Surface shape detection method and program product

Abstract

The invention discloses a surface shape detection method and a program product. The method comprises the following steps: continuously transmitting multiple types of detection waves to a target area, wherein the detection waves comprise communication waves and radar waves; aiming at each type of detection wave, respectively acquiring the echo intensity of the detection wave at a plurality of time points, and generating a target area imaging graph of the detection wave according to the echo intensity of the detection wave at the plurality of time points; and generating a surface form map of the target area according to the target area imaging maps of the multiple detection waves, wherein at least one subarea corresponding to the surface form of the preset type is marked in the surface form map. According to the technical scheme, through combination of various types of detection waves, the perception capability of diversified environments is effectively improved, and through combination of local features and global attention in the target area, features with higher discriminability are obtained, so that the recognition adaptability of complex scenes is improved.

Description

Technical Field

[0001] This invention relates to the field of artificial intelligence recognition technology, and in particular to a method and program product for detecting landforms. Background Technology

[0002] Surface morphology detection has important research significance, such as low-altitude UAV path planning, network planning and design, network operation and maintenance, and other related fields.

[0003] In related technologies, shape recognition is mainly achieved through radar perception. However, radar is susceptible to signal interference and the curvature of the earth, resulting in insufficient accuracy in terrain recognition. In addition, radar perception is difficult to apply to complex and ever-changing scenarios, and is prone to recognition confusion. Summary of the Invention

[0004] This invention provides a method and program for detecting landforms to solve the technical problem of inaccurate landform identification in related technologies.

[0005] According to one aspect of the present invention, a method for detecting landform features is provided, the method comprising:

[0006] Multiple types of detection waves, including communication waves and radar waves, are continuously emitted toward the target area;

[0007] For each type of probe wave, the echo intensity of the probe wave is collected at multiple time points, and an image map of the target area of ​​the probe wave is generated based on the echo intensity of the probe wave at multiple time points.

[0008] A surface morphology map of the target area is generated based on the target area imaging maps of various probe waves, and the surface morphology map is marked with sub-regions corresponding to at least one preset type of surface morphology.

[0009] According to another aspect of the present invention, a landform detection device is provided, the device comprising:

[0010] The area detection module is used to continuously transmit various types of detection waves to the target area, including communication waves and radar waves.

[0011] The data acquisition module is used to acquire the echo intensity of the probe wave at multiple time points for each type of probe wave, and generate a target area imaging map of the probe wave based on the echo intensity of the probe wave at multiple time points.

[0012] The surface morphology recognition module is used to generate a surface morphology map of the target area based on the target area imaging map of multiple probe waves, wherein the surface morphology map is marked with sub-regions corresponding to at least one preset type of surface morphology.

[0013] According to another aspect of the present invention, an electronic device is provided, the electronic device comprising:

[0014] At least one processor; and a memory communicatively connected to the at least one processor; wherein the memory stores a computer program executable by the at least one processor, the computer program being executed by the at least one processor to enable the at least one processor to perform a landform detection method according to any embodiment of the present invention.

[0015] According to another aspect of the present invention, a computer-readable storage medium is provided, the computer-readable storage medium storing computer instructions for causing a processor to execute and implement a landform detection method according to any embodiment of the present invention.

[0016] According to another aspect of the present invention, embodiments of the present disclosure also provide a computer program product, including a computer program that, when executed by a processor, implements a landform detection method as described in any of the embodiments of the present disclosure.

[0017] The technical solution of this invention involves continuously transmitting multiple types of probe waves, including communication waves and radar waves, towards a target area. This allows for the acquisition of rich target area feature data for subsequent accurate identification of various sites within the target area. Then, for each type of probe wave, the echo intensity is collected at multiple time points. Based on the echo intensity at these multiple time points, a target area imaging map is generated. Multiple target area imaging maps are constructed using the echo intensity, and by acquiring imaging maps from multiple times, the problem of inaccurate object identification in single-time-point imaging maps is reduced, thereby improving the accuracy of target object identification. Finally, a landform map of the target area is generated based on the target area imaging maps of the multiple probe waves. This landform map is marked with sub-regions corresponding to at least one preset type of landform. By comprehensively processing the target area imaging maps of multiple probe waves and focusing on the local and global features in the target area imaging during processing, the data correlation between various elements in the target image is effectively improved, thereby enhancing the recognition effect of subtle features and obtaining high-quality target area identification results. This technical solution acquires multi-time-lapse images of the target area and effectively combines the correlation between various local and global features during feature data processing. This allows the landform map to contain richer and more detailed object features, thereby improving the final recognition effect of each object.

[0018] It should be understood that the description in this section is not intended to identify key or essential features of the embodiments of the present invention, nor is it intended to limit the scope of the invention. Other features of the invention will become readily apparent from the following description. Attached Figure Description

[0019] To more clearly illustrate the technical solutions in the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0020] Figure 1 This is a flowchart of a landform detection method according to Embodiment 1 of the present invention;

[0021] Figure 2 This is a flowchart of a landform detection method according to Embodiment 2 of the present invention;

[0022] Figure 3A This is a flowchart of a landform detection method based on echo intensity data for acquiring a fused image according to a third embodiment of the present invention;

[0023] Figure 3B This is a flowchart illustrating the encoding and decoding operations in a landform detection method according to Embodiment 3 of the present invention.

