Mars mineral identification method and device, electronic equipment and storage medium

Abstract

The invention relates to a Mars mineral identification method and device, electronic equipment and a storage medium. The method comprises the following steps: acquiring a Mars surface image in a preset wavelength range; acquiring a plurality of mineral pattern spots of the Mars surface image in a pixel-by-pixel manner; extracting deep spatial spectral features of different mineral pattern spots through a pre-trained multi-scale adaptive convolutional neural network; and classifying minerals contained in the Mars surface image according to the deep spatial spectral features of all mineral pattern spots to obtain a classification result. According to the method, the Mars surface image is obtained by combining near-infrared hyperspectral imaging, deep spatial spectral features of different mineral pattern spots are extracted based on the pre-trained multi-scale adaptive convolutional neural network, and rapid, lossless and accurate identification of Mars minerals can be realized.

Description

Technical Field

[0001] The present invention relates to the field of machine learning and image processing technology. Specifically, the present invention relates to a Martian mineral identification method, device, electronic device and storage medium. Background Art

[0002] Traditional methods for identifying minerals on the Martian surface primarily rely on spectral summary parameters. While this method has achieved some research results, its manual calculations lead to a low degree of automation and significant noise sensitivity, leaving much room for improvement in mineral identification accuracy. Furthermore, CRISM images, as hyperspectral data, despite their high characteristic dimensionality, also suffer from information redundancy, which can hinder accurate mineral identification.

[0003] Currently, feature extraction methods are often used to obtain important information from hyperspectral imagery, using dimensionality reduction techniques to retain as much effective information or features as possible. However, conventional dimensionality reduction methods cannot fully account for the nonlinear relationships in hyperspectral data. Although densely linked random forest models can account for mineral absorption characteristics within spectral bands, they are still insufficient in extracting key information. This can make it difficult to fully extract the deep spectral characteristics of Martian surface minerals from CRISM hyperspectral imagery, resulting in inaccurate mineral identification results in hyperspectral imagery of the Martian surface. Summary of the Invention

[0004] The technical problem to be solved by the present invention is to provide a Martian mineral identification method, device, electronic device and storage medium, aiming to solve at least one of the above technical problems.

[0005] In a first aspect, the present invention provides a technical solution to the above-mentioned technical problem as follows: a method for identifying Martian minerals, the method comprising:

[0006] Acquire images of the Martian surface within a preset wavelength range;

[0007] Acquire a plurality of mineral patches of the Martian surface image in a pixel-by-pixel manner;

[0008] The deep spatial spectral features of different mineral patterns are extracted through pre-trained multi-scale adaptive convolutional neural networks;

[0009] The minerals contained in the Martian surface image are classified according to the deep spatial spectral characteristics of all mineral patches to obtain a classification result.

[0010] The beneficial effects of the present invention are: by combining near-infrared hyperspectral imaging to obtain images of the Martian surface, and extracting the deep spatial spectral characteristics of different mineral patches based on a pre-trained multi-scale adaptive convolutional neural network, rapid, non-destructive and accurate identification of Martian minerals can be achieved.

[0011] On the basis of the above technical solution, the present invention can also be improved as follows.

[0012] Furthermore, before acquiring the plurality of mineral patches of the Martian surface image in a pixel-by-pixel manner, the method further includes:

[0013] performing spectral normalization processing on the Martian surface image to obtain original spectral characteristics;

[0014] Using principal component analysis and linear discriminant analysis, the original spectral features are reduced in dimension and enhanced to obtain enhanced spectral features;

[0015] The enhanced spectral features are combined with the original spectral features through an adaptive weighted fusion algorithm to generate an enhanced hyperspectral image.

[0016] Furthermore, before extracting the deep spatial spectral features of different mineral patches through the pre-trained multi-scale adaptive convolutional neural network, the method further includes:

[0017] According to the adaptive neighborhood selection algorithm, the size of each mineral spot is dynamically adjusted to obtain each adjusted mineral spot.

[0018] Furthermore, dynamically adjusting the size of each mineral spot according to the adaptive neighborhood selection algorithm to obtain each adjusted mineral spot includes:

[0019] For each of the mineral spots, determining an adjacent region image of the mineral spot according to the local difference of the spectral characteristics;

[0020] For each mineral spot, determining the correlation between the central pixel of the mineral spot and each neighboring pixel in the corresponding adjacent area image by spectral similarity measurement;

[0021] For each of the mineral spots, the size of the mineral spot is adjusted according to the correlation between the central pixel of the mineral spot and each neighborhood pixel in the corresponding adjacent area image to obtain an adjusted mineral spot.

