Modern cemetery construction planning method and system

Abstract

A modern cemetery construction planning method and system are disclosed. First, the unmanned aerial vehicle collects multiple overhead views of the target area, and then splices these overhead views and uses artificial intelligence and large model-based image processing and generation technology to process them, thereby automatically generating a modern cemetery construction planning layout. In this way, different cemetery styles can be intelligently constructed according to the actual terrain, ensuring that the cemetery construction planning can focus on the utilization rate of land resources and the impact on the natural environment, ensuring that the ecological balance is not destroyed and sustainable development is achieved.

Description

Technical Field

[0001] This disclosure relates to the field of intelligent planning, specifically to a modern cemetery construction planning method and system. Background Technology

[0002] With the increasing aging of the global population and rising mortality rates, the demand for cemeteries is constantly increasing. This necessitates that cemetery construction planning adapt to changes in population structure and meet future needs. However, with population growth and accelerated urbanization, land resources available for cemetery construction are becoming increasingly scarce. Therefore, cemetery planning needs to achieve maximum utilization efficiency on limited land.

[0003] However, traditional cemetery construction often adopts a flat layout, which is relatively simple and occupies a large area of ​​land, failing to make full use of limited land resources. In addition, traditional cemetery construction planning usually relies on manual planning by professionals. This type of cemetery construction planning scheme has a low level of intelligence and is difficult to adapt to cemetery construction in different terrains. It often neglects the protection of the natural environment, and unreasonable cemetery layout may lead to environmental problems such as desertification and soil erosion.

[0004] Therefore, a modern cemetery construction planning scheme is desired. Summary of the Invention

[0005] This disclosure was made in consideration of the above issues. One object of this disclosure is to provide a modern method and system for cemetery construction and planning.

[0006] Embodiments of this disclosure provide a modern cemetery construction planning method, which includes:

[0007] Acquire multiple aerial images of the target area captured by a drone;

[0008] The multiple overhead images are stitched together to obtain a panoramic overhead image of the target area.

[0009] The panoramic aerial image of the target area is used to extract the region of interest features to obtain multiple region of interest feature maps;

[0010] Multi-scale semantic feature enhancement is performed on the feature maps of the multiple regions of interest to obtain multi-scale semantic feature vectors of the multiple regions of interest.

[0011] The multi-scale semantic feature vectors of the multiple regions of interest are input into the feature optimization module based on semantic association and semantic span to obtain multiple optimized multi-scale semantic feature vectors of the regions of interest.

[0012] Based on the multi-scale semantic feature vectors of the multiple optimized regions of interest, a recommended cemetery construction planning layout map is generated.

[0013] For example, in the modern cemetery construction planning method according to an embodiment of the present disclosure, the extraction of region of interest features from a panoramic aerial photograph of the target area to obtain multiple region of interest feature maps includes: inputting the panoramic aerial photograph of the target area into a region of interest feature extractor based on an R-CNN model to obtain the multiple region of interest feature maps.

[0014] For example, in the modern cemetery construction planning method according to an embodiment of the present disclosure, multi-scale semantic feature enhancement is performed on the feature maps of the plurality of regions of interest to obtain multi-scale semantic feature vectors of the plurality of regions of interest, including: passing the feature maps of the plurality of regions of interest through a multi-scale semantic feature extraction module to obtain multi-scale semantic feature vectors of the plurality of regions of interest.

[0015] For example, according to the modern cemetery construction planning method of the present disclosure, the method of extracting the feature maps of the plurality of regions of interest (ROIs) through a multi-scale semantic feature extraction module to obtain multi-scale semantic feature vectors of the plurality of ROIs includes:

[0016] The region of interest feature map is upsampled to obtain an upsampled region of interest feature map;

[0017] Global mean pooling is performed on each feature matrix along the channel dimension in the upsampled region of interest feature map to obtain the semantic channel feature vector of the upsampled region of interest;

[0018] The upsampled region of interest semantic channel feature vector is subjected to point convolution encoding to obtain the global semantic feature vector of the region of interest;

[0019] The region of interest feature map is upsampled to obtain an upsampled region of interest feature map;

[0020] Point convolutional encoding is performed on the upsampled region of interest feature map to obtain a channel-modulated upsampled region of interest feature map;

[0021] Two-dimensional convolutional encoding is performed on the feature map of the modulated upsampled region of interest in the channel to obtain the local semantic feature vector of the region of interest;

[0022] The global semantic feature vector of the region of interest and the local semantic feature vector of the region of interest are fused to obtain the multi-scale semantic feature vector of the region of interest.

[0023] For example, according to the modern cemetery construction planning method of the present disclosure, the plurality of multi-scale semantic feature vectors of regions of interest are input into a feature optimization module based on semantic association and semantic span to obtain a plurality of optimized multi-scale semantic feature vectors of regions of interest, including:

[0024] The multi-scale semantic feature vectors of each region of interest are concatenated to obtain a sequence of multiple local region of interest semantic row vectors.

[0025] Autocorrelation features are captured for each sequence of semantic row vectors of the plurality of local regions of interest to obtain a sequence of autocorrelation coefficients of the plurality of local regions of interest.

[0026] The sequence of autocorrelation coefficients of the multiple local regions of interest is arranged into multiple row vectors of autocorrelation coefficients of local regions of interest;

[0027] Calculate the class support weight vector of each local region of interest autocorrelation coefficient row vector in the multiple local regions of interest autocorrelation coefficient row vectors to obtain the semantic autocorrelation weight vector of multiple local regions of interest;

[0028] Calculate the number of vectors between every two multi-scale semantic feature vectors of the multiple regions of interest to obtain multiple local regions of interest semantic span count vectors;

[0029] Using the positional feature values ​​of each local region of interest semantic autocorrelation weight vector in the plurality of local regions of interest semantic autocorrelation weight vectors as weights, the weighted sum of the plurality of local regions of interest semantic span count vectors is calculated to obtain a sequence of local regions of interest semantic correlation representation vectors;

[0030] The sequence of semantic relevance representation vectors of the local region of interest is input into the activation function for activation processing to obtain the sequence of semantic relevance activation representation vectors of the local region of interest;

[0031] The product of each local region of interest semantic relevance activation vector in the sequence of local region of interest semantic relevance activation vectors and the first hyperparameter is calculated to obtain a sequence of local region of interest semantic relevance semantic weighted optimized feature vectors.

