CN-121996462-A - Portable cluster three-dimensional rapid modeling system and method

CN121996462ACN 121996462 ACN121996462 ACN 121996462ACN-121996462-A

Abstract

The invention relates to the technical field of modeling systems and discloses a portable cluster three-dimensional rapid modeling system and a method, wherein the system comprises an integrated chassis unit, an internal integrated main computing node and a sub computing node; the system comprises a power supply and power management module, a man-machine interaction module, a scheduling and modeling module and an environment adaptation module, wherein the power supply and power management module is used for monitoring loads of all computing nodes in real time and dynamically adjusting power supply, the man-machine interaction module is used for supporting a user to switch among a plurality of computing nodes through a keyboard or a mouse, the scheduling and modeling module is used for carrying out three-dimensional modeling on a target area according to image data, and the environment adaptation module is used for carrying out real-time monitoring and intelligent regulation and control on the internal environment temperature and humidity of an integrated chassis unit. According to the method, the rapid generation and the result derivation of the three-dimensional model of the target area are realized through the closed loop flow of unmanned aerial vehicle data acquisition, cluster node parallel calculation and local aggregation optimization.

Inventors

LI QUSHENG
LI ZHENXI
ZHONG BOJIN
LI LINCONG
Liang Renzheng
ZHANG ZHENLIN

Assignees

广西壮族自治区自然资源遥感院

Dates

Publication Date: 20260508
Application Date: 20260124

Claims (10)

1. A portable clustered three-dimensional rapid modeling system, the system comprising: The integrated chassis unit is internally integrated with a plurality of computing nodes, the computing nodes are interconnected through a back plate, and each computing node comprises a main computing node and a sub computing node; the power supply and power management module is used for monitoring the load of each computing node in real time and dynamically adjusting the power supply; The human-computer interaction module is arranged on the integrated chassis unit and comprises a KVM (keyboard-video-mouse) switching table, a keyboard, a mouse and a display, wherein the keyboard, the mouse and the display are connected with each computing node through the KVM switching table, and the KVM switching table is used for supporting a user to switch among a plurality of computing nodes through the keyboard or the mouse; the scheduling and modeling module is used for carrying out three-dimensional modeling on the target area according to the image data; The environment adaptation module is arranged in the integrated chassis unit and comprises an air cooling unit and a heat dissipation unit, and the environment adaptation module is used for monitoring and intelligently regulating the internal environment temperature and humidity of the integrated chassis unit in real time.
2. The portable clustered three-dimensional rapid modeling system of claim 1 wherein the scheduling and modeling module comprises a data receiving unit configured to receive image data of a target area; the task scheduling dividing unit is configured to guide the image data into a main computing node, divide the image data into a plurality of sub-tasks through the main computing node and distribute the sub-tasks to the sub-computing node; the task execution unit is configured to control the sub-computing nodes to execute the sub-tasks in parallel and output a sub-task result; the model construction unit is configured to aggregate the subtask results through a main computing node, perform global optimization and texture mapping and construct a three-dimensional model of the target area; And the model monitoring unit is configured to switch the computing nodes to check the modeling progress or the derived result through the KVM switch platform.
3. The portable clustered three-dimensional rapid modeling system according to claim 2, wherein the data receiving unit is connected with an unmanned aerial vehicle acquisition device, the unmanned aerial vehicle acquisition device is used for acquiring image data of a target area, the data receiving unit is provided with a plurality of data interfaces, and the unmanned aerial vehicle acquisition device transmits the image data to a main computing node through the data interfaces.
4. The portable clustered three-dimensional rapid modeling system of claim 2 wherein the task schedule partitioning unit imports the image data into a master computing node and partitions the image data into a number of subtasks through the master computing node, comprising: importing the image data into a main computing node and preprocessing, wherein the preprocessing comprises image distortion correction, image enhancement and data index establishment; reading the corresponding position of each image data through the established data index, and generating a space range boundary of the target area; dividing a target area into a plurality of space grids by adopting a grid dividing method, counting the number of images in each space grid, and dynamically adjusting the space grids according to the number of images; An independent subtask is generated for the image data of each spatial grid, the subtask including the image data, the preprocessing parameters, and the task execution instructions.
5. The portable cluster three-dimensional rapid modeling system according to claim 4, wherein when the task scheduling dividing unit distributes the subtasks to the sub-computing nodes, the task scheduling dividing unit specifically comprises: acquiring real-time load and CPU computing power of sub-computing nodes, and determining task processing capacity of each computing node according to the real-time load and the CPU computing power; and determining the number of images of each subtask, and distributing the subtasks according to the number of images and the task processing capacity.
6. The portable clustered three-dimensional rapid modeling system of claim 5 wherein the task execution unit controls the sub-compute nodes to execute the sub-tasks in parallel, the task execution instructions including feature extraction, feature matching, sparse reconstruction, and dense point cloud generation.
7. The portable cluster three-dimensional rapid modeling system of claim 6, wherein the task scheduling division unit is further configured to monitor task execution data of each sub-computing node in real time as the sub-tasks are executed, the task execution data including CPU utilization, GPU utilization, memory occupancy, and task progress of the sub-computing node; judging whether the corresponding sub-computing node fails according to the task execution data, and if so, reassigning the sub-tasks which are not completed by the sub-computing node.
8. The portable clustered three-dimensional rapid modeling system of claim 7 wherein the model building unit aggregates the subtask results through a master computing node, performs global optimization and texture mapping, builds a three-dimensional model of a target region, comprising: The method comprises the steps that a main computing node receives a subtask result uploaded by each sub computing node, wherein the subtask result comprises sparse point clouds, dense point clouds and image features of each space grid; Carrying out coordinate unification processing on the received subtask results, and converting sparse point clouds and dense point clouds generated by different sub-computing nodes under a local coordinate system into a global coordinate system; Global registration optimization is carried out based on the unified sparse point cloud and dense point cloud, and the point cloud edges of adjacent space grids are aligned through an iterative nearest point algorithm, so that a splicing gap is eliminated; and according to the mapping relation between the preprocessed image data and the sparse point cloud and the dense point cloud under the global coordinate system, performing texture coordinate calculation, attaching the image features to the surface of the three-dimensional point cloud, and generating a three-dimensional model of the target area.
9. The portable clustered three-dimensional rapid modeling system of claim 8 wherein the model monitoring unit switches computing nodes to view modeling progress or derived results through a KVM switch, comprising: the method comprises the steps that a KVM switching station is switched to a monitoring interface of a main computing node, and the task execution state of each sub computing node is displayed in real time, wherein the task execution state comprises the number of completed sub tasks, the progress percentage of the currently performed sub tasks and the predicted remaining time; simultaneously supporting a user to view the construction process preview of the global three-dimensional model through a display, and performing rotation, scaling and translation operations on the preview model through a keyboard and a mouse; after the modeling task is completed, a user sends a deriving instruction through the man-machine interaction module, the model monitoring unit packages the three-dimensional model data according to a preset format, and derives the three-dimensional model data to a designated position through external storage equipment or a network transmission mode.
10. Portable clustered three-dimensional rapid modeling method applied to a portable clustered three-dimensional rapid modeling system according to any one of claims 1 to 9, characterized in that it comprises: Receiving image data of a target area, importing the image data into a main computing node, dividing the image data into a plurality of sub-tasks through the main computing node, and distributing the sub-tasks to the sub-computing node; The sub-computing node is controlled to execute the sub-tasks in parallel, and a sub-task result is output; the subtask results are aggregated through a main computing node, global optimization and texture mapping are carried out, and a three-dimensional model of a target area is built; And switching the computing nodes to check the modeling progress or the derived result through the KVM switch.