[0024] Figure 4 This is a schematic diagram of the structure of a landform detection device according to Embodiment 4 of the present invention;

[0025] Figure 5 This is a schematic diagram of the structure of an electronic device that implements a landform detection method according to Embodiment 5 of the present invention. Detailed Implementation

[0026] To enable those skilled in the art to better understand the present invention, the technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments of the present invention. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort should fall within the scope of protection of the present invention.

[0027] It should be noted that the terms "instantaneous region," "target region," "local attention feature," "global attention feature," "first," "second," etc., used in the specification, claims, and accompanying drawings of this invention are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate so that embodiments of the invention described herein can be implemented in orders other than those illustrated or described herein. Furthermore, the terms "comprising" and "having," and any variations thereof, are intended to cover non-exclusive inclusion; for example, a process, method, system, product, or apparatus that comprises a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to such processes, methods, products, or apparatus.

[0028] It should be noted that the terms "a" and "a plurality of" used in this disclosure are illustrative rather than restrictive, and those skilled in the art should understand that, unless otherwise expressly indicated in the context, they should be understood as "one or more".

[0029] The names of messages or information exchanged between multiple devices in the embodiments of this disclosure are for illustrative purposes only and are not intended to limit the scope of such messages or information.

[0030] It is understood that before using the technical solutions disclosed in the various embodiments of this disclosure, users should be informed of the types, scope of use, and usage scenarios of the personal information involved in this disclosure in an appropriate manner in accordance with relevant laws and regulations, and user authorization should be obtained.

[0031] For example, upon receiving a user's active request, a prompt message is sent to the user to explicitly inform them that the requested operation will require the acquisition and use of the user's personal information. This allows the user to independently choose whether to provide personal information to the software or hardware, such as the electronic device, application, server, or storage medium performing the operations of this disclosed technical solution, based on the prompt message.

[0032] As an optional but non-limiting implementation, in response to a user's active request, sending a prompt message to the user can be done via a pop-up window, where the prompt message can be presented in text format. Furthermore, the pop-up window can also include a selection control allowing the user to choose "agree" or "disagree" to provide personal information to the electronic device.

[0033] It is understood that the above notification and user authorization process are merely illustrative and do not constitute a limitation on the implementation of this disclosure. Other methods that comply with relevant laws and regulations may also be applied to the implementation of this disclosure.

[0034] It is understood that the data involved in this technical solution (including but not limited to the data itself, the acquisition or use of the data) shall comply with the requirements of relevant laws, regulations and related provisions.

[0035] Example 1

[0036] Figure 1 This is a flowchart of a landform detection method provided in Embodiment 1 of the present invention. This embodiment is applicable to scenarios where the landform of a target area is identified, especially for landform identification using drones in low-altitude scenarios. The method can be executed by a landform detection device, which can be implemented in hardware and / or software, optionally through an electronic device such as a mobile terminal, PC, or server. Figure 1 As shown, the method may specifically include:

[0037] S110. Continuously transmit multiple types of detection waves toward the target area, including communication waves and radar waves.

[0038] The target area may include at least one of the following: grassland, river, road, wheat field, rice paddy, building, etc. In this technical solution, the communication wave specifically refers to microwaves corresponding to the 5G band.

[0039] Specifically, the drone carries signal transmitting equipment that continuously emits various types of signal probes into the target area to detect the area and collect a large number of data points for subsequent target area imaging. Detecting the target area using multiple probes improves the stability of target area detection and enhances the ability to perceive and identify complex and diverse environments within the target area.

[0040] S120. For each type of probe wave, the echo intensity of the probe wave is collected at multiple time points, and an image map of the target area of ​​the probe wave is generated based on the echo intensity of the probe wave at the multiple time points.

[0041] Echo intensity can be understood as the signal energy of different types of probe waves reflected back from the target area. Target area imaging can refer to an image of the target area constructed based on echo intensity data at multiple time points. For example, the display format of the target area imaging can include at least one of the following images: intensity map, interferogram, polarization map, point cloud map, grayscale image, etc.

[0042] Specifically, a sounding wave receiving device is used to collect sounding wave signals reflected back at multiple times, and the echo intensity of various types of sounding waves is detected. By normalizing the echo intensities of various types of sounding waves, an n x n x t convolutional image is obtained, which is used as the target region imaging map of the sounding wave. By constructing multiple sets of target region imaging maps using echo intensity data of the target region at multiple times, the blurring of target objects that easily occurs during UAV flight is reduced. Furthermore, by combining richer target object data features from multiple frames, the accuracy and robustness of target object identification are improved. Further, the target region imaging image can be mapped into a binarized image to facilitate simplified processing and convenient analysis of the target object.

[0043] In one embodiment, generating a target region imaging map of the probe wave based on the echo intensity of the probe wave at multiple time points includes: generating an instantaneous region imaging map of the probe wave based on the echo intensity collected by the probe wave at each time point, and fusing the instantaneous region imaging maps of the probe wave at multiple time points to obtain a target region imaging map of the probe wave.