[0032] Calculate the product of each region of interest multiscale semantic feature vector and the second hyperparameter in the multiple regions of interest multiscale semantic feature vectors to obtain a sequence of region of interest semantic weighted modulation representation vectors;

[0033] The positions of each corresponding set of the sequence of semantically weighted optimized feature vectors of the local region of interest and the sequence of semantically weighted modulated representation vectors of the region of interest are summed to obtain the plurality of optimized multi-scale semantic feature vectors of the region of interest.

[0034] For example, according to the modern cemetery construction planning method of the present disclosure, the sequence of semantic row vectors of each of the plurality of local regions of interest (ROIs) is subjected to autocorrelation feature capture to obtain a sequence of autocorrelation coefficients of the plurality of ROIs, including:

[0035] Calculate the product between each semantic row vector of the local region of interest and a preset weight feature vector in the sequence of semantic row vectors of the local region of interest to obtain a sequence of autocorrelation weight values ​​of the local region of interest;

[0036] The sequence of autocorrelation weight values ​​of the local region of interest is subjected to nonlinear activation processing to obtain the sequence of autocorrelation coefficients of the local region of interest.

[0037] For example, according to the modern cemetery construction planning method of the present disclosure, calculating the class support weight vector of each local region of interest autocorrelation coefficient row vector in the plurality of local regions of interest autocorrelation coefficient row vectors to obtain a plurality of local regions of interest semantic autocorrelation weight vectors includes:

[0038] Each feature value in the autocorrelation coefficient row vector of the local region of interest is used as the exponent of the natural constant to calculate the position-based exponential function value with the natural constant as the base, so as to obtain the autocorrelation class support row vector of the local region of interest;

[0039] The summation of the feature values ​​at each position in the support row vector of the autocorrelation class of the local region of interest is calculated to obtain the global autocorrelation metric coefficient of the local region of interest.

[0040] The semantic autocorrelation weight vector of the region of interest is obtained by calculating the positional division between each feature value in the support row vector of the autocorrelation class of the region of interest and the global autocorrelation metric coefficient of the region of interest.

[0041] For example, according to an embodiment of the present disclosure, a method for planning the construction of a modern cemetery includes generating a recommended cemetery construction planning layout map based on the plurality of optimized regions of interest multi-scale semantic feature vectors. This includes arranging the plurality of optimized regions of interest multi-scale semantic feature vectors into a global semantic feature matrix of the target region and then using a cemetery construction planning layout map generator based on an adversarial generative network to obtain the recommended cemetery construction planning layout map.

[0042] Embodiments of this disclosure also provide a modern cemetery construction planning system, which includes:

[0043] The overhead image acquisition module is used to acquire multiple overhead images of the target area captured by the drone;

[0044] An image stitching module is used to stitch together the multiple overhead images to obtain a panoramic overhead image of the target area.

[0045] The region of interest feature extraction module is used to extract region of interest features from the panoramic aerial image of the target area to obtain multiple region of interest feature maps.

[0046] The multi-scale semantic feature enhancement module is used to perform multi-scale semantic feature enhancement on the feature maps of the multiple regions of interest respectively to obtain multi-scale semantic feature vectors of the multiple regions of interest.

[0047] The optimization module is used to input the multi-scale semantic feature vectors of the multiple regions of interest into the feature optimization module based on semantic association and semantic span to obtain multiple optimized multi-scale semantic feature vectors of regions of interest;

[0048] The planning and recommendation module is used to generate a recommended cemetery construction planning layout map based on the multi-scale semantic feature vectors of the multiple optimized regions of interest.

[0049] For example, in a modern cemetery construction planning system according to an embodiment of the present disclosure, the region of interest feature extraction module is used to include: inputting a panoramic aerial photograph of the target area into a region of interest feature extractor based on an R-CNN model to obtain the plurality of region of interest feature maps.

[0050] The modern cemetery construction planning method and system according to embodiments of this disclosure uses drones to collect multiple aerial photographs of a target area. These photographs are then stitched together and processed using image processing and generation techniques based on artificial intelligence and large-scale models to automatically generate a modern cemetery construction planning layout map. This allows for the intelligent construction of different cemetery styles based on the actual terrain, ensuring that cemetery construction planning considers land resource utilization and the impact on the natural environment, preventing ecological imbalance and achieving sustainable development. Attached Figure Description

[0051] To more clearly illustrate the technical solutions of the embodiments of this disclosure, the accompanying drawings of the embodiments of this disclosure will be briefly described below. Obviously, the drawings described below only relate to some embodiments of this disclosure and are not intended to limit this disclosure.

[0052] Figure 1 This illustration shows a schematic diagram of the application architecture of the modern cemetery construction planning method in an embodiment of this disclosure;

[0053] Figure 2 A flowchart illustrating the planning method for modern cemetery construction in an embodiment of this disclosure is shown;

[0054] Figure 3A flowchart of sub-step S540 of the modern cemetery construction planning method in this embodiment of the present disclosure is shown;

[0055] Figure 4 A flowchart of sub-step S550 of the modern cemetery construction planning method in an embodiment of this disclosure is shown;

[0056] Figure 5 A schematic diagram of the structure of the modern cemetery construction planning system in an embodiment of this disclosure is shown;

[0057] Figure 6 This invention discloses an application scenario diagram of the modern cemetery construction planning method in an embodiment of this disclosure; and

[0058] Figure 7 A schematic diagram of a storage medium according to an embodiment of the present disclosure is shown. Detailed Implementation

[0059] The technical solutions in the embodiments of this disclosure will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this disclosure, and not all embodiments. Based on the embodiments of this disclosure, all other embodiments obtained by those skilled in the art without creative effort are also within the scope of protection of this disclosure.

