Village planning real scene three-dimensional data acquisition method, device and equipment and storage medium

Abstract

The present application relates to village planning space information processing technical field, disclose a kind of village planning real scene three-dimensional data acquisition method, device, equipment and storage medium, by establishing village rough three-dimensional scene and dividing planning observation unit, make data evaluation process by overall coverage determination turn to local planning expression determination, introduce planning interaction record and respectively calculate initial collection completeness, planning importance and required collection completeness, so that system can identify lane facade, courtyard edge, eaves lower interface and public space boundary and other local space area that influence planning expression effect, by calculating reinforcement requirement value and generating supplementary collection record table, so that supplementary collection process is transformed into differentiating reinforcement processing for key planning observation unit, finally by integrating initial oblique image and supplementary collection image, realize local depth estimation and real scene three-dimensional fusion, improve the three-dimensional expression quality of key area such as narrow lane, courtyard boundary and low-texture facade.

Description

Technical Field

[0001] This invention relates to the field of village planning spatial information processing technology, and in particular to a method, apparatus, equipment and storage medium for acquiring real-world three-dimensional data of village planning. Background Technology

[0002] In the process of village planning, it is usually necessary to investigate, represent, and analyze village buildings, alleyways, courtyard boundaries, public spaces, and relationships between land features. As the focus of village planning has gradually shifted from basic status quo surveys to refined planning, landscape improvement, and renovation of courtyard public spaces, with greater emphasis on public participation and results presentation, traditional methods relying on two-dimensional maps, on-site photographs, or uniformly routed oblique photography are no longer sufficient to meet the application needs of representing key local spaces.

[0003] Current methods for acquiring 3D village images primarily rely on oblique drone photography. These operations typically follow a standardized flight path with consistent altitude and overlap, and a 3D model is built based on the acquired images. This approach achieves good overall coverage in typical village settings with sparsely distributed buildings and open streetscapes. However, in traditional village clusters undergoing courtyard and alleyway renovations, challenges arise. These scenarios often involve narrow, continuous alleyways, protruding eaves, sheds, semi-enclosed walls, and difficulty in direct access to some courtyard edges. While the standardized flight path method effectively acquires data on rooftops and areas above main roads, it falls short in capturing data on critical areas representing planning, such as alleyway facades, courtyard edges, rooftop surfaces, and public space boundaries.

[0004] Furthermore, in the practical application of village planning, planners pay more attention to local spatial relationships that villagers can directly understand. For example, in the task of alleyway appearance improvement, the continuity of street facades, wall boundary relationships, and spatial connections at alleyway corners are more critical; in the task of courtyard public space renovation, the enclosure relationship between courtyard entrances, wall edges, surrounding building interfaces, and accessible open spaces is more critical. Existing technologies typically lack the technical means to link planning interaction needs with the data acquisition and processing process, resulting in the system's inability to dynamically adjust and supplement data acquisition and subsequent 3D reconstruction processing for the spatial areas truly relevant to the planning.

[0005] Therefore, how to effectively reinforce key local spaces in village planning and generate more suitable real-world 3D results for village planning expression and analysis remains a pressing technical problem in the field of real-world 3D data acquisition for village planning. Summary of the Invention

[0006] This invention provides a method, apparatus, equipment, and storage medium for acquiring real-world 3D data of village planning, which at least solves the problem that existing unified route acquisition methods are difficult to effectively represent key local spaces in the scenarios of courtyard public space renovation and alleyway landscape improvement in traditional settlement-type villages.

[0007] To achieve the above objectives, a first aspect of the present invention provides a method for acquiring real-world 3D data of village planning, comprising the following steps: Acquire an initial tilted image set of the village and an initial flight pose record corresponding to the initial tilted image set, perform coarse 3D reconstruction processing to generate a coarse 3D scene of the village, and divide the village into a set of planning observation units based on the coarse 3D scene of the village. Obtain planning interaction records based on the rough 3D scene of the village, and calculate the initial collection completeness, planning importance, and required collection completeness corresponding to each planning observation unit; Based on the initial data collection completeness, planning importance, and required data collection completeness corresponding to each planning observation unit, the reinforcement requirement value corresponding to each planning observation unit is calculated, and a supplementary data collection record table is generated based on the reinforcement requirement value. Receive the supplementary acquisition image set returned according to the supplementary acquisition record table and the supplementary flight pose record corresponding to the supplementary acquisition image set, and integrate the initial tilt image set and the supplementary acquisition image set into a complete image set, and integrate the initial flight pose record and the supplementary flight pose record into a complete pose record; Local depth estimation and 3D fusion processing are performed based on a complete image set, complete pose records, and a set of planning observation units to generate a real-world 3D model of the village plan, a planning representation quality map, and local spatial description data.

[0008] Furthermore, to achieve the above objectives, a second aspect of the present invention provides a village planning real-scene three-dimensional data acquisition device, comprising: The segmentation module is used to acquire the initial tilted image set of the village and the initial flight pose record corresponding to the initial tilted image set, perform coarse 3D reconstruction processing to generate a coarse 3D scene of the village, and divide the planning observation unit set based on the coarse 3D scene of the village. The calculation module is used to acquire planning interaction records based on the rough three-dimensional scene of the village, and to calculate the initial collection completeness, planning importance, and required collection completeness corresponding to each planning observation unit. The generation module is used to calculate the reinforcement requirement value corresponding to each planning observation unit based on the initial collection completeness, planning importance and required collection completeness of each planning observation unit, and generate a supplementary collection record table based on the reinforcement requirement value. The integration module is used to receive the supplementary acquisition image set returned according to the supplementary acquisition record table and the supplementary flight pose record corresponding to the supplementary acquisition image set, and to integrate the initial tilt image set and the supplementary acquisition image set into a complete image set, and to integrate the initial flight pose record and the supplementary flight pose record into a complete pose record. The fusion module is used to perform local depth estimation and 3D fusion processing based on the complete image set, complete pose record, and planning observation unit set to generate a real-world 3D model of village planning, a planning representation quality map, and local spatial description data.

[0009] Furthermore, in order to achieve the above objectives, a third aspect of the present invention provides a village planning real-scene 3D data acquisition device, the village planning real-scene 3D data acquisition device comprising: a memory, a processor, and a village planning real-scene 3D data acquisition program stored in the memory and executable on the processor, wherein when the village planning real-scene 3D data acquisition program is executed by the processor, it implements the steps of the village planning real-scene 3D data acquisition method as described above.

[0010] Furthermore, in order to achieve the above objectives, a fourth aspect of the present invention provides a storage medium storing a village planning real-scene 3D data acquisition program, which, when executed by a processor, implements the steps of the village planning real-scene 3D data acquisition method described above.