[0044] Instantaneous regional imaging refers to the target area image generated based on the echo intensity data corresponding to the acquired probe wave for any probe wave.

[0045] Specifically, for a given time point, an instantaneous regional imaging map corresponding to the target probe wave type is calculated based on the echo intensity data of various probe waves collected at that time. By fusing multiple frames of instantaneous regional imaging maps corresponding to the target probe wave at a preset time point, a target region imaging map of the probe wave is obtained. By combining echo intensity data collected at multiple time points, the inaccuracy of constructing a target region imaging map from data at a single time point can be effectively compensated for, thereby improving the accuracy and reliability of target region imaging.

[0046] S130. Generate a surface morphology map of the target area based on the target area imaging map of the various probe waves, wherein the surface morphology map is marked with at least one sub-region corresponding to a preset type of surface morphology.

[0047] Specifically, the landform map includes at least one of the following images: heat map, layered color map, and value line map.

[0048] Specifically, feature extraction is performed on target area images from multiple probe waves to obtain abstract information corresponding to each probe wave. This abstract information is then fused to obtain an abstract feature map. Decoding this abstract feature map yields a surface morphology map of the target area. Processing target area images generated by different probe waves effectively enriches the data features within the target area, improving the ability to identify complex scenes.

[0049] The technical solution of this invention involves continuously transmitting multiple types of probe waves, including communication waves and radar waves, towards a target area. This allows for the acquisition of rich target area feature data for subsequent accurate identification of various sites within the target area. Then, for each type of probe wave, the echo intensity is collected at multiple time points. Based on the echo intensity at these multiple time points, a target area imaging map is generated. Multiple target area imaging maps are constructed using the echo intensity, and by acquiring imaging maps from multiple times, the problem of inaccurate object identification in single-time-point imaging maps is reduced, thereby improving the accuracy of target object identification. Finally, a landform map of the target area is generated based on the target area imaging maps of the multiple probe waves. This landform map is marked with sub-regions corresponding to at least one preset type of landform. By comprehensively processing the target area imaging maps of multiple probe waves and focusing on the local and global features in the target area imaging during processing, the data correlation between various elements in the target image is effectively improved, thereby enhancing the recognition effect of subtle features and obtaining high-quality target area identification results. This technical solution acquires multi-time-lapse images of the target area and effectively combines the correlation between various local and global features during feature data processing. This allows the landform map to contain richer and more detailed object features, thereby improving the final recognition effect of each object.

[0050] Example 2

[0051] Figure 2 This is a flowchart of a landform detection method according to Embodiment 2 of the present invention. This embodiment is a refinement of the technical solution for generating a landform map of the target area based on the target area imaging map of various probe waves, building upon the above embodiments. Detailed implementation methods can be found in the description of this embodiment. Technical features that are the same as or similar to those in the foregoing embodiments will not be repeated here. Figure 2 As shown, the method may specifically include:

[0052] S210. Continuously transmit multiple types of detection waves toward the target area, including communication waves and radar waves.

[0053] S220. For each type of probe wave, the echo intensity of the probe wave is collected at multiple time points, and an image map of the target area of ​​the probe wave is generated based on the echo intensity of the probe wave at the multiple time points.

[0054] S230. Encode the target region imaging map of various probe waves based on the relative attention mechanism to obtain an abstract feature map. The relative attention mechanism is used to extract and fuse the local attention features and global attention features of the image.

[0055] Here, the abstract feature map refers to the feature fusion image obtained by fusing feature maps corresponding to target area imaging maps from multiple probe waves. Local attention features refer to the feature results obtained by extracting features from elements in the first sub-region. Global attention features can be used to characterize the degree of correlation or similarity between a certain element or region and all elements in other regions.

[0056] Specifically, multiple probe wave target region images are input, and local and global attention features are extracted from these images using an attention mechanism. These extracted features are then fused to obtain a feature map corresponding to the target region image of the target type probe wave. By fusing the feature maps of multiple probe waves in chronological order, an abstract feature map is obtained. This technical solution, by combining local and global attention features from the target region image, effectively considers the feature relationships between various local features and between various local features and global features. Furthermore, by comprehensively processing the feature maps obtained from multiple probe waves, relatively subtle feature data in the target region can be determined, thereby improving the accuracy of identifying various target objects within the target region.

[0057] In one embodiment, encoding the target region imaging maps of multiple probe waves based on a relative attention mechanism to obtain an abstract feature map includes: determining an attention feature map of the target region imaging map for each of the probe waves based on the relative attention mechanism, and performing a convolution operation on the attention feature map to obtain the abstract feature map.

[0058] Among them, the attention feature map refers to the feature image obtained after extracting features from the target region imaging map using the relative attention mechanism.

[0059] Specifically, by employing a relative attention mechanism, features are extracted from the target region images of each probe wave, resulting in local and global attention features corresponding to multiple target region images. Then, by fusing these local and global attention features and calculating the feature data of the input sliding window according to a preset step size, an abstract feature map corresponding to the target region images of each probe wave is obtained. Through comprehensive analysis and processing of the global and local features in the target region images, the ability to identify objects with subtle differences can be effectively improved.

