinate projection of semantic information refers to coordinate projection of the clusters of pixels on the image plane. The controller can obtain the camera intrinsic and extrinsic parameters through calibration, and further perform coordinate projection on each pixel in the clusters of pixels on the image plane. The controller can convert the semantic information into a cluster of pixels in the camera coordinate system based on the camera intrinsic parameter matrix, and then convert the clusters of pixels in the camera coordinate system into a cluster of pixels in the vehicle coordinate system based on the camera extrinsic parameter matrix. In one example, the coordinates of any pixel in the clusters of pixels on the image plane are given by K, the camera intrinsic parameter matrix is ​​K, and the camera extrinsic parameter matrix is ​​T. Multiplying the coordinates of any pixel, the camera intrinsic parameter matrix K, and the camera extrinsic parameter matrix T yields the corresponding point in the vehicle coordinate system. After completing the coordinate projection of all pixels in the clusters of pixels on the image plane, the cluster of pixels in the vehicle coordinate system is obtained.

[0067] In this embodiment of the application, step 105, determining the factor loading matrix corresponding to the point cluster in the vehicle coordinate system using factor analysis, may include:

[0068] Based on the point clusters in the vehicle coordinate system, the factor load matrix corresponding to the point clusters in the vehicle coordinate system is determined by combining the factor matrix and the special factor matrix.

[0069] Specifically, after obtaining the point cluster in the vehicle coordinate system, the controller can combine the factor matrix and the special factor matrix to determine the factor load matrix corresponding to the point cluster in the vehicle coordinate system.

[0070] In this embodiment, the factor loading matrix can satisfy formula (1):

[0071] X = AF + E; (1)

[0072] Where X is a variable, i.e. a cluster of points in the vehicle coordinate system, A is a factor loading matrix, F is a factor matrix, and E is a special factor matrix.

[0073] Specifically, the point cluster in the vehicle coordinate system contains n points, namely points x1…x n Assuming that the factors are independently and identically distributed in x ~ N(μ, ∑), the factor loading matrix can satisfy formula (1):

[0074] X = AF + E; (1)

[0075] The covariance of formula (1) is determined such that its covariance satisfies formula (2):

[0076] cov(X)=∑=Acov(F)A T +∑ ε (2)

[0077] Since the factors are independent of each other, let cov(F) = 1, then we get:

[0078] ∑=AA T +∑ ε (3)

[0079] For all points in the point cluster under the vehicle coordinate system, determine their Gaussian probability density function f(x) and multiply them together. From this, the likelihood probability satisfies formula (4):

[0080] L=f(x1)f(x2)…f(x n (4)

[0081] Where X is the variable, i.e., the point cluster in the vehicle coordinate system, A is the factor loading matrix, F is the factor matrix, E is the special factor matrix, and ∑ is the covariance of the point cluster in the vehicle coordinate system. ε Let L be the covariance of the special factor matrix and L be the likelihood probability.

[0082] When the likelihood probability reaches its maximum value, the covariance of the point cluster in the vehicle coordinate system can be obtained. Simultaneously, constraint A can be introduced. T ∑ -1 A = Λ, where Λ is a diagonal matrix. Thus, the factor loading matrix A can be obtained.

[0083] Figure 2 A schematic block diagram of a controller according to an embodiment of this application is shown. Figure 2 As shown in the figure, this application provides a controller that may include:

[0084] Memory 210 is configured to store instructions; and

[0085] Processor 220 is configured to retrieve instructions from memory 210 and, when executing instructions, to implement the aforementioned semantic information data dimensionality reduction method.

[0086] Specifically, in this embodiment of the application, the processor 220 can be configured to:

[0087] Acquire image data sent by the image acquisition device;

[0088] Image data is processed to obtain semantic images;

[0089] Extract semantic information from semantic images;

[0090] Determine the point cluster in the vehicle coordinate system based on semantic information;

[0091] The factor loading matrix corresponding to the point cluster in the vehicle coordinate system is determined by factor analysis.

[0092] The factor loading matrix corresponding to the point cluster in the vehicle coordinate system is used as the descriptor of the point cluster in the vehicle coordinate system to complete the data dimensionality reduction.

[0093] Furthermore, the processor 220 can also be configured to:

[0094] Traverse the pixels in the semantic image and classify the pixels into multiple categories based on their pixel values;

[0095] Connectivity analysis is performed on pixels in each category to extract semantic information.

[0096] Furthermore, the processor 220 can also be configured to:

[0097] Based on the point clusters in the vehicle coordinate system, the factor load matrix corresponding to the point clusters in the vehicle coordinate system is determined by combining the factor matrix and the special factor matrix.

[0098] In this embodiment, the factor loading matrix satisfies formula (1):

[0099] X = AF + E; (1)

[0100] Where X is a variable, i.e. a cluster of points in the vehicle coordinate system, A is a factor loading matrix, F is a factor matrix, and E is a special factor matrix.

[0101] Furthermore, the processor 220 can also be configured to:

[0102] The semantic information is projected onto coordinates to determine the point clusters in the vehicle coordinate system.

[0103] Furthermore, the processor 220 can also be configured to:

[0104] Determine the camera intrinsic and extrinsic parameters matrix;

[0105] Based on the camera intrinsic parameter matrix, semantic information is transformed into a cluster of pixels in the camera coordinate system;

[0106] The pixel clusters in the camera coordinate system are transformed into point clusters in the vehicle coordinate system based on the camera extrinsic matrix.