[0011] The beneficial effects of this invention are as follows: It proposes a method, device, equipment, and storage medium for real-scene 3D data acquisition of village planning. By first establishing a rough 3D scene of the village and dividing it into planning observation units after acquisition, the data evaluation process shifts from overall coverage judgment to local planning expression judgment. By introducing planning interaction records and calculating the initial acquisition completeness, planning importance, and required acquisition completeness, the system can identify local spatial areas that truly affect the planning expression effect, such as alleyway facades, courtyard edges, eaves interfaces, and public space boundaries. Then, by calculating reinforcement requirement values ​​and generating a supplementary acquisition record table, the supplementary acquisition process is no longer a repetitive acquisition of the entire village, but is transformed into differentiated reinforcement processing for key planning observation units. Finally, by integrating the initial tilted image and the supplementary acquisition image, and comprehensively considering texture differences, edge direction differences, occlusion relationships, and planning importance in subsequent local depth estimation and 3D fusion processing, the 3D expression quality of key areas such as narrow alleys, courtyard boundaries, and low-texture facades can be improved, thereby forming real-scene 3D results that are more suitable for village planning research, landscape improvement analysis, and public participation display. Attached Figure Description

[0012] Figure 1 This is a schematic diagram of the device structure of the hardware operating environment involved in the embodiments of the present invention; Figure 2 This is a flowchart illustrating an embodiment of the village planning real-scene 3D data acquisition method of the present invention; Figure 3 This is a structural block diagram of a village planning real-scene 3D data acquisition device according to an embodiment of the present invention. Detailed Implementation

[0013] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the invention.

[0014] like Figure 1 As shown, Figure 1 This is a schematic diagram of the device structure of the hardware operating environment involved in the embodiments of the present invention.

[0015] like Figure 1 As shown, the device may include: a processor 1001, such as a CPU; a communication bus 1002; a user interface 1003; a network interface 1004; and a memory 1005. The communication bus 1002 is used to enable communication between these components. The user interface 1003 may include a display screen or an input unit such as a keyboard; optionally, the user interface 1003 may also include a standard wired interface or a wireless interface. The network interface 1004 may optionally include a standard wired interface or a wireless interface (such as a Wi-Fi interface). The memory 1005 may be high-speed RAM or non-volatile memory, such as a disk drive. Optionally, the memory 1005 may also be a storage device independent of the aforementioned processor 1001.

[0016] Those skilled in the art will understand that Figure 1 The structure of the device shown does not constitute a limitation on the device and may include more or fewer components than shown, or combine certain components, or have different component arrangements.

[0017] like Figure 1 As shown, the memory 1005, which serves as a computer storage medium, may include an operating system, a network communication module, a user interface module, and a village planning real-scene 3D data acquisition program.

[0018] exist Figure 1In the terminal shown, network interface 1004 is mainly used to connect to the backend server and communicate with it; user interface 1003 is mainly used to connect to the client (user terminal) and communicate with it; while processor 1001 can be used to call the village planning real-scene 3D data acquisition program stored in memory 1005 and perform the following operations: Acquire an initial tilted image set of the village and an initial flight pose record corresponding to the initial tilted image set, perform coarse 3D reconstruction processing to generate a coarse 3D scene of the village, and divide the village into a set of planning observation units based on the coarse 3D scene of the village. Obtain planning interaction records based on the rough 3D scene of the village, and calculate the initial collection completeness, planning importance, and required collection completeness corresponding to each planning observation unit; Based on the initial data collection completeness, planning importance, and required data collection completeness corresponding to each planning observation unit, the reinforcement requirement value corresponding to each planning observation unit is calculated, and a supplementary data collection record table is generated based on the reinforcement requirement value. Receive the supplementary acquisition image set returned according to the supplementary acquisition record table and the supplementary flight pose record corresponding to the supplementary acquisition image set, and integrate the initial tilt image set and the supplementary acquisition image set into a complete image set, and integrate the initial flight pose record and the supplementary flight pose record into a complete pose record; Local depth estimation and 3D fusion processing are performed based on a complete image set, complete pose records, and a set of planning observation units to generate a real-world 3D model of the village plan, a planning representation quality map, and local spatial description data.

[0019] The specific embodiments of the present invention applied to the device are basically the same as the embodiments of the following methods for acquiring real-world 3D data of village planning, and will not be described in detail here.

[0020] This invention provides a method for acquiring real-world 3D data of village planning, referring to... Figure 2 , Figure 2 This is a flowchart illustrating an embodiment of the village planning real-scene 3D data acquisition method of the present invention.

[0021] In this embodiment, a method for acquiring real-world 3D data of village planning includes the following steps: S10: Obtain the initial tilted image set of the village and the initial flight pose record corresponding to the initial tilted image set, perform coarse 3D reconstruction processing to generate a coarse 3D scene of the village, and divide the planning observation unit set based on the coarse 3D scene of the village.

[0022] Specifically, a rough 3D scene of the village is established based on the initial oblique image set and the initial flight pose record; according to the spatial surface orientation, local elevation difference relationship and spatial continuity relationship in the rough 3D scene of the village, the village space is divided into multiple planning observation units to form a planning observation unit set; for each planning observation unit and each initial oblique image, the visibility quality score of the planning observation unit in the corresponding initial oblique image is calculated; the visibility quality score is stored as input data for subsequent calculation of the initial acquisition integrity.

[0023] In this embodiment of the invention, step S10 is used to acquire an initial tilted image set of the village and its corresponding initial flight pose records, and to establish a rough 3D scene based on the initial tilted image set and the initial flight pose records, so that subsequent processing steps have a unified spatial reference basis. The initial tilted image set is used to provide multi-angle image information of village buildings, alleys, courtyards, walls, and public spaces, and the initial flight pose records are used to characterize the spatial position and attitude relationships of each image at the time of acquisition, so that the image data can be organized and interpreted under the same reference system. It is easy to understand that if a unified modeling and evaluation is carried out directly around the whole village, although a relatively complete overall model can be obtained in form, it is difficult to identify which areas truly meet the planning expression requirements and which areas are only superficially photographed but do not have the conditions to further support planning judgments. Therefore, the rough 3D scene generated first in this embodiment of the invention is not directly used as the final planning result, but as the data basis for subsequent detailed analysis and reinforcement processing.

[0024] In practical applications, the rough 3D scene should at least reflect the relationships between the main building volumes, alleyway connections, courtyard boundaries, and the surrounding interfaces of public spaces within the village. Specifically, the rough 3D scene can be obtained by registering and reconstructing the initial oblique image with the initial flight pose record. The result is not required to reach the final display accuracy at this stage, but it needs to support spatial unit division, planning interactive mapping, and subsequent supplementary data acquisition.

[0025] It should be noted that the principle of establishing a rough 3D scene of the village based on the initial oblique image set and the initial flight pose record lies in utilizing the repeated observation relationship of the same spatial area of ​​the village from multiple perspectives, combined with the spatial position and attitude relationships of each image at the time of acquisition, to initially reconstruct the spatial distribution of building rooftops, building facades, alleyway interfaces, courtyard boundaries, and public space edges in the village. In simple terms, the initial oblique image set provides multi-angle observations of the village's real space, while the initial flight pose record provides the spatial arrangement relationship between these observations. Only by combining the two can the system determine whether the same ground features appearing in different images belong to the same spatial location, and further determine the relative geometric relationships between these spatial locations.

[0026] In one feasible implementation, the system first establishes initial spatial relationships between images based on initial flight pose records. Then, by analyzing the recurring village interface content in different images, it gradually determines the approximate location distribution of building surfaces, alleyway boundaries, and courtyard interfaces under a unified spatial coordinate system, thus forming a rough 3D scene. This rough 3D scene does not require precise restoration of all details at this stage; rather, it focuses on ensuring that the relative positional relationships between the main spatial interfaces in the village are correctly expressed. This allows subsequent planning observation unit division, planning interaction record mapping, and supplementary data acquisition generation to all be built upon a consistent spatial framework. This approach provides a basic scene with an overall spatial structure that still allows for subsequent focused enhancement, thus providing a stable starting point for subsequent steps.