[0060] In another embodiment, determining the attention feature map of the target region image for each type of probe wave based on a relative attention mechanism includes: for each type of probe wave target region image, determining multiple first sub-regions and multiple second sub-regions associated with the first sub-regions in the target region image based on a preset sliding window; determining local attention features of the first sub-region based on multiple elements within the first sub-region based on a relative attention mechanism, and determining global attention features of the first sub-region based on the first sub-region and the multiple second sub-regions associated with the first sub-region; determining sub-region attention features of the first sub-region based on the local attention features and the global attention features of the first sub-region; and determining the attention feature map of the target region image of the probe wave based on the sub-region attention features of the multiple first sub-regions.

[0061] Here, the first sub-region can be understood as the image region where data processing occurs each time the window slides. The second sub-region can be understood as an image region that has feature associations with the first sub-region and is usually larger than the first sub-region. An element can be understood as data content containing characterizing image features. Sub-region attention features refer to the data results obtained by processing data within the first sub-region.

[0062] Specifically, for any target region image of a probe wave, multiple first sub-regions and multiple second sub-regions associated with the first sub-regions are determined using a preset sliding window. A relative attention mechanism is used to extract local attention features of multiple elements within the first sub-regions; and global attention features of the first sub-regions are determined by processing the data from the first and second sub-regions. Sub-region attention features of the first sub-region are then determined based on its local and global attention features. By sequentially calculating the sub-region attention features of multiple first sub-regions, the attention feature map corresponding to the target region image of the probe wave is determined.

[0063] In another embodiment, determining the global attention features of the first sub-region based on multiple elements of the first sub-region and multiple second sub-regions associated with the first sub-region includes: for each element in the first sub-region, determining multiple global key vectors corresponding to the element based on multiple elements of the first sub-region and multiple elements of multiple second sub-regions associated with the first sub-region; determining the global attention features of the element based on the similarity between the query vector corresponding to the element and the multiple global key vectors, wherein the element corresponds to at least one pixel; and determining the global attention features of the first sub-region based on the global attention features of the multiple elements in the first sub-region.

[0064] The global key vector can be understood as a collection of feature information of elements at various locations within the second sub-region. The query vector can be understood as a feature vector of the data content at the location of the target element.

[0065] Specifically, for each element in the first sub-region, multiple global key vectors corresponding to the target element are determined based on multiple elements in the first sub-region and multiple elements in the second sub-region. The similarity between the query vector corresponding to each element and the multiple global key vectors is then calculated to determine the global attention feature of the target element. Furthermore, by calculating the global attention features of multiple target element pairs in the first sub-region, the global attention feature corresponding to the first sub-region as a whole is determined. By accumulating the global attention features of multiple target elements, more accurate attention feature data representing the sub-region can be obtained, improving the accuracy of target object recognition in that sub-region.

[0066] In another embodiment, determining the sub-region attention features of the first sub-region based on the local attention features and the global attention features of the first sub-region includes: determining the average value of the local attention features and the global attention features of the first sub-region as the sub-region attention features of the first sub-region.

[0067] Specifically, the local attention features and global attention features of the first sub-region are summed, and the average value of the two is determined as the sub-region attention feature of the first sub-region.

[0068] S240. Decode the abstract feature map to obtain a landform map of the target area, wherein the landform map is marked with sub-regions corresponding to at least one preset type of landform.

[0069] Specifically, by decoding the abstract feature map and obtaining a landform map of the target size through mapping, the final landform map of the target area is obtained.

[0070] In one implementation, decoding the abstract feature map to obtain a landform map of the target region includes performing a deconvolution operation on the abstract feature map to obtain a landform map of the target region.

[0071] The deconvolution operation can be understood as a method to transform small, low-resolution feature maps into large, high-resolution feature maps.

[0072] Specifically, deconvolution is used to process the encoded abstract feature map. By inserting data into the abstract feature map, it is enlarged to the target size, thus obtaining the landform map of the target area. Through deconvolution, a smoother and higher-quality landform map of the target area can be obtained, accurately outputting sub-regions corresponding to various landforms.

[0073] The technical solution of this invention encodes the target area imaging images of multiple probe waves based on a relative attention mechanism to obtain an abstract feature map. The relative attention mechanism is used to extract and fuse local and global attention features of the image. This allows for the acquisition of richer feature content of the target object from the image features corresponding to multiple probe waves, resulting in high-quality target object data and improving subsequent target object recognition and processing. Subsequently, the abstract feature map is decoded to obtain a surface morphology map of the target area. This decoding process yields a surface morphology map of the target area at the target size, providing a more realistic and detailed representation of the target area's surface morphology, thereby improving the display effect of various sites within the target area.

[0074] Example 3

[0075] Embodiment 3 of the present invention provides a flowchart of a landform detection method. To better illustrate the technical solution provided by this embodiment, the following steps are used to illustrate the method. A portion of the flowchart of this embodiment is shown below. Figure 3A , Figure 3B As shown, specific implementation methods can be found in the description of this embodiment. Technical features that are the same as or similar to those in the foregoing embodiments will not be repeated here.