[0060] The terminology used in this specification is that which is currently widely used in the art in consideration of the functionality of this disclosure; however, these terms may vary depending on the intent, precedent, or new technology of those skilled in the art. Furthermore, specific terms may be chosen by the applicant, and in such cases, their detailed meanings will be described in the detailed description of this disclosure. Therefore, the terminology used in this specification should not be construed as simple names, but rather based on the meaning of the terms and the overall description of this disclosure.

[0061] While this disclosure makes various references to certain modules of systems according to embodiments of this disclosure, any number of different modules may be used and run on user terminals and / or servers. The modules described are merely illustrative, and different aspects of the systems and methods may use different modules.

[0062] This disclosure uses flowcharts to illustrate the operations performed by a system according to embodiments of this disclosure. It should be understood that the preceding or following operations are not necessarily performed in exact order. Instead, various steps can be processed in reverse order or simultaneously, as needed. Furthermore, other operations can be added to these processes, or one or more steps can be removed from them.

[0063] Figure 1The diagram illustrates the application architecture of the modern cemetery construction planning method in this embodiment, including a server 100 and a terminal device 200.

[0064] Terminal device 200 and server 100 can be connected via the Internet to enable communication between them. Optionally, the Internet described above uses standard communication technologies and / or protocols. The Internet is typically the Internet, but can also be any network, including but not limited to any combination of Local Area Network (LAN), Metropolitan Area Network (MAN), Wide Area Network (WAN), mobile, wired or wireless networks, private networks or virtual private networks. In some embodiments, technologies and / or formats including Hyper Text Markup Language (HTML), Extensible Markup Language (XML), etc., are used to represent data exchanged over the network. Furthermore, conventional encryption technologies such as Secure Socket Layer (SSL), Transport Layer Security (TLS), Virtual Private Network (VPN), and Internet Protocol Security (IPsec) can be used to encrypt all or some links. In other embodiments, customized and / or dedicated data communication technologies can be used to replace or supplement the aforementioned data communication technologies.

[0065] Server 100 can provide various network services to terminal device 200. Server 100 can be a single server, a server cluster consisting of several servers, or a cloud computing center. Specifically, server 100 may include processor 110 (Center Processing Unit, CPU), memory 120, input device 130, and output device 140, etc. Input device 130 may include keyboard, mouse, touch screen, etc., and output device 140 may include display device, such as liquid crystal display (LCD), cathode ray tube (CRT), etc.

[0066] The memory 120 may include a read-only memory (ROM) and a random access memory (RAM), and provides the processor 110 with program instructions and data stored in the memory 120. In this embodiment of the present disclosure, the memory 120 may be used to store a program for a modern cemetery construction planning method according to this embodiment of the present disclosure.

[0067] The processor 110 executes the steps of any of the modern cemetery construction planning methods in this disclosure embodiment by calling the program instructions stored in the memory 120.

[0068] Furthermore, the application architecture diagrams in this disclosure are for the purpose of more clearly illustrating the technical solutions in this disclosure and do not constitute a limitation on the technical solutions provided in this disclosure. Of course, the technical solutions provided in this disclosure are also applicable to similar problems for other application architectures and business applications.

[0069] The following examples or embodiments illustrate, in a non-limiting manner, a method for planning and constructing a modern cemetery according to at least one embodiment of this disclosure. As described below, different features in these specific examples or embodiments can be combined with each other without conflict to obtain new examples or embodiments, all of which are also within the scope of protection of this disclosure.

[0070] To address the aforementioned technical issues, the present invention discloses a method for automatically generating a modern cemetery construction planning layout map by using drones to collect multiple aerial photographs of a target area, stitching these images together, and then processing them using image processing and generation techniques based on artificial intelligence and large-scale models. This allows for the intelligent construction of different cemetery styles, such as simulated tree-shaped or multi-story structures, based on the actual terrain. This ensures that cemetery construction planning considers the utilization rate of land resources and the impact on the natural environment, ensuring that the ecological balance is not disrupted and achieving sustainable development.

[0071] Figure 2 A flowchart of a modern cemetery construction planning method according to an embodiment of this disclosure is shown. For example, this modern cemetery construction planning method can be executed by a server, which can be... Figure 1 Server 100 is shown in the image. (As shown in the image...) Figure 2As shown, the modern cemetery construction planning method according to an embodiment of this disclosure includes the following steps: S510, acquiring multiple aerial photographs of a target area collected by a drone; S520, stitching the multiple aerial photographs together to obtain a panoramic aerial photograph of the target area; S530, extracting region of interest (ROI) features from the panoramic aerial photograph of the target area to obtain multiple ROI feature maps; S540, performing multi-scale semantic feature enhancement on the multiple ROI feature maps to obtain multiple ROI multi-scale semantic feature vectors; S550, inputting the multiple ROI multi-scale semantic feature vectors into a feature optimization module based on semantic association and semantic span to obtain multiple optimized ROI multi-scale semantic feature vectors; S560, generating a recommended cemetery construction planning layout map based on the multiple optimized ROI multi-scale semantic feature vectors.

