CN-121837990-B - Construction progress monitoring method based on unmanned aerial vehicle aerial photography

CN121837990BCN 121837990 BCN121837990 BCN 121837990BCN-121837990-B

Abstract

The invention relates to the technical field of construction progress monitoring, in particular to a construction progress monitoring method based on unmanned aerial vehicle aerial photography, which comprises the steps of acquiring image information through an unmanned aerial vehicle and determining a monitoring area of a panoramic image at each angle; the method comprises the steps of obtaining a panoramic image, carrying out matching processing on monitoring areas with different time sequences based on the visual field range and pitch angle of the panoramic image under different visual angles to obtain a matching feature set, configuring a trigger point and a trigger condition for each monitoring area with updated data based on semantic information of the matching feature set under each matching, checking the semantic information of the trigger point image by image frame to generate a progress deviation curve of each monitoring area, comparing the progress deviation curve of the current time sequence with the progress deviation curve of a preamble time sequence to form a progress advancing trend, dividing advancing stages and management priorities of each monitoring area, carrying out secondary verification on each monitoring area, and regarding checked data as an output construction progress list. The accuracy and the efficiency of construction progress monitoring are improved.

Inventors

RONG JIAN
GENG KAI
LIU SHUANGFEI
ZHEN ZHEN
GU YONGDING
DAI WEI
WANG WENBIN
WANG XUN

Assignees

安徽建工长丰建设工程有限公司
安徽建工水利开发投资集团有限公司

Dates

Publication Date: 20260508
Application Date: 20260310

Claims (9)