Description

Portable cluster three-dimensional rapid modeling system and method Technical Field The invention relates to the technical field of modeling systems, in particular to a portable cluster three-dimensional rapid modeling system and method. Background With the rapid development of unmanned aerial vehicle remote sensing, oblique photography and computer vision technology, three-dimensional modeling based on image data is widely applied to the fields of emergency mapping, disaster assessment, urban planning, military reconnaissance and the like. The traditional three-dimensional modeling system generally depends on a fixed high-performance computing center or cloud server, has the problems of long deployment period, poor environmental adaptability, low response speed and the like, and is difficult to meet urgent requirements for quick, efficient and portable modeling capability in complex scenes such as field operation, disaster sites and the like. The existing portable modeling equipment mostly adopts a single-machine architecture, is limited by computing resources and heat dissipation capacity, and is difficult to support real-time processing of large-scale image data, while the distributed cluster scheme has strong computing power, but is huge in size, high in power consumption and inconvenient in man-machine interaction, and lacks an environment adaptation mechanism and task scheduling optimization strategy aiming at a mobile operation scene. In addition, in the modeling process, if a certain computing node fails, the conventional system usually lacks a dynamic fault tolerance and task redistribution mechanism, which is easy to cause modeling interruption or incomplete result. Therefore, a portable three-dimensional modeling system integrating high computing power, strong robustness, good man-machine interaction and environment adaptation capability is needed. Disclosure of Invention The invention aims to provide a portable cluster three-dimensional rapid modeling system and method, which solve the problems that the existing portable modeling equipment adopts a single machine architecture, is difficult to support real-time processing of large-scale image data, and a distributed cluster is huge in size, high in power consumption and inconvenient in man-machine interaction, and lacks an environment adaptation mechanism and a task scheduling optimization strategy aiming at a mobile operation scene. The invention provides a portable cluster three-dimensional rapid modeling system, which comprises: The integrated chassis unit is internally integrated with a plurality of computing nodes, the computing nodes are interconnected through a back plate, and each computing node comprises a main computing node and a sub computing node; the power supply and power management module is used for monitoring the load of each computing node in real time and dynamically adjusting the power supply; The human-computer interaction module is arranged on the integrated chassis unit and comprises a KVM (keyboard-video-mouse) switching table, a keyboard, a mouse and a display, wherein the keyboard, the mouse and the display are connected with each computing node through the KVM switching table, and the KVM switching table is used for supporting a user to switch among a plurality of computing nodes through the keyboard or the mouse; the scheduling and modeling module is used for carrying out three-dimensional modeling on the target area according to the image data; The environment adaptation module is arranged in the integrated chassis unit and comprises an air cooling unit and a heat dissipation unit, and the environment adaptation module is used for monitoring and intelligently regulating the internal environment temperature and humidity of the integrated chassis unit in real time. Preferably, the scheduling and modeling module comprises a data receiving unit configured to receive image data of a target area; the task scheduling dividing unit is configured to guide the image data into a main computing node, divide the image data into a plurality of sub-tasks through the main computing node and distribute the sub-tasks to the sub-computing node; the task execution unit is configured to control the sub-computing nodes to execute the sub-tasks in parallel and output a sub-task result; the model construction unit is configured to aggregate the subtask results through a main computing node, perform global optimization and texture mapping and construct a three-dimensional model of the target area; And the model monitoring unit is configured to switch the computing nodes to check the modeling progress or the derived result through the KVM switch platform. Preferably, the data receiving unit is connected with unmanned aerial vehicle acquisition equipment, unmanned aerial vehicle acquisition equipment is used for gathering the image data of target area, the data receiving unit is provided with a plurality of data interfaces, unmanned aerial veh