CN-121985115-A - Video quality intelligent verification method and system for grouting construction

CN121985115ACN 121985115 ACN121985115 ACN 121985115ACN-121985115-A

Abstract

The invention relates to the technical field of construction video quality verification, in particular to an intelligent video quality verification method and system for grouting construction. According to the method, based on component edge characteristic information in a video frame sequence to be detected, shooting angles of component plane normal vectors and the direction of an optical axis of a camera are calculated, whether the shooting angles meet grouting recognition requirements is judged, when the grouting recognition requirements are met, boundary contours of target hole site areas in all clear video frames are recognized, local visible integrity of the target hole site areas is calculated, integral visibility of an acquired video is judged, integral translational motion quantity and rotational motion quantity between all adjacent frames are extracted to form an acquired video stability index, integral visibility of the acquired video and the acquired video stability index are combined, comprehensive quality of the acquired video is evaluated, video access results are output based on the comprehensive quality of the acquired video, the technical requirements of AI intelligent recognition are met for the acquired video, and the usability of the acquired video is improved.

Inventors

ZHANG YISHENG
CHENG WEIGUO
HU NAN
JIANG ZHIJIE
LIU YUBO

Assignees

中铁四局集团有限公司
安徽数智建造研究院有限公司

Dates

Publication Date: 20260505
Application Date: 20260408

Claims (10)