[0027] Furthermore, the spatial surfaces in the rough 3D scene can be decomposed according to surface orientation, local elevation differences, and spatial continuity to form multiple planning observation units. These planning observation units are not limited to a single fixed geometric shape; their essence is a discretized representation of local spatial interfaces within the village relevant to the planning expression. For example, a street-facing facade, the edge of a courtyard entrance, the outer interface of a wall, a corner facade of an alleyway, and the boundary interface of a public space can all be constructed as corresponding planning observation units. In this way, all subsequent evaluations, comparisons, and reinforcement actions can be carried out around the planning observation units, rather than being treated indiscriminately across the entire village.

[0028] Furthermore, in this embodiment of the invention, the visibility of each planned observation unit in each initial oblique image is analyzed, and the visibility quality score of the planned observation unit in the corresponding image is calculated. The visibility quality score satisfies the following formula:

[0029] in, Indicates the first The planning observation unit in the first Visible quality score in Zhang's initial oblique image; This represents the projection area factor, which characterizes whether the imaging area of ​​the planned observation unit in the corresponding image reaches the effective analysis range. The viewpoint adaptation factor is used to characterize the degree of adaptation between the image acquisition direction and the surface direction of the planned observation unit. The sharpness factor is used to characterize the resolvability of the corresponding image region. The occlusion impact factor is used to characterize the degree of occlusion impact of eaves, sheds, trees, fences, or adjacent buildings on the planning observation unit. It should be noted that the visibility quality score here does not simply determine whether an image was captured, but rather further determines whether it was captured clearly, effectively, and appropriately for the planning expression, thereby distinguishing between areas with general coverage and key areas with insufficient data collection.

[0030] In one specific implementation, a traditional settlement-type village can be selected as the implementation object. The village has narrow alleyways, some courtyard entrances have sheds, some buildings have prominently projecting eaves, and the walls and houses form semi-enclosed boundary spaces. After performing a conventional UAV oblique photogrammetry acquisition on the village, an initial oblique image set and corresponding initial flight pose records are obtained, and then a rough 3D scene is constructed. Subsequently, based on the interface change relationships and spatial transition relationships in this rough 3D scene, the alleyway facades, courtyard entrance edges, wall outer interfaces, and public space edges are discretized, resulting in multiple planning observation units. Analyzing the visibility of each planning observation unit reveals that rooftop areas typically have higher projected areas and lower occlusion impact, while narrow alleyway facades and courtyard edges are more prone to lower visibility quality scores due to insufficient viewing angles or occlusion. Through the method proposed in this embodiment, the system can identify potential critical weak areas before entering the supplementary acquisition decision stage.

[0031] S20: Obtain the planning interaction record based on the rough three-dimensional scene of the village, and calculate the initial collection completeness, planning importance and required collection completeness corresponding to each planning observation unit.

[0032] Specifically, based on the visible quality score corresponding to each planning observation unit, the initial acquisition completeness of the planning observation unit is calculated; planning interaction records formed by planners in the rough 3D scene of the village are obtained, and the planning interaction records are associated with each planning observation unit; for each planning observation unit, the planning importance of the planning observation unit is calculated based on the planning interaction records, the spatial type of the planning observation unit, and the positional relationship of the planning observation unit in the continuous spatial representation; based on the planning importance and the spatial type of the planning observation unit, the required acquisition completeness of the planning observation unit is calculated.

[0033] In this embodiment of the invention, step S20 is used to combine objective data collection quality with subjective planning needs for processing, thereby establishing differentiated objectives oriented towards planning applications. It is easy to understand that in the actual work of village planning, not all spatial interfaces are equally important. Some spatial interfaces, although objectively existing in the 3D model, may not be critical to the current specific task; conversely, other spatial interfaces, although small in area, may be directly related to landscape improvement, public space renovation, villager consultation and display, or grassroots review and judgment. Therefore, if all planning observation units are uniformly evaluated based solely on the visible quality score and rough scene obtained in step S10, results that truly conform to planning usage habits cannot be obtained. In this embodiment of the invention, by introducing planning interaction records in this step, the scope of the scene that planners focus on directly affects the subsequent data processing flow.

[0034] First, based on the visible quality score calculated in step S10, the initial acquisition completeness of each planned observation unit is calculated. The initial acquisition completeness satisfies the following formula:

[0035] in, Indicates the first The initial data collection completeness of each planning observation unit; Indicates the initial number of tilted images; Indicates the first The planning observation unit in the first The visible quality score in the initial oblique image. This formula aims to synthesize the effective coverage of the same planning observation unit by multiple images. Specifically, when a planning observation unit has good visible quality in multiple images, its initial acquisition completeness will be improved accordingly; when a planning observation unit appears in several images, but consistently suffers from unpleasant viewing angles, insufficient clarity, or severe occlusion, its initial acquisition completeness will remain at a low level.

[0036] After obtaining the initial data collection completeness, the planning interaction records generated by planners in the rough 3D scene are further acquired. These records can originate from actions performed by planners in the scene, such as selecting points, defining boxes, drawing, specifying paths, or annotating areas, on lanes, courtyard edges, public space interfaces, building exteriors, or other key areas. It should be noted that these planning interaction records are not supplementary information, but rather inputs used in this embodiment to drive subsequent differentiated reinforcement processing. Specifically, after establishing a correspondence between the planning interaction records and planning observation units, each planning observation unit obtains information related to the current planning task, thereby providing a basis for calculating the planning importance.

[0037] Furthermore, based on the planning interaction records, the spatial type of the planning observation units, and the positional relationship of the planning observation units in the continuous spatial representation, the planning importance of each planning observation unit is calculated. The planning importance satisfies the following formula:

[0038] in, Indicates the first The importance of planning for each planning observation unit; The interaction marker factor is used to characterize whether the planning observation unit is directly selected by the planning interaction record and the degree of attention it receives. This indicates the application matching factor, used to characterize the degree of matching between the planning observation unit and the current planning application scenario. ; represents the location importance factor, used to characterize the degree of locational importance of the planning observation unit in the continuous spatial representation; , and This represents the corresponding weighting coefficient. It should be noted that the application matching factor reflects the difference between alleyway appearance improvement tasks, which prioritize facade interfaces, and courtyard public space renovation tasks, which prioritize entrance edges and enclosure interfaces. The location importance factor reflects the characteristic that interfaces at alleyway corners, courtyard entrances, and spatial nodes are more important than ordinary continuous interfaces. Therefore, the planning importance is not simply a result of manual labeling, but a comprehensive value determined by human attention, task attributes, and spatial structure.

[0039] After obtaining the importance of the plan, this embodiment of the invention further calculates the required data collection completeness for each planning observation unit. The required data collection completeness satisfies the following formula:

[0040] in, Indicates the first The required data completeness for each planning observation unit; Indicates the basic completeness requirement; This indicates the maximum planning importance among all planning observation units; Indicates the space type coefficient; and This represents the adjustment coefficient. The spatial type coefficient is used to distinguish the differences in the integrity requirements of different spatial interfaces. For example, street-facing facades, courtyard entrance boundaries, and public space edges usually have higher expression requirements, and therefore require correspondingly higher data collection integrity; while general roofs or non-key areas can allow for lower integrity requirements. By mapping planning importance and spatial type together to the required data collection integrity, this embodiment of the invention obtains the target state that each planning observation unit should achieve under the current planning task, thus providing a clear reference for the calculation of subsequent reinforcement requirements.