[0107] Furthermore, the processor 220 can also be configured to:

[0108] Semantic images are obtained by processing image data using a neural network model.

[0109] The above technical solution acquires image data sent by an image acquisition device and processes the image data to obtain a semantic image. Then, semantic information is extracted from the semantic image, and point clusters in the vehicle coordinate system are determined based on this semantic information. Next, factor analysis is used to determine the factor loading matrix corresponding to the point clusters in the vehicle coordinate system. Finally, the factor loading matrix corresponding to the point clusters in the vehicle coordinate system is used as the descriptor of the point clusters in the vehicle coordinate system to complete data dimensionality reduction. This application uses the factor loading matrix corresponding to the point clusters in the vehicle coordinate system as the descriptor of the point clusters in the vehicle coordinate system, which can reduce storage space consumption while ensuring usability.

[0110] This application also provides a semantic information data dimensionality reduction system, including:

[0111] Controller;

[0112] The image acquisition device communicates with the controller and is configured to acquire image data.

[0113] Specifically, the semantic information data dimensionality reduction system includes a controller and an image acquisition device. The controller can be used to execute the aforementioned semantic information data dimensionality reduction method. The image acquisition device communicates with the controller, can acquire image data, and send the image data to the controller so that the controller can execute the semantic information data dimensionality reduction method based on the image data.

[0114] This application also provides a machine-readable storage medium storing instructions that cause a machine to execute the aforementioned data dimensionality reduction method for semantic information.

[0115] Those skilled in the art will understand that embodiments of this application can be provided as methods, systems, or computer program products. Therefore, this application can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, this application can take the form of a computer program product embodied on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0116] This application is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of this application. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart... Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0117] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0118] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0119] In a typical configuration, a computing device includes one or more processors (CPU), input / output interfaces, network interfaces, and memory.

[0120] Memory may include non-persistent memory in computer-readable media, such as random access memory (RAM) and / or non-volatile memory, such as read-only memory (ROM) or flash RAM. Memory is an example of computer-readable media.

[0121] Computer-readable media includes both permanent and non-permanent, removable and non-removable media that can store information using any method or technology. Information can be computer-readable instructions, data structures, modules of programs, or other data. Examples of computer storage media include, but are not limited to, phase-change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other memory technologies, CD-ROM, digital versatile optical disc (DVD) or other optical storage, magnetic tape, magnetic disk storage or other magnetic storage devices, or any other non-transferable medium that can be used to store information accessible by a computing device. As defined herein, computer-readable media does not include transient computer-readable media, such as modulated data signals and carrier waves.

[0122] It should also be noted that the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such process, method, article, or apparatus. Unless otherwise specified, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes that element.

[0123] The above are merely embodiments of this application and are not intended to limit the scope of this application. Various modifications and variations can be made to this application by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of this application should be included within the scope of the claims of this application.

Claims

1. A method for reducing the dimensionality of semantic information data, characterized in that, Applied to a controller that communicates with an image acquisition device, including: Acquire image data sent by the image acquisition device; The image data is processed to obtain a semantic image; Traverse the pixels in the semantic image and classify the pixels into multiple categories based on their pixel values; Connectivity analysis is performed on pixels in each category to extract semantic information from the semantic image; The semantic information is projected onto coordinates to determine the point clusters in the vehicle coordinate system; The factor loading matrix corresponding to the point cluster in the vehicle coordinate system is determined by factor analysis. The factor loading matrix corresponding to the point cluster in the vehicle coordinate system is used as the descriptor of the point cluster in the vehicle coordinate system to complete the data dimensionality reduction. The step of performing coordinate projection on the semantic information to determine the point cluster in the vehicle coordinate system includes: Determine the camera intrinsic and extrinsic parameters matrix; Based on the camera intrinsic parameter matrix, the semantic information is transformed into a cluster of pixels in the camera coordinate system; Based on the camera extrinsic matrix, the pixel clusters in the camera coordinate system are transformed into point clusters in the vehicle coordinate system; The step of determining the factor loading matrix corresponding to the point cluster in the vehicle coordinate system using factor analysis includes: Based on the point clusters in the vehicle coordinate system, the factor load matrix corresponding to the point clusters in the vehicle coordinate system is determined by combining the factor matrix and the special factor matrix.

2. The data dimensionality reduction method according to claim 1, characterized in that, The factor loading matrix satisfies formula (1): ； （1） in, For the point cluster in the vehicle coordinate system, The factor loading matrix is... The factor matrix is... This is the special factor matrix.

3. The data dimensionality reduction method according to claim 1, characterized in that, The process of processing the image data to obtain a semantic image includes: The image data is processed using a neural network model to obtain a semantic image.

4. A controller, characterized in that, include: The memory is configured to store instructions; as well as A processor is configured to retrieve the instructions from the memory and, when executing the instructions, to implement the data dimensionality reduction method for semantic information according to any one of claims 1 to 3.

5. A semantic information data dimensionality reduction system, characterized in that, include: The controller according to claim 4; An image acquisition device, which communicates with the controller, is configured to acquire image data.

6. A machine-readable storage medium, characterized in that, The machine-readable storage medium stores instructions for causing the machine to perform a data dimensionality reduction method for semantic information according to any one of claims 1 to 3.

Images