[0076] Step 1: Data Collection

[0077] By continuously transmitting 5G and radar waves to a designated target area using a drone, echo intensity data within the target area can be obtained.

[0078] Step 2: Data Processing

[0079] The system acquires echo intensity data within a preset time period, and performs normalization processing on data collected over three consecutive seconds to obtain multiple imaging images. These three consecutive images are then fused to obtain a 5G and radar sensing information matrix of size n×n×3 (n=1000), as shown below. Figure 3A The fused image is represented by the image result constructed based on echo intensity data at different times.

[0080] (1) The specific implementation of this process may also include: selecting an n×n detection space (the unit can be meters), normalizing the signal strength of the reflected wave (5G or radar) to a value between 0 and 1, and obtaining the following: Figure 3A The image in the image (in some processing, the image in the 3A can also be mapped to a black and white image of 0-255 visualization, i.e. a binarized image).

[0081] (2) By considering the jitter that often occurs during UAV flight, images from three consecutive moments can be fused to obtain, as shown below. Figure 3A The fusion of images can effectively improve the final image quality.

[0082] Step 3: Target Recognition Processing

[0083] (1) Encoding: First, features are extracted from the 5G and radar fusion maps based on a relative attention mechanism. Then, elements at corresponding positions in the feature maps are fused using a 1×1 convolution to obtain an abstract feature map containing semantic information. The process is as follows: Figure 3B As shown.

[0084] The encoding process specifically includes: both the 5G and radar fusion images are n×n×3 (n=1000) images. After calculating the fusion images using a relative attention mechanism, a feature map of (n-m+1)×(n-m+1)×3 is obtained (e.g., feature map). Figure 3B Features Figure 1 ,feature Figure 2 The feature map maintains the same input and output dimensions as the convolution-based approach; m is the window size, which can be set empirically; the stride of the sliding window is 1. Features Figure 1 , 2 After further 1×1 convolution and feature fusion along the time dimension, the final result is an abstract feature map (n-m+1)×(n-m+1) fused with joint features of 5G and radar, as shown below. Figure 3B The abstract feature map is shown in the figure.

[0085] Furthermore, to more clearly describe the encoding process, the following process will be used as an example:

[0086] Feature extraction preparation: Determine the sliding window parameters, set the window size to m (empirical setting), and the sliding step size to 1; for the input n×n×3 5G and radar imaging images, use the relative attention mechanism to extract features respectively.

[0087] Dual k-relative attention calculation:

[0088] 1. Global Attention Calculation: For each element in the window (e.g., the i-th element ai), obtain the query vector q through the preset parameter Wq, and obtain the global key vector k1 through the preset parameter Wq (applicable to the global n×n range); calculate the similarity between the query vector q and each global key vector k1, and then obtain the global attention weight a_ji through exponential operation and normalization; combine the value vector to calculate the global attention output b_i.

[0089] 2. Local attention calculation: The local key vector k2 is obtained through the parameter Wk2 (applicable to a local m×m window range). The similarity between the query vector q and each global key vector k2 is calculated. Then, the global attention weight a_ji is obtained through exponential operation and normalization. The local attention output c_i is obtained by combining the value vector.

[0090] 3. Attention Fusion: Calculate the average of the global attention output b_i and the local attention output c_j for each window element, i.e., d_i = (b_i + c_j) / 2, to obtain the final feature of a single window element; then slide the window with a step size of 1, calculating the feature of each window sequentially, thus obtaining a (n-m+1)×(n-m+1)×3 feature map after processing based on the 5G and radar fusion map (e.g., feature map). Figure 3B Features Figure 1 ,feature Figure 2 ).

[0091] (2) Decoding: The abstract feature map is mapped into a heat map (n×n) using deconvolution technology. In the heat map, 0 can be used to represent wheat field, 1 can be used to represent rice field, 2 can be used to represent grassland, etc.

[0092] This technical solution integrates rich data features from radar and 5G across multiple timeframes. By exploring the relationships between elements input into a sliding window while simultaneously paying attention to the relationships between each element and global elements (i.e., employing a dual-k relative attention mechanism), it can obtain more subtle features in the target image. This helps improve the robustness and accuracy of target object image processing and recognition, thereby enhancing the ability of the perception system in the UAV to adapt to diverse environments.

[0093] Example 4

[0094] Figure 4This is a schematic diagram of a landform detection device provided in Embodiment 4 of the present invention. Figure 4 As shown, the device includes: a region detection module 401, a data acquisition module 402, and a landform recognition module 403.

[0095] The system includes a region detection module 401, which continuously transmits multiple types of probe waves to a target region, including communication waves and radar waves; a data acquisition module 402, which collects the echo intensity of each type of probe wave at multiple time points and generates a target region imaging map based on the echo intensity of the probe wave at the multiple time points; and a landform recognition module 403, which generates a landform map of the target region based on the target region imaging map of the multiple types of probe waves, wherein the landform map is marked with at least one sub-region corresponding to a preset type of landform.