[0022] Furthermore, the multi-scale adaptive convolutional neural network includes a multi-scale convolutional layer and a deep residual network layer, and the multi-scale convolutional layer includes three parallel branches and a feature fusion layer;

[0023] The deep spatial spectral features of different mineral patterns are extracted by pre-trained multi-scale adaptive convolutional neural network, including:

[0024] For each of the mineral spots, extracting the spatial spectral features of the mineral spot through each branch of the multi-scale convolutional layer;

[0025] For each mineral spot, the spatial spectrum features of the three branches corresponding to the mineral spot are weightedly fused through the feature fusion layer to obtain the fusion feature corresponding to the mineral spot;

[0026] For each of the mineral patches, the deep spatial spectral features in the fusion features are extracted through the deep residual network layer.

[0027] Furthermore, the method further comprises:

[0028] According to the deep spatial spectral characteristics of all mineral patches, each mineral contained in the Martian surface image is regionally segmented to obtain a segmentation result.

[0029] Furthermore, the method further comprises:

[0030] According to the classification result and the segmentation result, the minerals contained in the Martian surface image are identified and classified through a multimodal fusion strategy, and a mineral distribution map is output.

[0031] In a second aspect, in order to solve the above technical problems, the present invention further provides a Martian mineral identification device, which includes:

[0032] An acquisition module, used to acquire images of the Martian surface within a preset wavelength range;

[0033] a mineral pattern determination module, configured to obtain a plurality of mineral patterns of the Martian surface image in a pixel-by-pixel manner;

[0034] Feature extraction module, used to extract deep spatial spectral features of different mineral patterns through pre-trained multi-scale adaptive convolutional neural network;

[0035] The classification module is used to classify the minerals contained in the Martian surface image according to the deep spatial spectral characteristics of all mineral patches to obtain a classification result.

[0036] In a third aspect, in order to solve the above-mentioned technical problems, the present invention also provides an electronic device, which includes a memory, a processor, and a computer program stored in the memory and runnable on the processor. When the processor executes the computer program, the Martian mineral identification method of the present application is implemented.

[0037] In a fourth aspect, in order to solve the above-mentioned technical problems, the present invention also provides a computer-readable storage medium, on which a computer program is stored. When the computer program is executed by a processor, the Martian mineral identification method of the present application is implemented.

[0038] Additional aspects and advantages of the present application will be given in part in the following description, which will become apparent from the following description, or will be learned through practice of the present application. BRIEF DESCRIPTION OF THE DRAWINGS

[0039] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the following briefly introduces the drawings required for describing the embodiments of the present invention.

[0040] Figure 1 A schematic flow chart of a method for identifying Martian minerals according to an embodiment of the present invention;

[0041] Figure 2 A schematic flow chart of another Martian mineral identification method provided by one embodiment of the present invention;

[0042] Figure 3 A schematic diagram of a median spectrum of the recognition results of the CRISM-based data FRT93BE and the MICA spectral library provided in one embodiment of the present invention;

[0043] Figure 4 A schematic diagram of mineral mapping of the Jezero region provided in accordance with one embodiment of the present invention;

[0044] Figure 5 A schematic structural diagram of a Martian mineral identification device provided by one embodiment of the present invention;

[0045] Figure 6 The present invention provides a schematic structural diagram of an electronic device according to an embodiment of the present invention. DETAILED DESCRIPTION

[0046] The principles and features of the present invention are described below. The examples given are only used to explain the present invention and are not used to limit the scope of the present invention.

[0047] The following describes in detail the technical solution of the present invention and how the technical solution of the present invention solves the above-mentioned technical problems using specific embodiments. The following specific embodiments can be combined with each other, and the same or similar concepts or processes may not be repeated in some embodiments. The following embodiments of the present invention are described in conjunction with the accompanying drawings.

[0048] The solution provided by the embodiments of the present invention can be applied to any application scenario requiring the classification and identification of minerals in Martian surface images. The solution provided by the embodiments of the present invention can be executed by any electronic device, for example, a user's terminal device. The terminal device can be any terminal device that can install applications and access web pages through applications, including at least one of the following: a smartphone, tablet computer, laptop computer, desktop computer, smart speaker, smart watch, smart TV, or smart car device.