1. The intelligent video quality verification method for grouting construction is characterized by comprising the following steps of: Extracting a video frame sequence to be detected from a video acquired from a grouting construction site, calculating a shooting angle between a component plane normal vector and the optical axis direction of a camera based on component edge characteristic information in the sequence, and judging whether the shooting angle meets grouting identification requirements; When the grouting identification requirement is met, carrying out uniform analysis on definition and brightness of all video frame images in a video frame sequence to be detected, and screening all the conforming clear video frames; Identifying boundary contours of all target hole site areas in all clear video frames, constructing an area visible distribution matrix through grid division, calculating the local visible integrity of all target hole site areas, and judging the overall visibility of the acquired video; acquiring pixel motion vector fields between all adjacent frames, and extracting the overall translational motion quantity and the rotational motion quantity between each adjacent frame to form an acquisition video stability index; based on the overall visibility of the acquired video and the acquired video stability index, the comprehensive quality of the acquired video is evaluated, and a video input result is output based on the comprehensive quality of the acquired video.
2. The intelligent video quality verification method for grouting construction of claim 1, wherein the step of judging whether the shooting angle meets the grouting identification requirement is as follows: Extracting frame images of the video frame sequence to be detected according to a preset time interval to obtain all the video frames to be analyzed; Performing edge pixel detection on each video frame to be analyzed, extracting component edge line segment sets in all the video frames to be analyzed, performing linear screening on the component edge line segment sets, and determining horizontal edge direction vectors and vertical edge direction vectors of components; Obtaining a normal vector of a component plane through vector cross multiplication operation, and calculating a shooting angle between the normal vector of the component plane and the direction of the optical axis of the camera by combining a preset vector of the direction of the optical axis of the camera; if the shooting angle meets the preset grouting AI identification corresponding shooting angle qualification judgment standard, judging that the shooting angle meets the grouting identification requirement, otherwise, judging that the shooting angle does not meet the grouting identification requirement, and outputting an admission result with unqualified acquired video.
3. The intelligent verification method for video quality for grouting construction according to claim 2, wherein the horizontal edge direction vector and the vertical edge direction vector of the member are determined in the following manner: Carrying out straight line recognition on the component edge line segment set, and extracting the corresponding horizontal direction edge line segment and vertical direction edge line segment of the component; Acquiring lengths and directions of the horizontal edge line segments and the vertical edge line segments corresponding to the members, and filtering the interference line segments based on the set sizes of the grouting sleeves corresponding to the grouting construction; And carrying out direction consistency analysis on the filtered horizontal edge line segment and the vertical edge line segment, and fitting to generate a horizontal datum line and a vertical datum line of the component to obtain a horizontal edge direction vector corresponding to the horizontal edge datum line and a vertical edge direction vector corresponding to the vertical datum line.
4. The intelligent video quality verification method for grouting construction of claim 1, wherein the step of screening all the clear video frames which meet the method is as follows: Carrying out graying treatment on each video frame image in a video frame sequence to be detected, counting brightness distribution characteristics of each video frame image based on a gray level histogram, and judging whether abnormal exposure exists in the video frame image according to the brightness distribution characteristics; if no abnormal exposure exists in a certain video frame image, the video frame image is marked as a video frame with uniform brightness; performing convolution operation on all the uniform-brightness video frames to obtain Laplacian response values of the uniform-brightness video frames, and calculating the variance to obtain the definition of each uniform-brightness video frame; and eliminating the uniform-brightness video frames with the definition not reaching the standard, and counting all the conforming clear video frames.
5. The intelligent verification method for video quality for grouting construction according to claim 1, wherein the calculation process of the local visible integrity of each target hole site area is as follows: Identifying sleeve grouting holes, grouting holes and plugging holes in all clear video frames to correspond to target hole site areas, and obtaining boundary position information of each target hole site area; dividing the boundary contour in the boundary position information of each target hole site area into a plurality of subarea grids, counting the covering area of the shielding object of each subarea grid, and calculating the visible coefficient of each subarea grid by combining the boundary contour area of each subarea grid; according to the space position of each subarea grid, arranging the visible coefficients of all subarea grids to construct an area visible distribution matrix; And calculating the local visible integrity of each target hole site region by combining the corresponding preset spatial position weight of each sub-region grid.
6. The intelligent verification method for video quality for grouting construction of claim 5, wherein the overall visibility determination process of the collected video is as follows: if the local visible integrity of a certain target hole site area in a certain clear video frame is lower than the basic hole site visible integrity threshold, marking the clear video frame as a hole site invisible video frame, and taking the target hole site area as an invisible hole site area; And when all the invisible hole site video frames are not continuous or the invisible hole site areas corresponding to the invisible hole site video frames are different, judging that the overall visibility of the acquired video is qualified, otherwise, judging that the overall visibility of the acquired video is unqualified.
7. The intelligent video quality verification method for grouting construction of claim 1, wherein the overall translational motion amount extraction mode is as follows: gray processing is carried out on each clear video frame to obtain positions of all pixel points, motion vectors of all pixel points between all adjacent frames are obtained through position comparison, and a pixel motion vector field between all adjacent frames is generated; Extracting motion vectors of all pixel points from a pixel motion vector field, decomposing the motion vectors, and generating horizontal displacement components and vertical displacement components of all the pixel points; Counting the horizontal displacement components and the vertical displacement components of all pixel points, determining the central tendency representation values of the horizontal displacement components and the vertical displacement components, and respectively marking the central tendency representation values as the integral horizontal translation quantity between frames and the integral vertical translation quantity between frames; vector synthesis is carried out on the inter-frame integral horizontal translation quantity and the inter-frame integral vertical translation quantity to obtain integral translation motion quantity between each two adjacent frames.
8. The intelligent video quality verification method for grouting construction of claim 7, wherein the rotation motion amount extraction mode is as follows: establishing a polar coordinate system by taking the center of a video frame as an origin, and dividing the video frame into a plurality of concentric ring areas according to different radial distances; According to the motion vector of each pixel point in the pixel motion vector field, acquiring a radial component and a tangential component of the motion vector of each pixel point, and carrying out average value calculation on the tangential components of all the pixel points in each concentric ring area to obtain the average tangential speed of each concentric ring area; carrying out ratio analysis on the average tangential velocity of each concentric ring region and the corresponding radius distance to obtain the local rotation angular velocity of each concentric ring region, and comparing to generate the estimated rotation angular velocity of the video frame; performing product operation on the estimated rotation angular velocity of the video frame and the time interval between adjacent frames to obtain the rotation angle variation between frames, and taking the rotation angle variation as the rotation motion quantity between each adjacent frame; and taking the integral translational motion quantity and the rotational motion quantity between every two adjacent frames as acquired video stability indexes.
9. The intelligent verification method for video quality for grouting construction according to claim 8, wherein the video admittance result output content is: When the overall visibility of the collected video is unqualified, judging that the comprehensive quality of the collected video is unqualified, and outputting a video admission result as failed; When the overall visibility of the acquired video is qualified, comparing the acquired video stability index with a maximum allowable position threshold value and a maximum allowable angle threshold value corresponding to the video stability; If any motion quantity in the acquired video stability index exceeds a corresponding threshold value, judging that the comprehensive quality of the acquired video is unqualified, and outputting a video input result as failed; Otherwise, judging the comprehensive quality of the acquired video to be qualified, and outputting a video admission result to pass.
10. A video quality intelligence check-up system for grout construction, its characterized in that: the shooting angle acquisition module is used for extracting a video frame sequence to be detected from a video acquired from a grouting construction site, calculating a shooting angle between a component plane normal vector and the optical axis direction of a camera based on component edge characteristic information in the sequence, and judging whether the shooting angle meets grouting identification requirements or not; The clear video frame screening module is used for carrying out uniform analysis on definition and brightness of all video frame images in a video frame sequence to be detected when the grouting identification requirement is met, and screening all the conforming clear video frames; The visibility judging module is used for identifying boundary outlines of all target hole site areas in all clear video frames, constructing an area visible distribution matrix through grid division, calculating the local visible integrity of all the target hole site areas and judging the overall visibility of the acquired video; the stability index forming module is used for obtaining pixel motion vector fields between all adjacent frames, extracting the integral translational motion quantity and the rotational motion quantity between each adjacent frame and forming an acquired video stability index; And the video admission evaluation module is used for evaluating the comprehensive quality of the acquired video based on the overall visibility of the acquired video and the acquired video stability index and outputting a video admission result based on the comprehensive quality of the acquired video.