[0096] In this embodiment of the invention, the area detection module 401 continuously emits various types of probe waves, including communication waves and radar waves, towards the target area to acquire rich target area feature data for subsequent accurate identification of various sites within the target area. Then, the data acquisition module 402 collects the echo intensity of each type of probe wave at multiple time points and generates a target area imaging map based on the echo intensity at these multiple time points. Multiple target area imaging maps are constructed using the echo intensity, and by acquiring imaging maps from multiple times, the problem of inaccurate object identification in a single-time-time imaging map is reduced, thereby improving the accuracy of target object identification. Finally, the surface morphology recognition module 403 generates a surface morphology map of the target area based on the target area imaging maps of the various probe waves. The surface morphology map is marked with sub-regions corresponding to at least one preset type of surface morphology. By comprehensively processing target area images from multiple probe waves and focusing on both local and global features during the processing, this approach effectively enhances the data correlation between various elements in the target image, thereby improving the recognition of subtle features and achieving high-quality target area identification results. This technical solution acquires target area images at multiple time points and effectively combines the correlation between various local and global features during feature data processing. This allows the surface morphology map to contain richer and more detailed object features, ultimately improving the identification effect of each object.

[0097] Based on the above-mentioned optional technical solutions, the data acquisition module 402 may optionally include a target area imaging unit. The target area imaging unit is used to generate instantaneous area imaging maps of the probe wave based on the echo intensity collected at each of the time points, and to fuse the instantaneous area imaging maps of the probe wave at multiple time points to obtain a target area imaging map of the probe wave.

[0098] Based on the above-mentioned optional technical solutions, the landform recognition module 403 may optionally include: an abstract feature acquisition unit and a landform acquisition unit. The abstract feature acquisition unit is used to encode the target area imaging images of multiple probe waves based on a relative attention mechanism to obtain an abstract feature map; the relative attention mechanism is used to extract and fuse local attention features and global attention features of the image. The landform acquisition unit is used to decode the abstract feature map to obtain a landform map of the target area.

[0099] Based on the above-mentioned optional technical solutions, the abstract feature acquisition unit may optionally include an attention feature processing unit. The attention feature processing unit is configured to determine attention feature maps of the target region imaging map for each type of probe wave based on a relative attention mechanism, and perform convolution operations on the attention feature maps to obtain the abstract feature map.

[0100] Based on the above-mentioned optional technical solutions, the attention feature processing unit may optionally include: a sub-region determination unit, a sub-region attention feature acquisition unit, and an attention feature map acquisition unit. Specifically, the sub-region determination unit is used to determine, for each type of probe wave's target region imaging map, multiple first sub-regions and multiple second sub-regions associated with the first sub-regions in the target region imaging map based on a preset sliding window; the sub-region attention feature acquisition unit is used to determine, based on a relative attention mechanism, the local attention features of the first sub-region according to multiple elements within the first sub-region, and the global attention features of the first sub-region according to the first sub-region and the multiple second sub-regions associated with the first sub-region, and to determine the sub-region attention features of the first sub-region according to the local attention features and the global attention features of the first sub-region; the attention feature map acquisition unit is used to determine the attention feature map of the target region imaging map of the probe wave based on the sub-region attention features of the multiple first sub-regions.

[0101] Based on the above-mentioned optional technical solutions, the sub-region attention feature acquisition unit may optionally include: a similarity calculation unit and a global attention feature determination unit. The similarity calculation unit is configured to, for each element in the first sub-region, determine multiple global key vectors corresponding to the element based on multiple elements of the first sub-region and multiple elements of multiple second sub-regions associated with the first sub-region, and determine the global attention feature of the element based on the similarity between the query vector corresponding to the element and the multiple global key vectors, wherein the element corresponds to at least one pixel. The global attention feature determination unit is configured to determine the global attention feature of the first sub-region based on the global attention features of the multiple elements in the first sub-region.

[0102] Based on the above-mentioned optional technical solutions, the attention feature map acquisition unit may optionally include: a sub-region attention feature determination unit. The sub-region attention feature determination unit is used to determine the average value of the local attention features and the global attention features of the first sub-region as the sub-region attention feature of the first sub-region.

[0103] Based on the above-mentioned optional technical solutions, the landform acquisition unit may optionally include a deconvolution unit. The deconvolution unit is used to perform a deconvolution operation on the abstract feature map to obtain a landform map of the target area.

[0104] Based on the above-mentioned optional technical solutions, the landform map may optionally include at least one of the following images: heat map, layered color map, and value line map.

[0105] The landform detection device provided in this embodiment of the invention can execute a landform detection method provided in any embodiment of the invention, and has the corresponding functional modules and beneficial effects for executing a landform detection method. Technical details not described in detail in this embodiment can be found in any of the landform detection methods described in this embodiment of the invention.

[0106] Example 5

[0107] Figure 5A schematic diagram of an electronic device 10, which can be used to implement embodiments of the present invention, is shown. The electronic device is intended to represent various forms of digital computers, such as laptop computers, desktop computers, workstations, personal digital assistants, servers, blade servers, mainframe computers, and other suitable computers. The electronic device can also represent various forms of mobile devices, such as personal digital processors, cellular phones, smartphones, wearable devices (e.g., helmets, glasses, watches, etc.), and other similar computing devices. The components shown herein, their connections and relationships, and their functions are merely illustrative and are not intended to limit the implementation of the invention described and / or claimed herein.