[0049] The embodiment of the present invention provides a possible implementation method, such as Figure 1 As shown in FIG, a flowchart of a method for identifying Martian minerals is provided. The method can be executed by any electronic device, for example, a terminal device, or by a terminal device and a server. For ease of description, the method provided by the embodiment of the present invention will be described below using a terminal device as an example of the execution subject. Figure 1 As shown in the flowchart, the method may include the following steps:

[0050] S10, acquiring an image of the Martian surface within a preset wavelength range;

[0051] S20, acquiring a plurality of mineral patches of the Martian surface image in a pixel-by-pixel manner;

[0052] S30, extracts deep spatial spectral features of different mineral patches through pre-trained multi-scale adaptive convolutional neural networks;

[0053] S40, classifying the minerals contained in the Martian surface image according to the deep spatial spectral characteristics of all mineral patches to obtain a classification result.

[0054] Through the method of the present invention, by combining near-infrared hyperspectral imaging to obtain images of the Martian surface, and extracting the deep spatial spectral characteristics of different mineral patches based on a pre-trained multi-scale adaptive convolutional neural network, rapid, non-destructive and accurate identification of Martian minerals can be achieved.

[0055] The following specific examples further illustrate the present invention. In this example, a Martian mineral identification method is provided. By combining near-infrared hyperspectral imaging technology with an adaptive spectral enhancement algorithm, a multi-scale adaptive convolutional neural network (MSA-CNN) is constructed to achieve rapid, non-destructive, and accurate identification of Martian minerals. This method aims to address the following issues with traditional Martian mineral identification methods:

[0056] Insufficient utilization of spectral features: Traditional methods have limited ability to extract deep spectral features from hyperspectral data, making it difficult to distinguish different minerals with similar spectral features.

[0057] Insufficient fusion of spatial and spectral features: Existing methods have limitations in the fusion of spatial and spectral information, resulting in insufficient accuracy and robustness of mineral identification.

[0058] Low real-time and automation level: Traditional methods rely on manual intervention and complex post-processing steps, which makes it difficult to meet the real-time and automated processing requirements of Mars exploration missions.

[0059] Poor adaptability: Existing methods have poor adaptability to the complex Martian surface environment and are unable to cope with practical problems such as lighting changes and noise interference.

[0060] Based on the above problems, a Martian mineral identification method provided in this embodiment may include the following steps:

[0061] S10, acquiring an image of the Martian surface within a preset wavelength range;

[0062] The preset wavelength range refers to the near-infrared wavelength range, for example, it can be 970-2500nm.

[0063] In the present application, an image of the Martian surface can be acquired in the wavelength range of 970-2500 nm based on near-infrared hyperspectral imaging technology, and the image of the Martian surface is a hyperspectral image.

[0064] Before S20, the method further includes:

[0065] The image is preprocessed using an adaptive spectral enhancement algorithm to enhance the discrimination of mineral spectral features in the Martian surface image. The specific steps include:

[0066] S11, performing spectral normalization processing on the Martian surface image to eliminate the effects of illumination and noise in the Martian surface image to obtain original spectral features;

[0067] S12, using principal component analysis (PCA) and linear discriminant analysis (LDA), reducing the dimension and enhancing the original spectral features, highlighting the spectral differences of the minerals, and obtaining enhanced spectral features;

[0068] S13, combining the enhanced spectral features with the original spectral features through an adaptive weighted fusion algorithm to generate an enhanced hyperspectral image.

[0069] In S20 , a plurality of mineral patches are obtained based on the enhanced hyperspectral image.

[0070] S20, acquiring a plurality of mineral patches of the Martian surface image in a pixel-by-pixel manner;

[0071] The mineral patch refers to a portion of an image of the Martian surface, and a point-centered strategy can be used to obtain multiple mineral patches of the Martian surface image in a pixel-by-pixel manner.

[0072] Before S30, the method further includes:

[0073] Adopting an adaptive neighborhood selection algorithm to dynamically adjust the size of each mineral patch to optimize the extraction of spatial spectral features, and obtaining each adjusted mineral patch, the following steps may be specifically included:

[0074] S21. For each mineral patch, determine an adjacent region image of the mineral patch based on local differences in spectral characteristics. If there are differences in spectral characteristics corresponding to different minerals, then specifically, the adjacent region image of the mineral patch may be selected based on the difference between the spectral characteristics corresponding to the mineral patch and the spectral characteristics corresponding to the surrounding region image of the mineral patch (the surrounding region image refers to the region image of the Martian surface excluding the mineral patch). The difference between the spectral characteristics corresponding to the adjacent region image and the spectral characteristics corresponding to the mineral patch is small.

[0075] Optionally, the size of the adjacent region image ranges from 5×5 to 11×11 pixels.

[0076] S22, for each of the mineral spots, determine the correlation between the central pixel of the mineral spot and each neighboring pixel in the corresponding adjacent area image through spectral similarity measurement; the greater the correlation between the central pixel and a neighboring pixel, the greater the possibility that the two pixels belong to the mineral spot corresponding to the same mineral.