1. The construction progress monitoring method based on unmanned aerial vehicle aerial photography is characterized by comprising the following steps of: Acquiring panoramic images of a current construction site at different moments by an unmanned aerial vehicle through image information acquisition, and determining monitoring areas of the panoramic images at all angles in a differential image superposition mode according to the actual layout of the construction site; based on the view field range and pitch angle of the panoramic image under different view angles, matching the monitoring areas with different time sequences to obtain a matching characteristic set of the monitoring areas under each view angle; based on the semantic information of the matching feature set under each matching, configuring a trigger point and a trigger condition for updating data for each monitoring area, checking the semantic information of the trigger point image frame by image frame, and generating a progress deviation curve of each monitoring area; Determining the construction progress of a current monitoring area by combining semantic information corresponding to trigger points from the position of each trigger point, comparing the image difference value of continuous image frames with the average gray value corresponding to the trigger point and the pixel point change proportion to determine whether the image difference value is consistent with the current construction progress or not, if the image difference value is inconsistent with the current construction progress, taking the number of pixels corresponding to the image difference value as a check basis, calling a pixel template in the current construction state, counting the number of pixels corresponding to the pixel template under each image frame, if the number of pixels matched with the continuous image frames is consistent with the number of pixels matched with the pixel template, recalculating the image difference value of the corresponding image frames, outputting the recalculated data, otherwise marking the corresponding image frames as output data by using the data except the image frames, and combining the semantic information and the image difference value in the output data as output trigger conditions if the image difference value is inconsistent with the current construction progress; Comparing the progress deviation curve of the current time sequence with the progress deviation curve of the preamble time sequence to form a progress advancing trend, and dividing advancing stages and management priorities of all monitoring areas; and (3) carrying out secondary verification on each monitoring area based on the trigger point corresponding to each propulsion stage, checking semantic errors when each monitoring area is updated, and taking checked data as an output construction progress list.
2. The method for monitoring the construction progress based on unmanned aerial vehicle aerial photography according to claim 1, wherein when the differential images are superimposed, the implementation manner further comprises: based on a characteristic point stitching algorithm, stitching the acquired image information to generate a panoramic image sequence; based on the obtained panoramic image sequence, superposing a vector image layer on the panoramic image, and dividing the construction site into a plurality of monitoring areas through a polygon segmentation algorithm; and carrying out differential calculation on panoramic images at adjacent moments in a monitoring area acquired at each view angle by using the panoramic image sequence, generating a binarized differential image, and stacking the differential image as a next layer of image into the original panoramic image.
3. The method for monitoring the construction progress based on unmanned aerial vehicle aerial photography according to claim 1, wherein the implementation manner for obtaining the matched feature set of the monitoring area at each view angle comprises the following steps: determining coordinates of each monitoring area in the panoramic image corresponding to the spherical coordinate system according to a spherical mapping mode of the panoramic image, and determining the visual field range and the pitch angle of the panoramic image under different visual angles according to a pitch angle distinguishing mode; based on the view field range and pitch angle of the panoramic image under different view angles, acquiring characteristic information corresponding to each view angle in a layering manner, and constructing view angle labels corresponding to each monitoring area; locking a pixel change sub-region in the monitoring region based on the differential image overlapped by each monitoring region; performing association mapping on each pixel change sub-region and the view angle label, determining the level of each pixel change sub-region, and connecting the pixel change sub-regions according to the corresponding levels; and carrying out semantic statistics on the connected pixel change subareas, and taking the data after the semantic statistics as an output matching feature set.
4. The method for monitoring the construction progress based on unmanned aerial vehicle aerial photography according to claim 3, wherein when the characteristic information corresponding to each view angle is obtained in a layered manner, the implementation mode further comprises: extracting boundary features of the corresponding monitoring area according to the maximum visual field range of the monitoring area under the corresponding visual angle aiming at the visual field range of the monitoring area under the corresponding visual angle; Determining building components corresponding to boundary features based on the positions of the boundary features under different view angles; And connecting boundary features under all view angles according to the relative positions and the connection relation between the building components to complete the configuration of multi-level feature information.
5. The method for monitoring the construction progress based on unmanned aerial vehicle aerial photography according to claim 1, wherein the implementation mode of the progress deviation curve comprises the following steps: determining a monitoring area with construction state change by taking semantic information of a matched feature set as a guide, and configuring a trigger point in updating by taking an average gray value of pixels around each pixel in the monitoring area and a pixel change proportion of the current monitoring area; Checking the construction progress of the panoramic image in continuous image frames by comparing the panoramic images of the monitoring area under a plurality of view angles based on the positions of the trigger points, and configuring the trigger conditions corresponding to the trigger points; And connecting the construction state changes of the monitoring areas according to the time sequence based on the acquired trigger points and trigger conditions, and generating a progress deviation curve of each monitoring area.
6. The method for monitoring the progress of construction based on aerial photography of an unmanned aerial vehicle according to claim 5, wherein when generating the progress deviation curve of each monitoring area, the implementation manner further comprises: Based on the time information corresponding to each trigger point in the same monitoring area, calling the construction progress corresponding to each trigger point; Correlating the construction progress with the time difference of each construction state update, and comparing the construction progress of each trigger point with the deviation of the construction plan step by step; and the compared trigger points are sequentially connected according to the time sequence to be used as an output progress deviation curve.
7. The method for monitoring the construction progress based on unmanned aerial vehicle aerial photography according to claim 1, wherein the implementation manner of dividing the advancing stage and the management priority of each monitoring area comprises the following steps: time alignment is carried out on the progress deviation curve of the current time sequence and the progress deviation curve of the preamble time sequence, a progress pushing trend is constructed, and pushing stages of corresponding monitoring areas are divided according to construction procedures corresponding to the progress pushing trend; Based on the progress advancing trend of each monitoring area, the slope value of the progress advancing trend is used as a guide to deduce the construction progress process of each monitoring area; carrying out preliminary classification representation on each monitoring area according to the construction progress process, then carrying out spatial clustering according to the relative distance between each monitoring area, and setting a cluster corresponding to each monitoring area; and configuring the management priority of each monitoring area by comparing the dominant progress process among different clusters.
8. The method for monitoring the progress of construction based on unmanned aerial vehicle aerial photography according to claim 7, wherein when comparing dominant progress processes among different clusters, the implementation manner further comprises: Fitting the progress deviation of each monitoring area under multiple time periods aiming at the monitoring areas of the same cluster, and determining the average progress deviation of each monitoring area under the corresponding dominant progress process; And configuring the management priority of each monitoring area by combining the initial priority of each leading progress process according to the average progress deviation corresponding to each monitoring area.
9. The method for monitoring the construction progress based on unmanned aerial vehicle aerial photography according to claim 1, wherein the implementation manner of performing secondary verification on each monitoring area comprises the following steps: Aiming at a trigger point in any propulsion stage, cross analysis is carried out by combining the adjacent areas of the current monitoring area by taking the monitoring area where the trigger point is positioned as a reference, and whether semantic errors exist in the corresponding monitoring area is determined; If the semantic error exists in the updating process of the monitoring area, correcting the monitoring area by using the minimum cost and reliability constraint corresponding to the current monitoring area, and determining corrected data; And if no semantic error exists, archiving all the data and then taking the data as an output construction progress list.