[0108] like Figure 5 As shown, the electronic device 10 includes at least one processor 11 and a memory, such as a read-only memory (ROM) 12 or a random access memory (RAM) 13, communicatively connected to the at least one processor 11. The memory stores computer programs executable by the at least one processor. The processor 11 can perform various appropriate actions and processes based on the computer program stored in the ROM 12 or loaded from storage unit 18 into the RAM 13. The RAM 13 can also store various programs and data required for the operation of the electronic device 10. The processor 11, ROM 12, and RAM 13 are interconnected via a bus 14. An input / output (I / O) interface 15 is also connected to the bus 14.

[0109] Multiple components in electronic device 10 are connected to I / O interface 15, including: input unit 16, such as keyboard, mouse, etc.; output unit 17, such as various types of displays, speakers, etc.; storage unit 18, such as disk, optical disk, etc.; and communication unit 19, such as network card, modem, wireless transceiver, etc. Communication unit 19 allows electronic device 10 to exchange information / data with other devices through computer networks such as the Internet and / or various telecommunications networks.

[0110] Processor 11 can be a variety of general-purpose and / or special-purpose processing components with processing and computing capabilities. Some examples of processor 11 include, but are not limited to, a central processing unit (CPU), a graphics processing unit (GPU), various special-purpose artificial intelligence (AI) computing chips, various processors running machine learning model algorithms, a digital signal processor (DSP), and any suitable processor, controller, microcontroller, etc. Processor 11 performs the various methods and processes described above, such as a landform detection method.

[0111] In some embodiments, a landform detection method may be implemented as a computer program tangibly contained in a computer-readable storage medium, such as storage unit 18. In some embodiments, part or all of the computer program may be loaded and / or installed on electronic device 10 via ROM 12 and / or communication unit 19. When the computer program is loaded into RAM 13 and executed by processor 11, one or more steps of the landform detection method described above may be performed. Alternatively, in other embodiments, processor 11 may be configured to perform a landform detection method by any other suitable means (e.g., by means of firmware).

[0112] Various embodiments of the systems and techniques described above herein can be implemented in digital electronic circuit systems, integrated circuit systems, field-programmable gate arrays (FPGAs), application-specific integrated circuits (ASICs), application-specific standard products (ASSPs), systems-on-a-chip (SoCs), payload-programmable logic devices (CPLDs), computer hardware, firmware, software, and / or combinations thereof. These various embodiments may include implementations in one or more computer programs that can be executed and / or interpreted on a programmable system including at least one programmable processor, which may be a dedicated or general-purpose programmable processor, capable of receiving data and instructions from a storage system, at least one input device, and at least one output device, and transmitting data and instructions to the storage system, the at least one input device, and the at least one output device.

[0113] Computer programs used to implement the methods of the present invention may be written in any combination of one or more programming languages. These computer programs may be provided to a processor of a general-purpose computer, a special-purpose computer, or other programmable data processing device, such that when executed by the processor, the computer programs cause the functions / operations specified in the flowcharts and / or block diagrams to be performed. The computer programs may be executed entirely on a machine, partially on a machine, or as a standalone software package, partially on a machine and partially on a remote machine, or entirely on a remote machine or server.

[0114] In the context of this invention, a computer-readable storage medium can be a tangible medium that may contain or store a computer program for use by or in conjunction with an instruction execution system, apparatus, or device. A computer-readable storage medium may include, but is not limited to, electronic, magnetic, optical, electromagnetic, infrared, or semiconductor systems, apparatus, or devices, or any suitable combination thereof. Alternatively, a computer-readable storage medium may be a machine-readable signal medium. More specific examples of machine-readable storage media include electrical connections based on one or more wires, portable computer disks, hard disks, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fibers, portable compact disk read-only memory (CD-ROM), optical storage devices, magnetic storage devices, or any suitable combination thereof.

[0115] To provide interaction with a user, the systems and techniques described herein can be implemented on an electronic device having: a display device (e.g., a CRT (cathode ray tube) or LCD (liquid crystal display) monitor) for displaying information to the user; and a keyboard and pointing device (e.g., a mouse or trackball) through which the user provides input to the electronic device. Other types of devices can also be used to provide interaction with the user; for example, feedback provided to the user can be any form of sensory feedback (e.g., visual feedback, auditory feedback, or tactile feedback); and input from the user can be received in any form (including sound input, voice input, or tactile input).

[0116] The systems and technologies described herein can be implemented in computing systems that include backend components (e.g., as data servers), or middleware components (e.g., application servers), or frontend components (e.g., user computers with graphical user interfaces or web browsers through which users can interact with implementations of the systems and technologies described herein), or any combination of such backend, middleware, or frontend components. The components of the system can be interconnected via digital data communication of any form or medium (e.g., communication networks). Examples of communication networks include local area networks (LANs), wide area networks (WANs), blockchain networks, and the Internet.