[0077] S23, for each of the mineral spots, adjusting the size of the mineral spot according to the correlation between the central pixel of the mineral spot and each neighboring pixel in the corresponding adjacent area image to obtain an adjusted mineral spot.

[0078] Specifically, the neighborhood pixels in the adjacent area image with a correlation greater than a set correlation threshold can be adjusted to the mineral patch to achieve adjustment of the size of the mineral patch.

[0079] S30, extracts deep spatial spectral features of different mineral patches through pre-trained multi-scale adaptive convolutional neural networks;

[0080] Optionally, the multi-scale adaptive convolutional neural network includes a multi-scale convolutional layer and a deep residual network layer, and the multi-scale convolutional layer includes three parallel branches and a feature fusion layer;

[0081] Among them, the three branches are composed of the following: 1×1 convolution, 1×1 convolution and 3×3 convolution, and 1×1 convolution and two consecutive 3×3 convolutions.

[0082] In the above S30, the deep spatial spectral features of different mineral patches are extracted through the pre-trained multi-scale adaptive convolutional neural network, including:

[0083] S301, for each mineral spot, extracting the spatial spectrum feature of the mineral spot through each branch of the multi-scale convolutional layer;

[0084] S302, for each of the mineral spots, the spatial spectral features of the three branches corresponding to the mineral spot are weightedly fused through the feature fusion layer (which can be a layer integrated with the adaptive attention mechanism) to obtain the fusion feature corresponding to the mineral spot; avoiding the gradient disappearance problem.

[0085] S303: For each mineral patch, extract the deep spatial spectral features in the fusion features through the deep residual network layer (ResNet) to avoid the gradient vanishing problem.

[0086] S40, classifying the minerals contained in the Martian surface image according to the deep spatial spectral characteristics of all mineral patches to obtain a classification result.

[0087] The above-mentioned S40 can also be obtained based on a pre-trained model, which can be the multi-scale adaptive convolutional neural network mentioned above, or other models, which are not limited in this solution. In this model, a weight adjustment layer can be included. This layer can use an adaptive attention mechanism to dynamically adjust the importance of features by calculating the weights of spectral and spatial features, that is, adjust the weight of each mineral pattern. The model can also include a classification layer, such as a SoftMax classification layer, which is used to output the classification probability of the mineral and obtain the classification result based on the classification probability.

[0088] Optionally, the optimal number of convolution kernels for the first layer convolution operation of the multi-scale adaptive convolutional neural network (MSA-CNN) is 8, and the number of convolution kernels for each layer thereafter can be dynamically adjusted according to feature complexity.

[0089] Optionally, the method further includes:

[0090] Based on the deep spatial spectral characteristics of all mineral patches, each mineral contained in the Martian surface image is regionally segmented to obtain a segmentation result (which may be a regional image, i.e., a regional image corresponding to one mineral).

[0091] Specifically, a threshold segmentation algorithm based on deep learning can be used to combine spectral features and spatial context information to achieve accurate separation of single mineral areas.

[0092] More specifically, the following processes are included:

[0093] Step 1: Using the 1279 nm wavelength as the mask band, combining the spectral characteristics and spatial context information, dynamically calculate the segmentation threshold (α) in the range of 0.1 to 0.3, and perform regional segmentation on each mineral contained in the Martian surface image based on the segmentation threshold to obtain the initial segmentation result.

[0094] Step 2: Optimize the initial segmentation result through convolutional neural network, eliminate noise and mis-segmented areas, and obtain the final segmentation result.

[0095] Optionally, the method further comprises:

[0096] According to the classification result and the segmentation result, the minerals contained in the Martian surface image are identified and classified through a multimodal fusion strategy, and a mineral distribution map is output.

[0097] Specifically, the classification and segmentation results of the multi-scale adaptive convolutional neural network can be combined to identify and classify minerals through a multimodal fusion strategy, and the mineral distribution map can be output.

[0098] More specifically, for each single mineral, the following processing steps are included:

[0099] Step a: Calculate the ratio of target mineral pixels to all pixels in a single mineral area (segmentation result corresponding to a single mineral), and determine the mineral type by combining the classification result and segmentation result corresponding to the single mineral.

[0100] Step b: The mineral type is determined by an adaptive threshold (θ), and the adaptive threshold range is 0.03 to 0.07.

[0101] Step c: Output the mineral distribution map and mark the mineral type and spatial distribution information.

[0102] This method can be applied to mineral resource assessment, geological structure analysis, and environmental evolution research during Mars exploration missions, significantly improving the accuracy and efficiency of Martian mineral identification. It can also be expanded to other fields such as mineral resource exploration on Earth, geological disaster monitoring, and environmental protection.