[0041] In one specific implementation, the aforementioned traditional settlement-type village can be used as an example. Planners highlight a lane to be renovated and its main facades on both sides in a rough 3D scene, while also marking the entrance and perimeter wall of a courtyard planned to be converted into a shared activity area. After mapping these interaction records to corresponding planning observation units, the system can find that the directly marked lane facade units and courtyard entrance boundary units have higher planning importance. Simultaneously, planning observation units located at lane corners and courtyard entrances, due to their more critical locations, will further increase their planning importance. Furthermore, since facade and entrance boundary space types have higher space type coefficients, the required data collection completeness for these areas will be automatically increased. Through this process, this embodiment of the invention accurately translates where planners want to focus their attention into the system's requirement for these areas to meet higher data quality standards, thereby truly realizing the pre-driving of data processing flow by planning needs.

[0042] S30: Based on the initial data collection completeness, planning importance, and required data collection completeness corresponding to each planning observation unit, calculate the reinforcement requirement value corresponding to each planning observation unit, and generate a supplementary data collection record table based on the reinforcement requirement value.

[0043] Specifically, based on the required data collection completeness and the initial data collection completeness, the basic deficiency of each planning observation unit is calculated; based on the set of planning observation units adjacent to the planning observation unit, the spatial continuity influence coefficient of the planning observation unit is calculated; based on the planning importance, the basic deficiency, and the spatial continuity influence coefficient, the reinforcement requirement value of the planning observation unit is calculated; based on the reinforcement requirement value corresponding to each planning observation unit, the screening criteria for candidate supplementary data collection actions are formed, and the screening results are written into the supplementary data collection record table.

[0044] Furthermore, based on the reinforcement requirement values ​​corresponding to each planning observation unit, a selection criterion for candidate supplementary acquisition actions is formed, and the selection results are written into the supplementary acquisition record table. Specifically, this includes: generating a set of candidate supplementary acquisition actions based on the rough 3D scene of the village, the set of planning observation units, the initial flight pose record, and the reinforcement requirement values; calculating the expected improvement value of the candidate supplementary acquisition action for each planning observation unit and for each planning observation unit; calculating the reinforcement benefit value of the candidate supplementary acquisition action based on the reinforcement requirement values ​​corresponding to each planning observation unit and the expected improvement value corresponding to the candidate supplementary acquisition action; sorting the candidate supplementary acquisition actions according to the reinforcement benefit value, and writing the sorting results into the supplementary acquisition record table.

[0045] In this embodiment of the invention, step S30 is used to further determine where reinforcement is most needed and how to organize supplementary data collection, based on the current data collection results and the required level of completion in the planning. It is easy to understand that if supplementary data collection is determined solely based on the condition that the initial data collection completeness is lower than the required completeness, while individual deficient areas can be identified, it is difficult to reflect the spatial integrity of continuous alleyways, continuous wall interfaces, or entire courtyard boundaries. For village planning, the key issue is often not the lack of clarity in a single location, but rather the overall lack of clarity in a certain spatial relationship, leading to an inability to continuously understand the landscape improvement interface or courtyard renovation boundary. Therefore, in this embodiment of the invention, in addition to calculating the basic deficiencies, spatial continuity influencing factors are further introduced in this step, and candidate supplementary data collection actions and corresponding supplementary data collection record tables are generated based on this.

[0046] First, based on the required data completeness and the initial data completeness, the basic deficiency of each planned observation unit is calculated. The basic deficiency satisfies the following formula:

[0047] in, Indicates the first The basic requirements of each planning observation unit are insufficient; Indicates the first The required data completeness for each planning observation unit; Indicates the first The initial data collection completeness of each planning observation unit. This formula directly quantifies the gap between the target requirement and the current state. Specifically, when the initial data collection completeness of a planning observation unit has reached or exceeded the required data collection completeness, its basic deficiency is zero, indicating that from an individual perspective, this planning observation unit does not require reinforcement; when the initial data collection completeness is lower than the required data collection completeness, its basic deficiency reflects the data gap that still exists for this planning observation unit under the current task.

[0048] Building upon this, the embodiments of the present invention further introduce a spatial continuity influence coefficient to reflect the joint deficiency relationship between adjacent planning observation units. The spatial continuity influence coefficient satisfies the following formula:

[0049] in, Indicates the first Spatial continuity influence coefficient of each planning observation unit; Indicates the relationship with the first A set of adjacent planning observation units; Indicates the number of adjacent planning observation units; Indicates the first The spatial correlation strength between a planning observation unit and its adjacent planning observation units; This indicates the degree of deficiency in adjacent planning observation units. It's important to note that this formula represents the situation where, if a planning observation unit itself has deficiencies, and its adjacent spaces also have deficiencies, then these deficiencies are no longer isolated problems but will have a greater impact on the representation of continuous space. For example, if three consecutive interfaces in a lane facade have insufficient clarity, its planning representation obstacle will be significantly higher than if only one interface has a slight deficiency. Therefore, through the spatial continuity influence coefficient, large areas of weakness in continuous space can be transformed from single-point problems into overall reinforcement problems.

[0050] Furthermore, based on the planning importance, the amount of basic deficiencies, and the spatial continuity impact coefficient, the reinforcement requirement value for each planning observation unit is calculated. The reinforcement requirement value satisfies the following formula:

[0051] in, Indicates the first The reinforcement requirements for each planning observation unit; Indicates the first The importance of planning for each planning observation unit; Indicates the first The basic requirements of each planning observation unit are insufficient; Indicates the first The spatial continuity impact coefficient of a planning observation unit. This formula means that only when a planning observation unit simultaneously possesses three characteristics—importance in planning, current insufficiency, and impact on surrounding continuous space—will its reinforcement demand value increase significantly. This can effectively prevent the system from misjudging a large number of ordinary areas as priority targets for reinforcement, concentrating limited reinforcement resources on areas that truly affect the planning's expressive effect.

[0052] After obtaining the reinforcement requirement value, this embodiment of the invention does not immediately output a conclusion that supplementary acquisition is needed. Instead, it further generates a set of candidate supplementary acquisition actions based on the rough 3D scene, the planned observation unit set, the initial flight pose record, and the reinforcement requirement value. It should be noted that supplementary acquisition actions are not equivalent to simply repeating the original flight path, but rather regenerating a set of acquisition action schemes oriented towards the local space around the high reinforcement requirement area. These candidate actions may include suggested supplementary shooting location, suggested shooting direction, suggested shooting altitude, suggested number of shots, and spatial relationship with the existing flight path. Furthermore, in one executable implementation, the system calculates the expected improvement value of each candidate supplementary acquisition action for each planned observation unit. The expected improvement value satisfies the following formula:

[0053] in, Indicates the first The candidate supplementary collection action for the first The projected improvement value for each planning observation unit; Indicates the projected area factor; Indicates the expected viewpoint adaptation factor; Indicates the expected clarity factor; This indicates the safe and executable factor.