[0072] Specifically, in the technical solution disclosed herein, firstly, multiple aerial photographs of the target area are acquired using a drone. Next, considering that each aerial photograph may contain different local features and semantics of the target area collected by the drone, and a single aerial photograph cannot fully display all the details and environmental conditions of the target area, the technical solution disclosed herein requires image stitching of the multiple aerial photographs to obtain a panoramic aerial photograph of the target area. Stitching multiple aerial photographs provides a more comprehensive view of the target area, including topographical and geomorphological information from different angles and positions, enabling a more comprehensive understanding of the target area for cemetery construction planning.

[0073] Then, considering that the panoramic aerial image of the target area contains panoramic features and state information of the target area, some of which are important for the planning and layout of the cemetery, while others are redundant or irrelevant information, in order to extract the feature information of the region of interest from the panoramic aerial image of the target area to facilitate suitable cemetery construction planning for the target area, the technical solution of this disclosure further inputs the panoramic aerial image of the target area into a region of interest feature extractor based on an R-CNN model to obtain multiple region of interest feature maps. It should be understood that the R-CNN model can help accurately locate the region of interest in the image, i.e., the important features, structures, or objects in the target area, and capture the implicit state features of these regions of interest. Therefore, by using the region of interest feature extractor based on the R-CNN model, feature information of each region of interest in the target area can be learned and captured, including shape, color, texture, etc. These features help reveal the internal structure and characteristics of the target area, thereby helping to more accurately analyze and understand the characteristics of the target area, and providing more accurate and useful data support for subsequent planning work.

[0074] Accordingly, in step S530, region of interest feature extraction is performed on the panoramic aerial image of the target area to obtain multiple region of interest feature maps, including: inputting the panoramic aerial image of the target area into a region of interest feature extractor based on an R-CNN model to obtain the multiple region of interest feature maps.

[0075] Furthermore, considering that each region of interest (ROI) feature map in the target area contains different local feature information, and that terrain and landform features within the ROI may exhibit different state characteristics at different scales, and that different ROIs may require features at different scales to describe them (e.g., flat open areas may require larger-scale features, while complex terrain may require smaller-scale features to capture details), in order to extract semantic information from the image at different scales within each ROI, enabling planners to more comprehensively and accurately understand the features and structure of the target area, the technical solution of this disclosure further processes the multiple ROI feature maps through a multi-scale semantic feature extraction module to obtain multiple ROI multi-scale semantic feature vectors. Through the processing of the multi-scale semantic feature extraction module, feature information within the target ROI can be extracted at different scales, thereby obtaining richer and more comprehensive terrain feature representation information. This information helps to more accurately identify and understand the semantic content of the terrain, which is crucial for understanding the complexity and diversity of the region.

[0076] Accordingly, in step S540, multi-scale semantic feature enhancement is performed on the feature maps of the multiple regions of interest to obtain multi-scale semantic feature vectors of the multiple regions of interest, including: passing the feature maps of the multiple regions of interest through the multi-scale semantic feature extraction module to obtain multi-scale semantic feature vectors of the multiple regions of interest.

[0077] Specifically, such as Figure 3As shown, the method involves extracting the feature maps of the multiple regions of interest (ROIs) through a multi-scale semantic feature extraction module to obtain multi-scale semantic feature vectors for the multiple ROIs, including: S541, upsampling the ROI feature maps to obtain upsampled ROI feature maps; S542, performing global mean pooling on each feature matrix along the channel dimension of the upsampled ROI feature maps to obtain upsampled ROI semantic channel feature vectors; S543, performing point convolutional encoding on the upsampled ROI semantic channel feature vectors to obtain global ROI semantic feature vectors; S544, upsampling the ROI feature maps to obtain upsampled ROI feature maps; S545, performing point convolutional encoding on the upsampled ROI feature maps to obtain channel-modulated upsampled ROI feature maps; S546, performing two-dimensional convolutional encoding on the channel-modulated upsampled ROI feature maps to obtain local ROI semantic feature vectors; and S547, fusing the global ROI semantic feature vectors and the local ROI semantic feature vectors to obtain multi-scale ROI semantic feature vectors.

[0078] It should be understood that since each of the multiple region-of-interest (ROI) multi-scale semantic feature vectors contains multi-scale semantic feature information about each ROI, these semantic features of different ROIs have correlation relationships and hidden association features based on the panoramic view of the target area. This association relationship is related to the regional span; for example, the larger the span between ROIs, the weaker the correlation between the topographic semantics and state features of these two regions. Based on this, in order to further optimize and improve the expressive power of each ROI feature to better reflect the characteristics and structure of the target area, the technical solution of this disclosure further inputs the multiple ROI multi-scale semantic feature vectors into a feature optimization module based on semantic association and semantic span to obtain multiple optimized ROI multi-scale semantic feature vectors. Through the processing of the feature optimization module based on semantic association and semantic span, the inherent association patterns between the semantic features of each ROI based on the regional span can be learned and captured, thereby optimizing the expression of the semantic features of each ROI, making the features of different ROIs more consistent and global, which helps to ensure the consistency and integrity of the overall planning and layout of the cemetery construction.