[0103] This paper proposes a Martian mineral identification method based on multimodal deep learning and adaptive spectral enhancement. This method addresses the large volume of Martian CRISM hyperspectral imagery data and the complex spectral characteristics, providing an automated, efficient, and accurate mineral identification solution. Compared with traditional spectral aggregation parameter methods and existing deep learning methods, this paper achieves the following significant advantages through technological innovation:

[0104] 1. High-precision spectral feature extraction and enhancement

[0105] Adaptive spectral enhancement algorithm: Through spectral normalization, PCA dimensionality reduction, and adaptive weighted fusion, it effectively eliminates light and noise interference and enhances the discrimination of mineral spectra. For example, for the diagnostic absorption features of ferromagnesian montmorillonite at 1.9μm, 2.31μm, and 2.39μm, the spectral signal-to-noise ratio is improved by approximately 30%, significantly improving the detection capability of weak spectral features.

[0106] Multi-Scale Adaptive Convolutional Neural Network (MSA-CNN): This model combines multi-scale convolutional layers, a deep residual network (ResNet), and an adaptive attention mechanism to fully exploit the nonlinear spatial-spectral characteristics of hyperspectral data. Experiments show that the model achieves 98.2% classification accuracy for ferromagnesian montmorillonite, an improvement of approximately 25% over traditional methods.

[0107] 2. Dynamic optimization and intelligent processing

[0108] Dynamic Neighborhood Selection Algorithm: Adaptively adjust the size of mineral patches (5×5 to 11×11 pixels) based on local spectral differences to optimize the extraction of spatial-spectral features. For example, see Figure 4 ,In the complex surface environment of the Jezero area, the algorithm improves the ,feature extraction efficiency by about 40%.

[0109] Threshold segmentation based on deep learning: Using the 1279nm band as a mask, the segmentation threshold (α=0.1-0.3) is dynamically calculated by combining spectral features and spatial context information, and the segmentation results are optimized through convolutional neural networks, reducing the missegmentation rate to below 5%.

[0110] 3. Efficient automated processing and multimodal fusion

[0111] End-to-end automated process: From data pre-processing to mineral identification, the entire process requires no human intervention, increasing processing speed by approximately 60% compared to traditional methods. For example, processing time for seven hyperspectral images (235 bands) of the Jezero region was reduced to under 10 minutes per image.

[0112] Multimodal fusion strategy: Combining the classification results with the segmentation results, the mineral type is determined by an adaptive threshold (θ = 0.03 to 0.07), and the mineral distribution map is output. Experimental verification shows that the degree of agreement between the ferromagnesian montmorillonite identification results and the standard spectrum of the MICA spectral library is 97.5%, and the spatial distribution of minerals can be accurately marked. For details, please refer to Figure 3 .

[0113] Based on Figure 1 Based on the same principle as the method shown in , the embodiment of the present invention also provides a Martian mineral identification device 20, such as Figure 2 and Figure 5 As shown in FIG, the Martian mineral identification device 20 may include an acquisition module 210, a mineral pattern determination module 220, a feature extraction module 230, and a classification module 240, wherein:

[0114] Acquisition module 210, for acquiring the surface image of Mars within a preset wavelength range (corresponding to Figure 2 The processing of the data acquisition module in

[0115] a mineral patch determination module 220 for acquiring a plurality of mineral patches of the Martian surface image in a pixel-by-pixel manner;

[0116] Feature extraction module 230 is used to extract deep spatial spectral features of different mineral patterns (corresponding to Figure 2 The processing of the multi-scale adaptive convolutional neural network module shown in );

[0117] The classification module 240 is used to classify the minerals contained in the Martian surface image according to the deep spatial spectral characteristics of all mineral patches, and obtain the classification results (corresponding to Figure 2 The processing of the multi-scale adaptive convolutional neural network module shown in ).

[0118] Optionally, before acquiring the plurality of mineral patches of the Martian surface image in a pixel-by-pixel manner, the apparatus further comprises:

[0119] The preprocessing module is used to perform spectral normalization processing on the Mars surface image to obtain the original spectral features; use principal component analysis and linear discriminant analysis to reduce the dimension and enhance the original spectral features to obtain enhanced spectral features; combine the enhanced spectral features with the original spectral features through an adaptive weighted fusion algorithm to generate an enhanced hyperspectral image (corresponding to Figure 2 The processing of the adaptive spectrum enhancement module is shown in ).