Description

Construction progress monitoring method based on unmanned aerial vehicle aerial photography Technical Field The invention relates to the technical field of construction progress monitoring, in particular to a construction progress monitoring method based on unmanned aerial vehicle aerial photography. Background During construction progress monitoring, unmanned aerial vehicles are used for collecting aerial images of construction sites at different moments, changes of construction areas are recorded by means of image identification, construction progress completion conditions are judged by combining construction plan comparison, but in the process of monitoring construction progress, the most main data sources are image information and corresponding logs, if the construction logs do not accurately record construction progress change conditions, the false identification of the whole construction progress can be caused, further follow-up construction conditions are affected, and the problems of construction progress stagnation and the like are caused. For example, chinese patent publication No. CN119027069a discloses a photovoltaic project construction progress recognition method based on unmanned aerial vehicle images and neural networks. The method comprises the steps of dividing a photovoltaic project factory into a plurality of square matrixes, taking the square matrixes as units, shooting construction progress field images according to a preset route task based on an unmanned aerial vehicle, acquiring an orthographic image corresponding to each square matrix based on a related orthographic image generation algorithm, obtaining a proper photovoltaic project construction progress picture by performing processing procedures such as cutting, format conversion and the like on the obtained image, recognizing a project progress recognition model obtained by training based on an optimized YOLT algorithm model, and finally obtaining the actual project construction progress by comparing the recognition quantity and the design quantity of pile foundations, photovoltaic supports and photovoltaic modules under the square matrixes. In the construction process of any engineering project, according to each sub-project task of each task level divided by the engineering project, the method respectively obtains the construction progress of each sub-project task in each engineering day, obtains the overall construction progress time sequence of the engineering project according to the construction progress of each sub-project task in each engineering day, obtains the fitting weight corresponding to the engineering date, optimizes the overall construction progress time sequence according to the fitting weight corresponding to each engineering date contained in the overall construction progress time sequence, obtains the optimized overall construction progress time sequence, and predicts the construction progress of the engineering project according to the optimized overall construction progress time sequence. The construction state of the orthographic image is recognized, the project construction progress is recognized according to the number of corresponding components in the state recognition result, or the construction progress is predicted in a time sequence analysis mode according to window change of engineering date, the prior art is biased to analysis processing under a single data scene, other areas under the construction scene are easy to ignore, continuous area conflict and area data interference problems can exist in the construction state under the scene recognition, the area data statistical distortion problem is caused, and the accuracy of the construction progress summarization under the construction scene is affected. Disclosure of Invention In order to solve the technical problems, the technical scheme includes that the construction progress monitoring method based on unmanned aerial vehicle aerial photography comprises the steps of acquiring panoramic images of a current construction site at different moments through image information acquisition of unmanned aerial vehicles, and determining monitoring areas of the panoramic images at all angles in a differential image superposition mode according to actual layout of the construction site. And carrying out matching processing on the monitoring areas with different time sequences based on the view field range and the pitch angle of the panoramic image under different view angles, and obtaining a matching characteristic set of the monitoring areas under each view angle. Based on the semantic information of the matching feature set under each matching, configuring the trigger points and trigger conditions for each monitoring area with data update, checking the semantic information of the trigger points image frame by image frame, and generating a progress deviation curve of each monitoring area. And comparing the progress deviation curve of the current time sequ