[0054] Subsequently, based on the reinforcement demand value corresponding to each planning observation unit and the expected improvement value corresponding to the candidate supplementary collection actions, the reinforcement benefit value of the candidate supplementary collection actions is further calculated. The reinforcement benefit value satisfies the following formula:

[0055] in, Indicates the first The enhancement benefit value of each candidate supplementary collection action; Indicates the total number of planning observation units; Indicates the first The reinforcement requirements for each planning observation unit; Indicates the first The candidate supplementary collection action for the first The projected improvement value for each planning observation unit; Indicates the first Additional flight distance corresponding to each candidate supplementary data collection action; Indicates the first The number of influence ranges corresponding to each candidate supplementary data collection action; and This represents the adjustment coefficient. The formula aims to comprehensively balance the improvement effect and the execution cost. Specifically, if a candidate action theoretically offers good improvement, but requires a significant deviation from the original path, affects many sensitive areas, or substantially increases on-site execution costs, its enhancement benefit value will decrease accordingly. Conversely, if a candidate action can cover multiple high-demand planning observation units with relatively low additional costs, its enhancement benefit value will be higher.

[0056] Finally, the candidate supplementary acquisition actions are ranked according to their enhancement benefit values, and the ranking results are written into the supplementary acquisition record table. This supplementary acquisition record table serves not only as the basis for subsequent supplementary acquisition execution but also as the foundation for data correspondence when receiving supplementary acquisition images and supplementary flight pose records. In one specific implementation, taking the aforementioned alleyways and courtyard entrances that are of particular interest to planners as examples, the system can generate multiple candidate supplementary acquisition actions, such as oblique supplementary acquisition along the alleyway direction, lateral supplementary acquisition at alleyway corners, and oblique supplementary acquisition at the outer interface of courtyard entrances. After ranking by enhancement benefit values, the system prioritizes retaining actions that can simultaneously improve multiple units with high enhancement needs and have relatively small additional flight distances, and writes these actions into the supplementary acquisition record table. Through this process, supplementary acquisition no longer relies entirely on manual experience but becomes a calculable result driven by both planning needs and spatial analysis.

[0057] S40: Receive the supplementary acquisition image set returned according to the supplementary acquisition record table and the supplementary flight pose record corresponding to the supplementary acquisition image set, and integrate the initial tilt image set and the supplementary acquisition image set into a complete image set, and integrate the initial flight pose record and the supplementary flight pose record into a complete pose record.

[0058] In this embodiment of the invention, step S40 is used to reintegrate the supplementary acquisition results into the unified data processing chain, so that the aforementioned planning-driven supplementary acquisition decision is truly transformed into an improvement in subsequent 3D representation capabilities. It is easy to understand that the supplementary acquisition record table is merely an intermediate process document; its value lies not only in indicating where more data needs to be acquired, but more importantly, in enabling subsequent new acquisition results to establish a strict one-to-one correspondence with the original data and further participate in unified modeling.

[0059] In practical applications, after supplementary image acquisition is completed based on the supplementary acquisition record table, the system receives the supplementary image set and its corresponding supplementary flight pose record. These supplementary images typically focus on spatial areas that are insufficient in the original images but are crucial for planning, such as the facades on both sides of narrow alleys, the edges of courtyard entrances, the outer interface of walls, the interface under eaves, or the boundaries of public spaces. The supplementary flight pose record describes the spatial positional relationships, acquisition directions, and attitude relationships of these newly added images. It should be noted that the supplementary images are not isolated data, but rather enhanced data specifically acquired to meet the reinforcement requirements of the aforementioned planning observation units; therefore, their role in subsequent processing is not entirely the same as that of the initial oblique images.

[0060] Furthermore, this embodiment of the invention integrates the initial tilted image set and the supplementary acquired image set into a complete image set, and integrates the initial flight pose record and the supplementary flight pose record into a complete pose record. This integration is not merely a simple stitching together; it also includes unified numbering, unified coordinate reference, unified image-pose correspondence, and a unified data retrieval entry point. It is easy to understand that if the supplementary acquired images are processed separately, although they may produce high-resolution results in certain areas, they often weaken the spatial consistency between the supplementary acquired images and the original village scene. If only the original initial images are relied upon, the new perspectives and details obtained through the supplementary acquisition cannot truly participate in the reconstruction. Therefore, the complete image set and complete pose record formed through this step can simultaneously retain the overall supporting role of the original data for the village scene, as well as the local enhancement role of the supplementary acquired data for key areas.

[0061] Furthermore, in one feasible implementation, source identifiers can be established for supplementary acquired images during the integration phase, so that the enhancement effect of supplementary images on key planning observation units can be appropriately reflected during subsequent local depth estimation. Through this method, this step not only achieves data-level merging but also provides a unified and traceable input structure for local depth estimation and 3D fusion in step S50. In a specific implementation, taking the aforementioned traditional settlement-type village as an example, after receiving newly acquired images around key alleyways and key courtyard entrances, the system can organize them with the original initial oblique images in a unified scene coordinate system and establish spatial connections between all images through complete pose recording. Thus, when proceeding to the next step, the system no longer distinguishes between images captured by the original flight path and those added during supplementary acquisition; instead, it treats all data as a complete image system for unified analysis, only reflecting differences in weight and contribution methods between images from different sources. This processing method ensures that local enhancement does not disrupt the continuity of the overall scene and significantly enhances the readability of key localities while preserving the overall model framework.

[0062] S50: Based on the complete image set, complete pose record, and planning observation unit set, perform local depth estimation and 3D fusion processing to generate a real-world 3D model of village planning, a planning representation quality map, and local spatial description data.

[0063] Specifically, based on the complete pose record, the spatial projection relationship between each planning observation unit and each image in the complete image set is determined, and a subset of images that can form an effective observation of each planning observation unit is selected; for any point in the planning observation unit and the corresponding depth candidate values, a comprehensive matching cost is calculated based on the image subset; based on the comprehensive matching cost, local depth estimation and three-dimensional fusion processing are performed to form a village planning real-scene three-dimensional model, a planning expression quality map, and local spatial description data.

[0064] Furthermore, based on the comprehensive matching cost, local depth estimation and 3D fusion processing are performed to form a village planning real-scene 3D model, a planning expression quality map, and local spatial description data. Specifically, this includes: selecting the depth value that minimizes the comprehensive matching cost from the depth candidate set corresponding to each location as the final depth value; performing 3D fusion processing based on the final depth values ​​of each location to generate a village planning real-scene 3D model; calculating the expression quality value of each planning observation unit based on the completeness of the supplementary acquisition and the updated acquisition after 3D fusion, the planning importance, and the local clarity of the model; generating a planning expression quality map based on the expression quality value; and generating local spatial description data based on the spatial type of the planning observation unit, the change in completeness before and after supplementary acquisition, and the expression quality value.

[0065] In this embodiment of the invention, step S50 is used to unify the original acquisition, planning interaction, reinforcement decision, and supplementary acquisition results into a final 3D result that can be used for planning analysis and display. It is easy to understand that key interfaces in traditional settlement-type villages often possess characteristics such as low texture, local occlusion, narrow space, and limited viewing angle. This means that if only conventional reconstruction processing methods based on texture consistency are used, problems such as unstable depth estimation, blurred boundaries, or local breaks can easily occur in areas such as alleyway facades, courtyard entrances, wall boundaries, and interfaces under eaves. Therefore, this embodiment of the invention does not simply perform ordinary fusion of the complete image set, but simultaneously considers texture differences, edge direction differences, occlusion adaptation relationships, planning importance, and the contribution of supplementary images during the local depth estimation stage, thereby forming a 3D representation result that better suits the planning application goals.