[0120] Optionally, before extracting the deep spatial spectral features of different mineral patches through the pre-trained multi-scale adaptive convolutional neural network, the device further includes:

[0121] The adjustment module is used to dynamically adjust the size of each mineral spot according to the adaptive neighborhood selection algorithm to obtain each adjusted mineral spot (corresponding to Figure 2 The point-centric strategy and the processing process of the dynamic area selection module are shown in ).

[0122] Optionally, when the adjustment module dynamically adjusts the size of each mineral patch according to the adaptive neighborhood selection algorithm to obtain each adjusted mineral patch, it is specifically configured to:

[0123] For each of the mineral spots, determining an adjacent region image of the mineral spot according to the local difference of the spectral characteristics;

[0124] For each mineral spot, determining the correlation between the central pixel of the mineral spot and each neighboring pixel in the corresponding adjacent area image by spectral similarity measurement;

[0125] For each of the mineral spots, the size of the mineral spot is adjusted according to the correlation between the central pixel of the mineral spot and each neighborhood pixel in the corresponding adjacent area image to obtain an adjusted mineral spot.

[0126] Optionally, the multi-scale adaptive convolutional neural network includes a multi-scale convolutional layer and a deep residual network layer, and the multi-scale convolutional layer includes three parallel branches and a feature fusion layer;

[0127] When extracting deep spatial spectral features of different mineral patches through a pre-trained multi-scale adaptive convolutional neural network, the feature extraction module 230 is specifically used to:

[0128] For each of the mineral spots, extracting the spatial spectral features of the mineral spot through each branch of the multi-scale convolutional layer;

[0129] For each mineral spot, the spatial spectrum features of the three branches corresponding to the mineral spot are weightedly fused through the feature fusion layer to obtain the fusion feature corresponding to the mineral spot;

[0130] For each of the mineral patches, the deep spatial spectral features in the fusion features are extracted through the deep residual network layer.

[0131] Optionally, the device further comprises:

[0132] The segmentation module is used to perform regional segmentation on each mineral contained in the Mars surface image according to the deep spatial spectral characteristics of all mineral patches, and obtain the segmentation result (corresponding to Figure 2 The processing of the deep learning-based threshold segmentation module is shown in ).

[0133] Optionally, the device further comprises:

[0134] The mineral distribution map output module is used to identify and classify the minerals contained in the Mars surface image according to the classification result and the segmentation result through a multimodal fusion strategy, and output a mineral distribution map (corresponding to Figure 2 The processing of multimodal fusion and mineral recognition modules is shown in ).

[0135] The Martian mineral identification device of the embodiment of the present invention can execute the Martian mineral identification method provided by the embodiment of the present invention. The implementation principle is similar. The actions performed by each module and unit in the Martian mineral identification device in each embodiment of the present invention correspond to the steps in the Martian mineral identification method in each embodiment of the present invention. For the detailed functional description of each module of the Martian mineral identification device, please refer to the description of the corresponding Martian mineral identification method shown in the previous text, which will not be repeated here.

[0136] Among them, the above-mentioned Martian mineral identification device can be a computer program (including program code) running in a computer device, for example, the Martian mineral identification device is an application software; the device can be used to execute the corresponding steps in the method provided in the embodiment of the present invention.

[0137] In some embodiments, the Martian mineral identification device provided by the embodiments of the present invention can be implemented by a combination of software and hardware. As an example, the Martian mineral identification device provided by the embodiments of the present invention can be a processor in the form of a hardware decoding processor, which is programmed to execute the Martian mineral identification method provided by the embodiments of the present invention. For example, the processor in the form of a hardware decoding processor can adopt one or more application-specific integrated circuits (ASICs), DSPs, programmable logic devices (PLDs), complex programmable logic devices (CPLDs), field-programmable gate arrays (FPGAs), or other electronic components.

[0138] In other embodiments, the Martian mineral identification device provided by the embodiments of the present invention can be implemented in a software manner. Figure 5 A Martian mineral identification device stored in a memory is shown, which can be software in the form of a program and a plug-in, and includes a series of modules, including an acquisition module 210, a mineral pattern determination module 220, a feature extraction module 230 and a classification module 240, for implementing the Martian mineral identification method provided in an embodiment of the present invention.

[0139] The modules involved in the embodiments of the present invention may be implemented in software or hardware, wherein the name of a module does not necessarily limit the module itself.

[0140] Based on the same principle as the method shown in the embodiments of the present invention, an electronic device is also provided in the embodiments of the present invention, which may include but is not limited to: a processor and a memory; the memory is used to store computer programs; the processor is used to execute the method shown in any embodiment of the present invention by calling the computer program.