[0079] Accordingly, in step S550, as Figure 4As shown, the method involves inputting the multiple region-of-interest (ROI) multi-scale semantic feature vectors into a feature optimization module based on semantic association and semantic span to obtain multiple optimized ROI multi-scale semantic feature vectors, including: S5501, concatenating every two ROI multi-scale semantic feature vectors to obtain a sequence of multiple local ROI semantic row vectors; S5502, performing autocorrelation feature capture on the sequence of each local ROI semantic row vector to obtain a sequence of multiple local ROI autocorrelation coefficients; S5503, inputting the... The sequence of autocorrelation coefficients of multiple regions of interest is arranged into multiple row vectors of autocorrelation coefficients of regions of interest; S5504, the class support weight vector of each row vector of autocorrelation coefficients of regions of interest is calculated to obtain multiple semantic autocorrelation weight vectors of regions of interest; S5505, the number of vectors between every two multi-scale semantic feature vectors of regions of interest is calculated to obtain multiple semantic span of regions of interest; S5506, each of the multiple semantic autocorrelation weight vectors of regions of interest is used to calculate the class support weight vector of each row vector of autocorrelation coefficients of regions of interest. Using the positional feature values ​​of the semantic autocorrelation weight vector of the region of interest as weights, the weighted sum of the multiple semantic span count vectors of the regions of interest is calculated to obtain a sequence of semantic autocorrelation representation vectors of the regions of interest; S5507, the sequence of semantic autocorrelation representation vectors of the regions of interest is input into an activation function for activation processing to obtain a sequence of semantic autocorrelation activation representation vectors of the regions of interest; S5508, the product of each semantic autocorrelation activation representation vector of the regions of interest in the sequence of semantic autocorrelation activation representation vectors of the regions of interest with the first hyperparameter is calculated to obtain the semantic autocorrelation of the regions of interest. S5509, calculate the product of each region-of-interest (ROI) multi-scale semantic feature vector and the second hyperparameter in the plurality of ROI multi-scale semantic feature vectors to obtain the sequence of ROI semantic weighted modulation representation vectors; S5510, calculate the positional summation between each corresponding group of local ROI semantic relevance semantic weighted optimized feature vectors and ROI semantic weighted modulation representation vectors in the sequence of local ROI semantic relevance semantic weighted optimized feature vectors and the sequence of ROI semantic weighted modulation representation vectors to obtain the plurality of optimized ROI multi-scale semantic feature vectors.

[0080] In step S5502, the sequence of semantic row vectors of the plurality of local regions of interest is subjected to autocorrelation feature capture to obtain a sequence of autocorrelation coefficients of the plurality of local regions of interest. This includes: calculating the product between each semantic row vector of the local regions of interest and a preset weight feature vector to obtain a sequence of autocorrelation weight values ​​of the local regions of interest; and performing nonlinear activation processing on the sequence of autocorrelation weight values ​​of the local regions of interest to obtain a sequence of autocorrelation coefficients of the local regions of interest.

[0081] In step S5504, calculating the class support weight vector of each local region of interest (ROI) autocorrelation coefficient row vector in the plurality of ROI autocorrelation coefficient row vectors to obtain a plurality of ROI semantic autocorrelation weight vectors includes: using each feature value in the ROI autocorrelation coefficient row vectors as the exponent of the natural constant to calculate the positional exponential function value with the natural constant as the base to obtain the ROI autocorrelation class support row vectors; calculating the sum of the positional feature values ​​in the ROI autocorrelation class support row vectors to obtain the ROI global autocorrelation metric coefficient; and calculating the positional division between each feature value in the ROI autocorrelation class support row vectors and the ROI global autocorrelation metric coefficient to obtain the ROI semantic autocorrelation weight vectors.

[0082] Specifically, in one example, inputting the multiple multi-scale semantic feature vectors of regions of interest into a feature optimization module based on semantic association and semantic span to obtain multiple optimized multi-scale semantic feature vectors of regions of interest includes: inputting the multiple multi-scale semantic feature vectors of regions of interest into the feature optimization module based on semantic association and semantic span and processing them with the following feature optimization formula to obtain the multiple optimized multi-scale semantic feature vectors of regions of interest; wherein, the feature optimization formula is:

[0083] S ij =f1(W[v i ||v j(j≠i) ])

[0084]

[0085] Among them, v i ,v j Let i and j represent the i-th and j-th multi-scale semantic feature vectors of the multiple regions of interest, respectively, and W represent the weight feature vector. i ||v j ] represents the vector concatenation operation, f1 is the non-linear activation function, and Sij Let represent the eigenvalues ​​at each position in the semantic autocorrelation matrix of the local region of interest, exp(·) is the exponential operation with the natural constant e as the base, n is the number of vectors in the multi-scale semantic feature vectors of the multiple regions of interest minus one, and α ij v′ represents the feature value at each position in the semantic autocorrelation weight vector of the local region of interest. j For v j With v i The number of feature vectors separated by each other, α and β represent trainable hyperparameters, σ(·) represents the activation function, v′ i This represents the i-th multi-scale semantic feature vector of the multiple optimized regions of interest.

[0086] Subsequently, the multi-scale semantic feature vectors of the multiple optimized regions of interest are arranged into a global semantic feature matrix of the target region. This matrix is ​​then used by a cemetery construction planning layout generator based on an adversarial generative network to obtain a recommended cemetery construction planning layout. In other words, the global semantic feature matrix of the target region, formed by arranging the multi-scale semantic features of different regions of interest, is used to generate an adversarial cemetery construction planning layout. This allows for the intelligent construction of different cemetery styles based on the actual terrain, such as simulated tree-shaped or multi-story structures, ensuring that cemetery construction planning considers the utilization rate of land resources and the impact on the natural environment, ensuring that the ecological balance is not disrupted and achieving sustainable development.

[0087] Accordingly, in step S560, a recommended cemetery construction planning layout map is generated based on the multiple optimized regions of interest multi-scale semantic feature vectors, including: arranging the multiple optimized regions of interest multi-scale semantic feature vectors into a global semantic feature matrix of the target region, and then using a cemetery construction planning layout map generator based on an adversarial generative network to obtain the recommended cemetery construction planning layout map.

[0088] In this embodiment, each of the plurality of region-of-interest (ROI) multi-scale semantic feature vectors represents the topographic image semantic features of each ROI in the panoramic aerial view of the target area. When the plurality of ROI multi-scale semantic feature vectors are input into the feature optimization module based on semantic association and semantic span, the feature optimization module uses the semantic association information and semantic span information between the ROI multi-scale semantic feature vectors to adaptively enhance the feature expression, thereby strengthening the saliency between the feature distributions of the plurality of ROI multi-scale semantic feature vectors. However, feature optimization based on semantic association and semantic span may also cause local overflow features in the plurality of optimized ROI multi-scale semantic feature vectors due to differences in attention weights, thereby affecting the accuracy of the recommended cemetery construction planning layout map obtained by the cemetery construction planning layout map generator based on adversarial generative networks of the target area global semantic feature matrix.