[0066] First, for each planned observation unit, a subset of images capable of observing that unit is selected from the complete image set. This is done to avoid introducing images irrelevant to the current local spatial context into the depth estimation process, thereby reducing invalid comparisons and unnecessary interference. Then, for any point within the planned observation unit and its corresponding depth candidate values, the comprehensive matching cost is calculated. The comprehensive matching cost satisfies the following formula:

[0067] in, Indicates the first Location within each planning observation unit In depth candidate value The comprehensive matching value; Indicates that the first can be observed A subset of images from each planning observation unit; Indicates the first Zhang's image is in position and depth candidate values The corresponding weights; Indicates the first Texture and edge blending coefficients of a planned observation unit; Indicates the value of texture differences; Indicates the difference in cost value along the edge direction; This indicates an extremely small positive number that prevents the denominator from being zero; Indicates the space smoothness adjustment coefficient; This represents the local spatial smoothing term. It's important to note that this formula incorporates texture information, structural edge information, and spatial continuity constraints into the calculation, enabling the system to stably estimate the depth location of key local spaces even when facing low-texture and complex occlusion areas. Furthermore, because the spatial smoothing term is adjusted for planning importance, when a planning observation unit is more critical in the planning task, its local details are more fully preserved in the final estimate, rather than being eliminated by excessive smoothing.

[0068] Furthermore, in the above formula, the consistency weight satisfies the following formula:

[0069] in, Indicates the first Zhang's image is in position and depth candidate values The corresponding weights; Indicates the first Zhang's image is the first The overall visible quality score of each planning observation unit; Indicates the occlusion adaptation coefficient; This represents the contribution coefficient of the image source. The formula explains that whether an image should play a significant role in the current location and depth candidate is not determined by a single factor, but rather by a combination of factors including the image's visibility quality to the planned observation unit, its fit with the current depth and occlusion boundaries, and whether the image came from supplementary acquisitions. This approach allows truly clear images with suitable perspectives and direct contribution to enhancing key areas to have a higher weight in the estimation, while images that are severely occluded or offer little help to the current local space automatically have a reduced impact.

[0070] The occlusion adaptation coefficient satisfies the following formula:

[0071] in, Indicates the occlusion adaptation coefficient; Indicates position In depth candidate value The following and the first Distance deviation between foreground occlusion boundaries in Zhang image; This indicates the occlusion adaptation range parameter. In this way, when a certain depth candidate value does not match the visibility and occlusion relationship in the image, the weight corresponding to that depth candidate will be automatically reduced, thereby effectively reducing the interference of eaves, trellises, branches or temporary occlusions on the reconstruction of key interfaces.

[0072] After completing the comprehensive matching cost calculation, this embodiment of the invention further calculates the texture richness of the planning observation unit. The texture richness satisfies the following formula:

[0073] in, Indicates the first Texture richness of each planning observation unit; Indicates the first The projection area of ​​each planning observation unit in the image; Represents the pixels in the projection area; Represents pixels Image gradient at the location; Indicates the gradient adjustment parameter; This represents the number of pixels within the projected area. This formula is used to determine whether a given planning observation unit belongs to a texture-rich or texture-sparse region. It's easy to understand that walls, painted walls, gray-white facades, and areas under the shadow of eaves are generally considered to have relatively sparse textures, while facades with dense doors and windows, obvious brick and stone textures, or many decorative components tend to be texture-rich regions. Quantifying texture richness provides a basis for the subsequent balanced use of texture and edge information.

[0074] Furthermore, based on texture richness and planning importance, the texture and edge blending coefficient of the planning observation unit is calculated. The texture and edge blending coefficient satisfies the following formula:

[0075] in, Indicates the first Texture and edge blending coefficients of a planned observation unit; Indicates the percentage of the base edge; Indicates a low texture enhancement factor; This represents the enhancement coefficient of the planning importance; This represents the maximum planning importance among all planning observation units. The formula embodies the idea that when a planning observation unit has weaker texture and greater planning importance, edge direction information should occupy a higher proportion in the comprehensive estimation; conversely, when a planning observation unit has relatively rich texture and moderate planning importance, texture differences can still be used as the main basis. Through this adaptive balancing method, embodiments of the present invention can better maintain spatial structures such as door and window outlines, eaves boundaries, and wall fold lines in low-texture critical areas.

[0076] Subsequently, regarding the location The final depth value is selected from the corresponding candidate depth set that minimizes the overall matching cost. The final depth value satisfies the following formula:

[0077] in, Indicates position The final depth value; Indicates position The system generates a candidate depth set. By selecting the optimal depth value for each location, the system can gradually recover the three-dimensional geometric relationships of key local spaces. Subsequently, a three-dimensional fusion process is performed based on the final depth values ​​of all locations to generate a real-world 3D model of the village planning. It should be noted that the 3D fusion result here is a planning-oriented real-world 3D model that has undergone planning demand-driven, supplementary data acquisition, and local adaptation processing. It is significantly superior to a general model obtained solely based on data acquisition from a unified flight path in terms of the continuity of alleyway facades, the readability of courtyard boundaries, and the clarity of public space interfaces.

[0078] In this embodiment of the invention, the principle of performing 3D fusion processing based on the final depth values ​​of all locations is to uniformly map the final depth results obtained from local depth estimation at different locations within each planning observation unit back to the same 3D space. Based on the adjacency, continuity, and surface attribution relationships between these locations, the overall 3D morphology of building interfaces, wall interfaces, alleyway boundaries, courtyard entrance boundaries, and public space edges in the village is restored. It is easy to understand that the aforementioned final depth values ​​essentially provide spatial distance information relative to the observation viewpoint. However, if only the depth results at a single location are considered, it remains a discrete local spatial description and cannot form a continuous 3D scene suitable for planning expression. Therefore, this embodiment of the invention further organizes the dispersed depth results into an overall spatial expression result with continuous surface relationships through 3D fusion processing.

[0079] In practical applications, the 3D fusion processing is not a simple stacking of all depth results. Instead, it requires consistent integration of the final depth values ​​from different locations to ensure that spatial locations belonging to the same building facade, the same section of wall boundary, or the same courtyard entrance interface can form a continuous, harmonious, and clearly defined spatial structure. Furthermore, in one feasible implementation, the system can prioritize merging adjacent locations with similar surface orientations into the same local interface based on the positional distribution relationship corresponding to the final depth values, and coordinate the edge connection relationships between local interfaces. This ensures that the final generated 3D model can reflect the overall spatial structure while maintaining the detailed continuity of alleyway interfaces, courtyard boundaries, and public space transitions.

[0080] Furthermore, 3D fusion processing also helps suppress the spread of local discrete errors. Since the final depth values ​​at different locations originate from local estimations under multiple image conditions, slight deviations may still exist at individual locations. Without unified fusion, these deviations can easily manifest as surface fragmentation, abnormal interface undulations, or boundary discontinuities in the final result. By performing fusion processing on the final depth values ​​of all locations within the same planning observation unit system and the same complete pose reference system, the local depth results can be coordinated holistically. This ensures that the final 3D model retains the detailed representation of key locations while avoiding the impact of local estimation fluctuations on the overall interface quality. Through this method, the resulting village planning real-world 3D model can simultaneously meet the needs of overall scene browsing, local spatial analysis, and planning display and communication.