[0141] In an alternative embodiment, an electronic device is provided, such as Figure 6 As shown, Figure 6 The electronic device 4000 shown includes: a processor 4001 and a memory 4003. The processor 4001 and the memory 4003 are connected, for example, via a bus 4002. Optionally, the electronic device 4000 may further include a transceiver 4004, which can be used for data exchange between the electronic device and other electronic devices, such as data transmission and / or data reception. It should be noted that in actual applications, the number of transceivers 4004 is not limited to one, and the structure of the electronic device 4000 does not constitute a limitation on the embodiments of the present invention.

[0142] Processor 4001 may be a CPU (Central Processing Unit), a general-purpose processor, a DSP (Digital Signal Processor), an ASIC (Application Specific Integrated Circuit), an FPGA (Field Programmable Gate Array), or other programmable logic devices, transistor logic devices, hardware components, or any combination thereof. It may implement or execute the various exemplary logic blocks, modules, and circuits described in conjunction with the present disclosure. Processor 4001 may also be a combination that implements computing functions, such as a combination of one or more microprocessors, a combination of a DSP and a microprocessor, and the like.

[0143] Bus 4002 may include a path for transmitting information between the above components. Bus 4002 may be a PCI (Peripheral Component Interconnect) bus or an EISA (Extended Industry Standard Architecture) bus. Bus 4002 may be divided into an address bus, a data bus, a control bus, etc. For ease of representation, Figure 6 Only one thick line is used in the diagram, but this does not mean that there is only one bus or one type of bus.

[0144] The memory 4003 may be a ROM (Read Only Memory) or other types of static storage devices that can store static information and instructions, a RAM (Random Access Memory) or other types of dynamic storage devices that can store information and instructions, or an EEPROM (Electrically Erasable Programmable Read Only Memory), a CD-ROM (Compact Disc Read Only Memory) or other optical disk storage, optical disk storage (including compact discs, laser discs, optical discs, digital versatile discs, Blu-ray discs, etc.), a magnetic disk storage medium or other magnetic storage device, or any other medium that can be used to carry or store desired program code in the form of instructions or data structures and can be accessed by a computer, but is not limited to these.

[0145] The memory 4003 is used to store application code (computer program) for executing the solution of the present invention, and is controlled by the processor 4001. The processor 4001 is used to execute the application code stored in the memory 4003 to implement the content shown in the above method embodiment.

[0146] Among them, the electronic device can also be a terminal device, Figure 6 The electronic device shown is only an example and should not limit the functions and scope of use of the embodiments of the present invention.

[0147] An embodiment of the present invention provides a computer-readable storage medium having a computer program stored thereon. When the computer-readable storage medium is run on a computer, the computer can execute the corresponding contents of the aforementioned method embodiment.

[0148] According to another aspect of the present invention, a computer program product or computer program is provided, comprising computer instructions stored in a computer-readable storage medium. A processor of a computer device reads the computer instructions from the computer-readable storage medium and executes the computer instructions, causing the computer device to perform the methods provided in the various implementations described above.

[0149] Computer program code for performing the operations of the present invention may be written in one or more programming languages, or a combination thereof, including object-oriented programming languages such as Java, Smalltalk, C++, and conventional procedural programming languages such as "C" or similar programming languages. The program code may be executed entirely on the user's computer, partially on the user's computer, as a stand-alone software package, partially on the user's computer and partially on a remote computer, or entirely on the remote computer or server. In cases involving a remote computer, the remote computer may be connected to the user's computer through any type of network, including a local area network (LAN) or a wide area network (WAN), or may be connected to an external computer (e.g., through the Internet using an Internet service provider).

[0150] It should be understood that the flowcharts and block diagrams in the accompanying drawings illustrate the possible implementation architecture, functions and operations of the methods and computer program products according to various embodiments of the present invention. In this regard, each box in the flowchart or block diagram can represent a module, program segment, or a part of code, and the module, program segment, or a part of code contains one or more executable instructions for implementing the specified logical function. It should also be noted that in some alternative implementations, the functions marked in the box can also occur in a different order than the order marked in the accompanying drawings. For example, two boxes represented in succession can actually be executed substantially in parallel, and they can sometimes be executed in the opposite order, depending on the functions involved. It should also be noted that each box in the block diagram and / or flowchart, and the combination of the boxes in the block diagram and / or flowchart, can be implemented using a dedicated hardware-based system that performs the specified function or operation, or can be implemented using a combination of dedicated hardware and computer instructions.

[0151] The computer-readable storage medium provided by the embodiments of the present invention may be, for example, but not limited to, an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, device or component, or any combination thereof. More specific examples of computer-readable storage media may include, but are not limited to, an electrical connection with one or more wires, a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination thereof. In the present invention, a computer-readable storage medium may be any tangible medium containing or storing a program that can be used by or in conjunction with an instruction execution system, device or device.