Description

Video quality intelligent verification method and system for grouting construction Technical Field The invention relates to the technical field of construction video quality verification, in particular to an intelligent video quality verification method and system for grouting construction. Background In order to ensure traceability and quality controllability of the grouting construction process of the sleeve, the whole grouting operation process is generally recorded in a video monitoring mode on a construction site, and the grouting quality is identified and judged through intelligent analysis of video data. However, the quality of the video determines the accuracy and reliability of the subsequent AI identification, and if the video picture has problems of deviation of shooting angle, blurring of picture, shielding of key area or severe shaking, the identification algorithm cannot effectively extract the target feature, which causes quality misjudgment or omission, and thus creates a great hidden trouble for engineering safety burial. In the prior art, manual handheld shooting is usually adopted when grouting construction video acquisition is carried out, but the conditions of excessively small angle, side shooting, inclination and the like easily occur in manual random shooting, so that the follow-up intelligent recognition is difficult to extract the characteristic information of target areas such as grouting holes, grout outlet holes and the like, and the quality judgment standard is disjointed from the intelligent recognition requirement, so that recognition failure or the misjudgment rate of grouting quality is improved. Meanwhile, people walk, equipment move and the like in the construction process often block the grouting hole, so that the slurry state is invisible, video data in the shielding period is still used for quality identification, the reliability of quality identification results is reduced, the standardization of the grouting construction process cannot be effectively analyzed, an error video basis is provided for construction quality supervision, and further follow-up grouting quality responsibility disputes are caused. Disclosure of Invention The invention aims to provide an intelligent video quality verification method and system for grouting construction, which are used for solving the problems in the prior art. In order to achieve the aim, the invention adopts the following technical scheme that the invention provides an intelligent video quality verification method for grouting construction, which comprises the steps of extracting a video frame sequence to be detected from a video acquired at a grouting construction site, calculating a shooting angle of a component plane normal vector and a camera optical axis direction based on component edge characteristic information in the sequence, and judging whether the shooting angle meets grouting identification requirements. When the grouting identification requirement is met, performing uniform definition and brightness analysis on all video frame images in the video frame sequence to be detected, and screening all the conforming clear video frames. And identifying boundary outlines of all target hole site areas in all clear video frames, constructing an area visible distribution matrix through grid division, calculating the local visible integrity of all target hole site areas, and judging the overall visibility of the acquired video. And acquiring pixel motion vector fields between all adjacent frames, and extracting the overall translational motion quantity and the rotational motion quantity between each adjacent frame to form an acquisition video stability index. Based on the overall visibility of the acquired video and the acquired video stability index, the comprehensive quality of the acquired video is evaluated, and a video input result is output based on the comprehensive quality of the acquired video. On the other hand, the invention provides an intelligent video quality verification system for grouting construction, which comprises a shooting angle acquisition module, a clear video frame screening module, a visibility judging module, a stability index forming module and a video admission evaluation module. The connection mode among the modules is that the shooting angle acquisition module is connected with the clear video frame screening module, the visibility judging module is respectively connected with the clear video frame screening module and the stability index forming module, and the video admission evaluation module is connected with the stability index forming module. The shooting angle acquisition module is used for extracting a video frame sequence to be detected from a grouting construction site acquisition video, calculating the shooting angle of a component plane normal vector and the direction of the optical axis of the camera based on component edge characteristic information in the sequence, and judging whe