[0081] After generating a 3D model of the village plan, this embodiment of the invention does not stop at model output, but continues to generate a planning representation quality map and local spatial description data. To this end, based on the completeness of the supplementary data collection and the updated data collection after 3D fusion, the importance of the plan, and the local clarity of the model, the representation quality value of each planning observation unit is calculated. The representation quality value satisfies the following formula:

[0082] in, Indicates the first The expression quality value of each planning observation unit; Indicates the first The completeness of the updated data collection for each planning observation unit after supplementary data collection and 3D fusion; Indicates the first The formula represents the local clarity of the model corresponding to each planning observation unit. It is used to characterize that if a region has high planning importance but still lacks clarity after supplementary collection and fusion, its expression quality value will not be artificially high, thus reminding planners to use it with caution; if a region has high planning importance, high update collection completeness, and good local model clarity, its expression quality value will be correspondingly improved, and the region is sufficient to support planning presentation and discussion.

[0083] Furthermore, a planning expression quality map is generated based on the expression quality value, and local spatial description data is generated based on the spatial type of the planning observation unit, the change in completeness before and after supplementary data collection, and the expression quality value. It should be noted that the planning expression quality map can be used to visually identify which areas in the village have reached a high expression quality and which areas still require attention; the local spatial description data can further explain whether a certain area is suitable for judging the appearance of alleyways, suitable for discussing the renovation of courtyard public spaces, or should continue to be manually reviewed. Through this process, the final output is no longer just a technical 3D model, but a truly understandable expression result for planners, managers, and villagers.

[0084] In one specific implementation, taking the aforementioned traditional settlement-type village as an example, the system integrates the initial oblique image and the supplementary acquired image, and then performs local depth estimation and 3D fusion processing on the facades of key alleyways, the edges of courtyard entrances, and the boundaries of public spaces. Through the processing of this embodiment, the facade interfaces of alleyways, which were previously unclear due to roof obstruction in a rough 3D scene, can be significantly improved, and areas lacking continuous boundary relationships at courtyard entrances can be more fully represented. The representation quality of key alleyway corners, courtyard entrance boundaries, and continuous wall interfaces is significantly higher than before supplementary acquisition, while non-key areas maintain their original general representation level. Through local spatial description data, planners can clearly identify which areas already meet the conditions for showcasing landscape improvement and spatial transformation, and which areas are still only suitable for auxiliary browsing. Thus, this embodiment deeply integrates planning needs, supplementary acquisition decisions, and 3D reconstruction, ultimately forming a real-scene 3D result system for village planning applications.

[0085] Reference Figure 3 , Figure 3 This is a schematic diagram of an embodiment of the village planning real-scene 3D data acquisition device of the present invention.

[0086] like Figure 3 As shown, the village planning real-scene 3D data acquisition device proposed in this embodiment of the invention includes: The segmentation module is used to acquire the initial tilted image set of the village and the initial flight pose record corresponding to the initial tilted image set, perform coarse 3D reconstruction processing to generate a coarse 3D scene of the village, and divide the planning observation unit set based on the coarse 3D scene of the village. The calculation module is used to acquire planning interaction records based on the rough three-dimensional scene of the village, and to calculate the initial collection completeness, planning importance, and required collection completeness corresponding to each planning observation unit. The generation module is used to calculate the reinforcement requirement value corresponding to each planning observation unit based on the initial collection completeness, planning importance and required collection completeness of each planning observation unit, and generate a supplementary collection record table based on the reinforcement requirement value. The integration module is used to receive the supplementary acquisition image set returned according to the supplementary acquisition record table and the supplementary flight pose record corresponding to the supplementary acquisition image set, and to integrate the initial tilt image set and the supplementary acquisition image set into a complete image set, and to integrate the initial flight pose record and the supplementary flight pose record into a complete pose record. The fusion module is used to perform local depth estimation and 3D fusion processing based on the complete image set, complete pose record, and planning observation unit set to generate a real-world 3D model of village planning, a planning representation quality map, and local spatial description data.

[0087] Other embodiments or specific implementations of the village planning real-scene three-dimensional data acquisition device of the present invention can refer to the above-mentioned method embodiments, and will not be repeated here.

[0088] Furthermore, the present invention also proposes a village planning real-scene 3D data acquisition device, which includes: a memory, a processor, and a village planning real-scene 3D data acquisition program stored in the memory and executable on the processor. When the village planning real-scene 3D data acquisition program is executed by the processor, it implements the steps of the village planning real-scene 3D data acquisition method as described above.

[0089] The specific implementation method of the village planning real-scene 3D data acquisition device in this application is basically the same as the above-mentioned village planning real-scene 3D data acquisition method embodiments, and will not be repeated here.

[0090] Furthermore, this invention also proposes a readable storage medium, which includes a computer-readable storage medium storing a village planning real-scene 3D data acquisition program thereon. The readable storage medium may be... Figure 1 The memory 1005 in the terminal can also be at least one of ROM (Read-Only Memory) / RAM (Random Access Memory), magnetic disk, optical disk, etc. The readable storage medium includes several instructions to cause a village planning real-scene three-dimensional data acquisition device with a processor to execute the village planning real-scene three-dimensional data acquisition method described in various embodiments of the present invention.

[0091] The specific implementation in the readable storage medium of this application is basically the same as the embodiments of the above-described method for acquiring real-world 3D data of village planning, and will not be described again here.

[0092] It is understood that in the description of this specification, references to terms such as "an embodiment, another embodiment, other embodiments, or first to Nth embodiments," etc., indicate that a specific feature, structure, material, or characteristic described in connection with that embodiment or example is included in at least one embodiment or example of the present invention. In this specification, illustrative expressions of the above terms do not necessarily refer to the same embodiment or example. Furthermore, the specific features, structures, materials, or characteristics described may be combined in any suitable manner in one or more embodiments or examples.

[0093] It should be noted that, in this document, the terms “comprising,” “including,” or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or system that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or system.

[0094] Through the above description of the embodiments, those skilled in the art can clearly understand that the methods of the above embodiments can be implemented by means of software plus necessary general-purpose hardware platforms. Of course, they can also be implemented by hardware, but in many cases the former is a better implementation method. Based on this understanding, the technical solution of the present invention, or the part that contributes to the prior art, can be embodied in the form of a software product. This computer software product is stored in a storage medium (such as ROM / RAM, magnetic disk, optical disk) as described above, and includes several instructions to cause a terminal device (which may be a mobile phone, computer, server, or network device, etc.) to execute the methods described in the various embodiments of the present invention.

[0095] The above are merely preferred embodiments of the present invention and do not limit the scope of the patent. Any equivalent structural or procedural transformations made based on the description and drawings of the present invention, or direct or indirect applications in other related technical fields, are similarly included within the scope of patent protection of the present invention.