[0152] The computer-readable storage medium carries one or more programs. When the one or more programs are executed by the electronic device, the electronic device executes the method shown in the above embodiment.

[0153] The above description is merely a preferred embodiment of the present invention and an illustration of the technical principles employed. Those skilled in the art should understand that the scope of disclosure involved in the present invention is not limited to the technical solutions formed by the specific combination of the above-mentioned technical features, but also includes other technical solutions formed by any combination of the above-mentioned technical features or their equivalents without departing from the above-mentioned disclosed concepts. For example, a technical solution formed by replacing the above-mentioned features with (but not limited to) technical features with similar functions disclosed in the present invention.

Claims

1. A method for identifying Martian minerals, characterized in that: The following steps are involved: Acquire images of the Martian surface within a preset wavelength range; Acquire a plurality of mineral patches of the Martian surface image in a pixel-by-pixel manner; The deep spatial spectral features of different mineral patterns are extracted through pre-trained multi-scale adaptive convolutional neural networks; The minerals contained in the Martian surface image are classified according to the deep spatial spectral characteristics of all mineral patches to obtain a classification result.

2. The method according to claim 1, characterized in that Before acquiring a plurality of mineral patches of the Martian surface image in a pixel-by-pixel manner, the method further includes: performing spectral normalization processing on the Martian surface image to obtain original spectral characteristics; Using principal component analysis and linear discriminant analysis, the original spectral features are reduced in dimension and enhanced to obtain enhanced spectral features; The enhanced spectral features are combined with the original spectral features through an adaptive weighted fusion algorithm to generate an enhanced hyperspectral image.

3. The method according to claim 1, characterized in that Before extracting the deep spatial spectral features of different mineral patterns through the pre-trained multi-scale adaptive convolutional neural network, the method further includes: According to the adaptive neighborhood selection algorithm, the size of each mineral spot is dynamically adjusted to obtain each adjusted mineral spot.

4. The method according to claim 3, characterized in that The method of dynamically adjusting the size of each mineral patch according to the adaptive neighborhood selection algorithm to obtain each adjusted mineral patch includes: For each of the mineral spots, determining an adjacent region image of the mineral spot according to the local difference of the spectral characteristics; For each mineral spot, determining the correlation between the central pixel of the mineral spot and each neighboring pixel in the corresponding adjacent area image by spectral similarity measurement; For each of the mineral spots, the size of the mineral spot is adjusted according to the correlation between the central pixel of the mineral spot and each neighborhood pixel in the corresponding adjacent area image to obtain an adjusted mineral spot.

5. The method according to any one of claims 1 to 4, characterized in that The multi-scale adaptive convolutional neural network includes a multi-scale convolutional layer and a deep residual network layer, and the multi-scale convolutional layer includes three parallel branches and a feature fusion layer; The deep spatial spectral features of different mineral patterns are extracted by pre-trained multi-scale adaptive convolutional neural network, including: For each of the mineral spots, extracting the spatial spectral features of the mineral spot through each branch of the multi-scale convolutional layer; For each mineral spot, the spatial spectrum features of the three branches corresponding to the mineral spot are weightedly fused through the feature fusion layer to obtain the fusion feature corresponding to the mineral spot; For each of the mineral patches, the deep spatial spectral features in the fusion features are extracted through the deep residual network layer.

6. The method according to any one of claims 1 to 4, characterized in that The method further comprises: According to the deep spatial spectral characteristics of all mineral patches, each mineral contained in the Martian surface image is regionally segmented to obtain a segmentation result.

7. The method according to claim 6, characterized in that The method further comprises: According to the classification result and the segmentation result, the minerals contained in the Martian surface image are identified and classified through a multimodal fusion strategy, and a mineral distribution map is output.

8. A Martian mineral identification device, characterized in that: include: An acquisition module, used to acquire images of the Martian surface within a preset wavelength range; a mineral pattern determination module, configured to obtain a plurality of mineral patterns of the Martian surface image in a pixel-by-pixel manner; Feature extraction module, used to extract deep spatial spectral features of different mineral patterns through pre-trained multi-scale adaptive convolutional neural network; The classification module is used to classify the minerals contained in the Martian surface image according to the deep spatial spectral characteristics of all mineral patches to obtain a classification result.

9. An electronic device, characterized in that: The method comprises a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor implements the method according to any one of claims 1 to 7 when executing the computer program.

10. A computer-readable storage medium, characterized in that The computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the method according to any one of claims 1 to 7 is implemented.

Images