[0089] Preferably, obtaining a recommended cemetery construction planning layout map by using a cemetery construction planning layout map generator based on an adversarial generative network from the global semantic feature matrix of the target area includes the following steps: expanding the global semantic feature matrix of the target area into a global semantic feature vector of the target area; calculating the global semantic sum representation matrix and the global semantic difference representation matrix of the target area global semantic feature vector, wherein the value at position (i,j) of the global semantic sum representation matrix of the target area is the mean of the eigenvalues ​​at positions i and j of the global semantic feature vector of the target area, and the value at position (i,j) of the global semantic difference representation matrix of the target area is the absolute value of the difference between the eigenvalues ​​at positions i and j of the global semantic feature vector of the target area; performing matrix multiplication between the transpose of the global semantic feature vector of the target area and the global semantic sum representation matrix of the target area to obtain the global semantic sum representation vector of the target area and then performing matrix multiplication on the transpose of the global semantic feature vector of the target area and the global semantic sum representation matrix of the target area. The target region global semantic difference representation matrix is ​​multiplied by the target region global semantic feature vector to obtain the target region global semantic difference representation vector, wherein the target region global semantic feature vector is a column vector; the matrix product of the target region global semantic sum representation matrix and the target region global semantic difference representation matrix is ​​calculated, and the transpose of the target region global semantic feature vector is multiplied by the matrix product, and then added to the transpose of the leakage point aggregation inference sum representation vector and the leakage point aggregation inference difference representation vector to obtain the corrected target region global semantic feature vector; the corrected target region global semantic feature vector is restored to the corrected target region global semantic feature matrix according to the expansion of the target region global semantic feature matrix; and the corrected target region global semantic feature matrix is ​​used to generate a recommended cemetery construction planning layout map through a cemetery construction planning layout map generator based on adversarial generative networks.

[0090] In other words, by using the sum and difference representation matrices of the group aggregation statistical evaluation used for the local distribution of eigenvalue granularity of the global semantic feature matrix of the target region as the retrieval and response distribution enhancement of the global semantic feature matrix of the target region, and constructing a referenceless distribution retrieval response framework for the global semantic feature matrix of the target region in an open domain based on the sum-difference feature distribution of the group aggregation, the redundancy of the distribution response caused by the local overflow features of the global semantic feature matrix of the target region is avoided through response superposition. This achieves the fidelity constraint of the self-aggregation statistical correlation of the response from the global semantic feature matrix of the target region to the target generation regression domain, and improves the accuracy of the recommended cemetery construction planning layout map obtained by the cemetery construction planning layout map generator based on the global semantic feature matrix of the target region. In this way, modern cemetery construction planning layout maps can be generated more accurately according to the actual terrain, thereby realizing the intelligent construction of different cemetery styles, ensuring that cemetery construction planning can pay attention to the utilization rate of land resources and the impact on the natural environment, ensuring that the ecological balance is not destroyed, and achieving sustainable development.

[0091] It is understandable that the construction of modern cemeteries addresses the current regulations prohibiting the development of forest and arable land, and the negative phenomena of farmers burying their dead indiscriminately, damaging the ecological environment (and easily causing fires). Utilizing abandoned mines, bare rocky areas, and barren land with poor ecological conditions (difficult to restore and requiring significant capital investment) as development sites, and fully introducing social capital, allows for the development of beautiful, intensive multi-story (building-style) burial plots. This further promotes the restoration of fragile environments and cultural tourism development. The shape of the cemetery is not limited to a tower shape; it can also be designed in special styles according to the overall landscape plan, such as mosques, Fujian Tulou (earthen buildings), or even simulated tree shapes. Currently, most cemeteries are flat, resulting in low utilization and environmental damage. With multi-story cemeteries, several times more burial plots can be accommodated in the same area, and after beautification and lighting, cultural tourism can also be developed.

[0092] Based on the above embodiments, see Figure 5The diagram shown is a structural schematic of a modern cemetery construction planning system 800 according to an embodiment of this disclosure. The modern cemetery construction planning system 800 includes: an aerial image acquisition module 810, used to acquire multiple aerial images of a target area collected by a drone; an image stitching module 820, used to stitch the multiple aerial images together to obtain a panoramic aerial image of the target area; a region of interest (ROI) feature extraction module 830, used to extract ROI features from the panoramic aerial image of the target area to obtain multiple ROI feature maps; a multi-scale semantic feature enhancement module 840, used to enhance the ROI feature maps at multiple scales to obtain multiple ROI multi-scale semantic feature vectors; an optimization module 850, used to input the multiple ROI multi-scale semantic feature vectors into a feature optimization module based on semantic association and semantic span to obtain multiple optimized ROI multi-scale semantic feature vectors; and a planning recommendation module 860, used to generate a recommended cemetery construction planning layout map based on the multiple optimized ROI multi-scale semantic feature vectors.

[0093] Furthermore, the region of interest feature extraction module 830 is used to include: inputting the panoramic aerial image of the target region into a region of interest feature extractor based on an R-CNN model to obtain the plurality of region of interest feature maps.

[0094] Here, those skilled in the art will understand that the specific functions and operations of each module in the aforementioned modern cemetery construction planning system 800 have been referenced above. Figures 2 to 4 The description of the modern cemetery construction planning method is detailed here, and therefore, its repeated description will be omitted.