Claims

1. A method for acquiring real-world 3D data of village planning, characterized in that, Includes the following steps: Acquire an initial tilted image set of the village and an initial flight pose record corresponding to the initial tilted image set, perform coarse 3D reconstruction processing to generate a coarse 3D scene of the village, and divide the village into a set of planning observation units based on the coarse 3D scene of the village. Obtain planning interaction records based on the rough 3D scene of the village, and calculate the initial collection completeness, planning importance, and required collection completeness corresponding to each planning observation unit; Based on the initial data collection completeness, planning importance, and required data collection completeness corresponding to each planning observation unit, the reinforcement requirement value corresponding to each planning observation unit is calculated, and a supplementary data collection record table is generated based on the reinforcement requirement value. Receive the supplementary acquisition image set returned according to the supplementary acquisition record table and the supplementary flight pose record corresponding to the supplementary acquisition image set, and integrate the initial tilt image set and the supplementary acquisition image set into a complete image set, and integrate the initial flight pose record and the supplementary flight pose record into a complete pose record; Local depth estimation and 3D fusion processing are performed based on a complete image set, complete pose records, and a set of planning observation units to generate a real-world 3D model of the village plan, a planning representation quality map, and local spatial description data.

2. The method for acquiring real-world 3D data of village planning as described in claim 1, characterized in that, Acquire an initial set of oblique images of the village and initial flight pose records corresponding one-to-one with the initial oblique image set. Perform coarse 3D reconstruction processing to generate a coarse 3D scene of the village. Based on the coarse 3D scene of the village, divide and plan a set of observation units, specifically including: A rough 3D scene of the village is established based on the initial oblique image set and the initial flight attitude record; Based on the spatial surface orientation, local elevation differences, and spatial continuity in the rough three-dimensional scene of the village, the village space is divided into multiple planning observation units to form a set of planning observation units; For each planned observation unit and each initial tilted image, calculate the visibility quality score of the planned observation unit in the corresponding initial tilted image; The visible quality score is stored as input data for subsequent calculation of the initial acquisition integrity.

3. The method for acquiring real-world 3D data of village planning as described in claim 1, characterized in that, Obtain planning interaction records based on the rough 3D scene of the village, and calculate the initial data collection completeness, planning importance, and required data collection completeness for each planning observation unit, specifically including: Based on the visible quality score corresponding to each planning observation unit, the initial acquisition completeness of the planning observation unit is calculated; Acquire the planning interaction records generated by planners in the rough 3D scene of the village, and associate the planning interaction records with each planning observation unit; For each planning observation unit, the planning importance of the planning observation unit is calculated based on the planning interaction record, the spatial type of the planning observation unit, and the positional relationship of the planning observation unit in the continuous spatial representation. Based on the planning importance and the spatial type of the planning observation unit, the required data collection completeness of the planning observation unit is calculated.

4. The method for acquiring real-world 3D data of village planning as described in claim 1, characterized in that, Based on the initial data collection completeness, planning importance, and required data collection completeness for each planning observation unit, the reinforcement requirement value for each planning observation unit is calculated, and a supplementary data collection record table is generated according to the reinforcement requirement value, specifically including: Based on the required data collection completeness and the initial data collection completeness, calculate the basic deficiency of each planned observation unit; Based on the set of planning observation units adjacent to the planning observation unit, calculate the spatial continuity influence coefficient of the planning observation unit; Based on the planning importance, the basic deficiency, and the spatial continuity influence coefficient, the reinforcement requirement value of the planning observation unit is calculated; Based on the reinforcement requirement values ​​corresponding to each planning observation unit, a selection criterion for candidate supplementary collection actions is formed, and the selection results are written into the supplementary collection record table.

5. The method for acquiring real-world 3D data of village planning as described in claim 4, characterized in that, Based on the reinforcement requirement values ​​corresponding to each planning observation unit, a selection criterion for candidate supplementary data collection actions is formed, and the selection results are written into the supplementary data collection record table, specifically including: Based on the rough 3D scene of the village, the planned set of observation units, the initial flight pose record, and the reinforcement requirement value, a set of candidate supplementary acquisition actions is generated. For each candidate supplementary acquisition action and each planned observation unit, calculate the expected improvement value of the candidate supplementary acquisition action for the planned observation unit; Based on the reinforcement demand value corresponding to each planning observation unit and the expected improvement value corresponding to the candidate supplementary collection action, the reinforcement benefit value of the candidate supplementary collection action is calculated. The candidate supplementary collection actions are sorted according to the reinforcement benefit value, and the sorting results are written into the supplementary collection record table.

6. The method for acquiring real-world 3D data of village planning as described in claim 1, characterized in that, Based on a complete image set, complete pose records, and a set of planning observation units, local depth estimation and 3D fusion processing are performed to generate a real-world 3D model of the village plan, a planning representation quality map, and local spatial description data, specifically including: Based on the complete pose record, the spatial projection relationship between each planned observation unit and each image in the complete image set is determined, and a subset of images that can form an effective observation of each planned observation unit is selected. For any point in the planned observation unit and the corresponding depth candidate values, the comprehensive matching cost is calculated based on the image subset. Based on the comprehensive matching cost, local depth estimation and three-dimensional fusion processing are performed to form a real-world three-dimensional model of village planning, a planning expression quality map, and local spatial description data.

7. The method for acquiring real-world 3D data of village planning as described in claim 6, characterized in that, Based on the comprehensive matching cost, local depth estimation and 3D fusion processing are performed to generate a village planning real-scene 3D model, a planning expression quality map, and local spatial description data, specifically including: The depth value that minimizes the comprehensive matching cost is selected from the depth candidate set corresponding to each location as the final depth value. Based on the final depth value of each location, three-dimensional fusion processing is performed to generate a real-world three-dimensional model of the village planning. Based on the completeness of the updated acquisition after supplementary acquisition and 3D fusion, the planning importance, and the local clarity of the model, the expression quality value of each planning observation unit is calculated. Based on the expression quality value, a planning expression quality map is generated, and local spatial description data is generated based on the spatial type of the planning observation unit, the change in completeness before and after supplementary collection, and the expression quality value.

8. A three-dimensional data acquisition device for village planning, characterized in that, include: The segmentation module is used to acquire the initial tilted image set of the village and the initial flight pose record corresponding to the initial tilted image set, perform coarse 3D reconstruction processing to generate a coarse 3D scene of the village, and divide the planning observation unit set based on the coarse 3D scene of the village. The calculation module is used to acquire planning interaction records based on the rough three-dimensional scene of the village, and to calculate the initial collection completeness, planning importance, and required collection completeness corresponding to each planning observation unit. The generation module is used to calculate the reinforcement requirement value corresponding to each planning observation unit based on the initial collection completeness, planning importance and required collection completeness of each planning observation unit, and generate a supplementary collection record table based on the reinforcement requirement value. The integration module is used to receive the supplementary acquisition image set returned according to the supplementary acquisition record table and the supplementary flight pose record corresponding to the supplementary acquisition image set, and to integrate the initial tilt image set and the supplementary acquisition image set into a complete image set, and to integrate the initial flight pose record and the supplementary flight pose record into a complete pose record. The fusion module is used to perform local depth estimation and 3D fusion processing based on the complete image set, complete pose record, and planning observation unit set to generate a real-world 3D model of village planning, a planning representation quality map, and local spatial description data.

9. A village planning real-scene 3D data acquisition device, characterized in that, The village planning real-scene 3D data acquisition device includes: a memory, a processor, and a village planning real-scene 3D data acquisition program stored in the memory and executable on the processor. When the village planning real-scene 3D data acquisition program is executed by the processor, it implements the steps of the village planning real-scene 3D data acquisition method as described in any one of claims 1 to 7.

10. A storage medium, characterized in that, The storage medium stores a village planning real-scene 3D data acquisition program, which, when executed by a processor, implements the steps of the village planning real-scene 3D data acquisition method as described in any one of claims 1 to 7.

Images