[0117] A computing system can include clients and servers. Clients and servers are generally located far apart and typically interact through communication networks. The client-server relationship is created by computer programs running on the respective computers and having a client-server relationship with each other. The server can be a cloud server, also known as a cloud computing server or cloud host, which is a hosting product within the cloud computing service system to address the shortcomings of traditional physical hosts and VPS services, such as high management difficulty and weak business scalability.

[0118] In particular, according to embodiments of the present invention, the processes described above with reference to the flowcharts can be implemented as computer software programs. For example, embodiments of the present invention include a computer program product comprising a computer program carried on a non-transitory computer-readable medium, the computer program containing program code for performing the methods shown in the flowcharts. In such embodiments, the computer program can be downloaded and installed from a network via communication unit 19, or installed from storage unit 18, or installed from ROM 12. When the computer program is executed by processor 11, it performs the functions defined in the methods of the embodiments of the present invention.

[0119] It should be understood that the various forms of processes shown above can be used, with steps reordered, added, or deleted. For example, the steps described in this invention can be executed in parallel, sequentially, or in different orders, as long as the desired result of the technical solution of this invention can be achieved, and this is not limited herein.

[0120] The specific embodiments described above do not constitute a limitation on the scope of protection of this invention. Those skilled in the art should understand that various modifications, combinations, sub-combinations, and substitutions can be made according to design requirements and other factors. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of this invention should be included within the scope of protection of this invention.

Claims

1. A method for detecting landforms, characterized in that, include: Multiple types of detection waves, including communication waves and radar waves, are continuously emitted toward the target area; For each type of probe wave, the echo intensity of the probe wave is collected at multiple time points, and an image map of the target area of ​​the probe wave is generated based on the echo intensity of the probe wave at multiple time points. A surface morphology map of the target area is generated based on the target area imaging maps of various probe waves, and the surface morphology map is marked with sub-regions corresponding to at least one preset type of surface morphology.

2. The landform detection method according to claim 1, characterized in that, The step of generating a target area image map of the probe wave based on the echo intensity of the probe wave at multiple time points includes: An instantaneous regional imaging map of the probe wave is generated based on the echo intensity collected at each time point. The instantaneous regional imaging maps of the probe wave at multiple time points are then fused to obtain a target region imaging map of the probe wave.

3. The landform detection method according to claim 1, characterized in that, The step of generating a surface morphology map of the target area based on the target area imaging maps of multiple probe waves includes: The target region imaging maps of various probe waves are encoded based on a relative attention mechanism to obtain an abstract feature map. The relative attention mechanism is used to extract and fuse the local attention features and global attention features of the image. The abstract feature map is decoded to obtain a surface morphology map of the target area.

4. The landform detection method according to claim 3, characterized in that, The encoding of the target region imaging map based on the relative attention mechanism to obtain an abstract feature map includes: Based on the relative attention mechanism, attention feature maps of the target region imaging map for each of the probe waves are determined respectively, and convolution operations are performed on the attention feature maps to obtain the abstract feature map.

5. The landform detection method according to claim 4, characterized in that, The step of determining the attention feature map of the target region imaging map for each type of probe wave based on the relative attention mechanism includes: For each type of probe wave, a target region image is formed, and a plurality of first sub-regions and a plurality of second sub-regions associated with the first sub-regions are determined based on a preset sliding window. Based on the relative attention mechanism, the local attention features of the first sub-region are determined according to multiple elements in the first sub-region, and the global attention features of the first sub-region are determined according to the first sub-region and multiple second sub-regions associated with the first sub-region. The sub-region attention features of the first sub-region are determined according to the local attention features and the global attention features of the first sub-region. The attention feature map of the target region imaging map of the probe wave is determined based on the sub-region attention features of multiple first sub-regions.

6. The landform detection method according to claim 5, characterized in that, The step of determining the global attention features of the first sub-region based on multiple elements of the first sub-region and multiple second sub-regions associated with the first sub-region includes: For each element in the first sub-region, multiple global key vectors corresponding to the element are determined based on multiple elements in the first sub-region and multiple elements in multiple second sub-regions associated with the first sub-region. The global attention feature of the element is determined based on the similarity between the query vector corresponding to the element and the multiple global key vectors. The element corresponds to at least one pixel. The global attention features of the first sub-region are determined based on the global attention features of multiple elements in the first sub-region.

7. The landform detection method according to claim 5, characterized in that, Determining the sub-region attention features of the first sub-region based on the local attention features and the global attention features of the first sub-region includes: The average value of the local attention features and the global attention features of the first sub-region is determined as the sub-region attention feature of the first sub-region.

8. The landform detection method according to claim 3, characterized in that, Decoding the abstract feature map to obtain the surface morphology map of the target area includes: Perform a deconvolution operation on the abstract feature map to obtain a surface morphology map of the target region.

9. The landform detection method according to claim 1, characterized in that, The landform map includes a heat map.

10. A computer program product, comprising a computer program, characterized in that, When the computer program is executed by the processor, it implements the landform detection method as described in any one of claims 1-9.

Images