[0095] Figure 6 This is an application scenario diagram of the modern cemetery construction planning method according to embodiments of this disclosure. For example... Figure 6 As shown, in this application scenario, firstly, multiple aerial images of the target area are acquired by the drone (e.g., Figure 6 As shown in D), the multiple aerial images are then input into a server deployed with modern cemetery construction planning algorithms (e.g., Figure 6 As shown in S), the server is able to use the modern cemetery construction planning algorithm to process the multiple aerial images to obtain a recommended cemetery construction planning layout map.

[0096] Based on the above embodiments, this disclosure also provides an electronic device with another exemplary implementation. In some possible implementations, the electronic device in this disclosure may include a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor, when executing the program, can implement the steps of the modern cemetery construction planning method described above.

[0097] For example, disclosure based on electronic devices Figure 1 Taking server 100 as an example, the processor in the electronic device is the processor 110 in server 100, and the memory in the electronic device is the memory 120 in server 100.

[0098] Embodiments of this disclosure also provide a computer-readable storage medium. Figure 7 A schematic diagram of a computer-readable storage medium 1000 according to an embodiment of the present disclosure is shown. Figure 7 As shown, the computer-readable storage medium 1000 stores computer-executable instructions 1001. When the computer-executable instructions 1001 are executed by a processor, the modern cemetery construction planning method according to embodiments of this disclosure, as described with reference to the above figures, can be executed. The computer-readable storage medium includes, but is not limited to, volatile memory and / or non-volatile memory. The volatile memory may, for example, include random access memory (RAM) and / or cache memory. The non-volatile memory may, for example, include read-only memory (ROM), hard disk, flash memory, etc.

[0099] Embodiments of this disclosure also provide a computer program product or computer program including computer-executable instructions stored in a computer-readable storage medium. A processor of a computer device reads the computer-executable instructions from the computer-readable storage medium and executes the computer-executable instructions, causing the computer device to perform a modern cemetery construction planning method according to embodiments of this disclosure.

[0100] Those skilled in the art will understand that the contents disclosed herein can be varied and modified in many ways. For example, the various devices or components described above can be implemented in hardware, or in software, firmware, or a combination of some or all of the three.

[0101] Furthermore, while this disclosure makes various references to certain units in systems according to embodiments of this disclosure, any number of different units may be used and run on clients and / or servers. The units described are merely illustrative, and different aspects of the systems and methods may use different units.

[0102] Those skilled in the art will understand that all or part of the steps in the above methods can be implemented by a program instructing related hardware, and the program can be stored in a computer-readable storage medium, such as a read-only memory, a disk, or an optical disk. Optionally, all or part of the steps in the above embodiments can also be implemented using one or more integrated circuits. Accordingly, each module / unit in the above embodiments can be implemented in hardware or as a software functional module. This disclosure is not limited to any particular combination of hardware and software.

[0103] Unless otherwise defined, all terms used herein (including technical and scientific terms) shall have the same meaning as commonly understood by one of ordinary skill in the art to which this disclosure pertains. It should also be understood that terms such as those defined in a common dictionary shall be interpreted as having a meaning consistent with their meaning in the context of the relevant art, and not as having an idealized or highly formalized meaning, unless expressly defined herein.

[0104] The foregoing description is a description of the present disclosure and should not be construed as limiting it. Although several exemplary embodiments of the present disclosure have been described, those skilled in the art will readily understand that many modifications can be made to the exemplary embodiments without departing from the novel teachings and advantages of the present disclosure. Therefore, all such modifications are intended to be included within the scope of the present disclosure as defined by the claims. It should be understood that the foregoing description is a description of the present disclosure and should not be construed as limiting it to the specific embodiments disclosed, and modifications to the disclosed embodiments and other embodiments are intended to be included within the scope of the appended claims. The present disclosure is defined by the claims and their equivalents.

Claims

1. A modern cemetery construction and planning method, characterized in that, include: Acquire multiple aerial images of the target area captured by a drone; The multiple overhead images are stitched together to obtain a panoramic overhead image of the target area. The panoramic aerial image of the target area is used to extract the region of interest features to obtain multiple region of interest feature maps; Multi-scale semantic feature enhancement is performed on the feature maps of the multiple regions of interest to obtain multi-scale semantic feature vectors of the multiple regions of interest. The multi-scale semantic feature vectors of the multiple regions of interest are input into the feature optimization module based on semantic association and semantic span to obtain multiple optimized multi-scale semantic feature vectors of the regions of interest. Based on the multi-scale semantic feature vectors of the multiple optimized regions of interest, a recommended cemetery construction planning layout map is generated.

2. The modern cemetery construction planning method according to claim 1, characterized in that, Extracting region of interest features from the panoramic aerial image of the target area to obtain multiple region of interest feature maps includes: inputting the panoramic aerial image of the target area into a region of interest feature extractor based on an R-CNN model to obtain the multiple region of interest feature maps.

3. The modern cemetery construction planning method according to claim 2, characterized in that, Performing multi-scale semantic feature enhancement on the feature maps of the multiple regions of interest to obtain multi-scale semantic feature vectors of the multiple regions of interest includes: passing the feature maps of the multiple regions of interest through a multi-scale semantic feature extraction module to obtain multi-scale semantic feature vectors of the multiple regions of interest.

4. The modern cemetery construction planning method according to claim 3, characterized in that, The feature maps of the multiple regions of interest are respectively processed by a multi-scale semantic feature extraction module to obtain multi-scale semantic feature vectors for the multiple regions of interest, including: The region of interest feature map is upsampled to obtain an upsampled region of interest feature map; Global mean pooling is performed on each feature matrix along the channel dimension in the upsampled region of interest feature map to obtain the semantic channel feature vector of the upsampled region of interest; The upsampled region of interest semantic channel feature vector is subjected to point convolution encoding to obtain the global semantic feature vector of the region of interest; The region of interest feature map is upsampled to obtain an upsampled region of interest feature map; Point convolutional encoding is performed on the upsampled region of interest feature map to obtain a channel-modulated upsampled region of interest feature map; Two-dimensional convolutional encoding is performed on the feature map of the modulated upsampled region of interest in the channel to obtain the local semantic feature vector of the region of interest; The global semantic feature vector of the region of interest and the local semantic feature vector of the region of interest are fused to obtain the multi-scale semantic feature vector of the region of interest.

5. The modern cemetery construction planning method according to claim 4, characterized in that, The multi-scale semantic feature vectors of the multiple regions of interest are input into a feature optimization module based on semantic association and semantic span to obtain multiple optimized multi-scale semantic feature vectors of the regions of interest, including: The multi-scale semantic feature vectors of each region of interest are concatenated to obtain a sequence of multiple local region of interest semantic row vectors. Autocorrelation features are captured for each sequence of semantic row vectors of the plurality of local regions of interest to obtain a sequence of autocorrelation coefficients of the plurality of local regions of interest. The sequence of autocorrelation coefficients of the multiple local regions of interest is arranged into multiple row vectors of autocorrelation coefficients of local regions of interest; Calculate the class support weight vector of each local region of interest autocorrelation coefficient row vector in the multiple local regions of interest autocorrelation coefficient row vectors to obtain the semantic autocorrelation weight vector of multiple local regions of interest; Calculate the number of vectors between every two multi-scale semantic feature vectors of the multiple regions of interest to obtain multiple local regions of interest semantic span count vectors; Using the positional feature values ​​of each local region of interest semantic autocorrelation weight vector in the plurality of local regions of interest semantic autocorrelation weight vectors as weights, the weighted sum of the plurality of local regions of interest semantic span count vectors is calculated to obtain a sequence of local regions of interest semantic correlation representation vectors; The sequence of semantic relevance representation vectors of the local region of interest is input into the activation function for activation processing to obtain the sequence of semantic relevance activation representation vectors of the local region of interest; The product of each local region of interest semantic relevance activation vector in the sequence of local region of interest semantic relevance activation vectors and the first hyperparameter is calculated to obtain a sequence of local region of interest semantic relevance semantic weighted optimized feature vectors. Calculate the product of each region of interest multiscale semantic feature vector and the second hyperparameter in the multiple regions of interest multiscale semantic feature vectors to obtain a sequence of region of interest semantic weighted modulation representation vectors; The positions of each corresponding set of the sequence of semantically weighted optimized feature vectors of the local region of interest and the sequence of semantically weighted modulated representation vectors of the region of interest are summed to obtain the plurality of optimized multi-scale semantic feature vectors of the region of interest.

6. The method for planning and constructing a modern cemetery according to claim 5, characterized in that, The sequence of semantic row vectors of the plurality of local regions of interest is subjected to autocorrelation feature capture to obtain a sequence of autocorrelation coefficients of the plurality of local regions of interest, including: Calculate the product between each semantic row vector of the local region of interest and a preset weight feature vector in the sequence of semantic row vectors of the local region of interest to obtain a sequence of autocorrelation weight values ​​of the local region of interest; The sequence of autocorrelation weight values ​​of the local region of interest is subjected to nonlinear activation processing to obtain the sequence of autocorrelation coefficients of the local region of interest.

7. The modern cemetery construction planning method according to claim 6, characterized in that, Calculate the class support weight vector of each local region of interest autocorrelation coefficient row vector in the plurality of local regions of interest autocorrelation coefficient row vectors to obtain the semantic autocorrelation weight vectors of the plurality of local regions of interest, including: Each feature value in the autocorrelation coefficient row vector of the local region of interest is used as the exponent of the natural constant to calculate the position-based exponential function value with the natural constant as the base, so as to obtain the autocorrelation class support row vector of the local region of interest; The summation of the feature values ​​at each position in the support row vector of the autocorrelation class of the local region of interest is calculated to obtain the global autocorrelation metric coefficient of the local region of interest. The semantic autocorrelation weight vector of the region of interest is obtained by calculating the positional division between each feature value in the support row vector of the autocorrelation class of the region of interest and the global autocorrelation metric coefficient of the region of interest.

8. The method for planning and constructing a modern cemetery according to claim 7, characterized in that, Based on the multi-scale semantic feature vectors of the multiple optimized regions of interest, a recommended cemetery construction planning layout map is generated, including: arranging the multi-scale semantic feature vectors of the multiple optimized regions of interest into a global semantic feature matrix of the target region, and then using a cemetery construction planning layout map generator based on an adversarial generative network to obtain the recommended cemetery construction planning layout map.

9. A modern cemetery construction and planning system, characterized in that, include: The overhead image acquisition module is used to acquire multiple overhead images of the target area captured by the drone; An image stitching module is used to stitch together the multiple overhead images to obtain a panoramic overhead image of the target area. The region of interest feature extraction module is used to extract region of interest features from the panoramic aerial image of the target area to obtain multiple region of interest feature maps. The multi-scale semantic feature enhancement module is used to perform multi-scale semantic feature enhancement on the feature maps of the multiple regions of interest respectively to obtain multi-scale semantic feature vectors of the multiple regions of interest. The optimization module is used to input the multi-scale semantic feature vectors of the multiple regions of interest into the feature optimization module based on semantic association and semantic span to obtain multiple optimized multi-scale semantic feature vectors of regions of interest; The planning and recommendation module is used to generate a recommended cemetery construction planning layout map based on the multi-scale semantic feature vectors of the multiple optimized regions of interest.

10. The modern cemetery construction planning system according to claim 9, characterized in that, The region of interest feature extraction module is used to include: inputting the panoramic aerial image of the target area into a region of interest feature extractor based on an R-CNN model to obtain the multiple region of interest feature maps.

Images