Research On Recognition And Spatial Location Method Of Construction Land Based On Tower-based Monitoring Image

In view of the low efficiency of manual duty in current land space supervision,this paper takes construction land as the main research object and proposes a set of intelligent method for construction land target recognition and spatial location based on the existing tower-based monitoring equipment supervision.This method cross-applies the target detection and recognition model in the field of deep learning,the stereo matching and the three-dimensional spatial location model in the field of digital photogrammetry.It mainly analyzes the theories and methods related to the integrated recognition and spatial location of the construction land targets in monitoring images,and the simulation experiment under the high-tower scene was carried out in the construction site.It not only verifies the effectiveness of the integration of geographic target recognition model and spatial location model,but also clarifies the hardware design framework of the model integration,and finally obtains valuable experimental results.The main work and research contents of this paper are as follows:(1)Definition and accurate recognition of construction land targets.In this paper,the characteristics of the site in scale-based monitoring image are summarized,and the concept of construction land object is defined in detail.Based on YOLO(You Only Look Once)model,the training method of target recognition model of construction land is summarized.The effective recognition model is obtained through multiple training iterations and continuous hyperparametric adjustment,which can quickly and accurately identify the target area of construction land in the GPU-supported environment.(2)Accurate spatial location of geographic target areas.Aiming at the spatial location method of geographic target areas under high-tower monitoring,this paper constructs a three-dimensional spatial location model SITCOL(Space Intersection of Two Cameras and One Location)for the calibrated binocular camera and the target based on the dual-image analytical photogrammetry theory.By acquiring the stereo pair of the binocular camera and the corresponding internal and external orientation parameters,the model realizes the accurate spatial location of the target area based on the relationships between the solid geometry.(3)Construction and implementation of geographic target recognition and spatial location combination model(YOLO-SITCOL).With the binocular advantage of the binocular camera,this paper uses the left eye to recognise the geographic target,the right eye to match the same target in the left eye,and finally uses the binocular to locate the geographic target.The research realizes the integration of geographic target recognition model YOLO and spatial location model SITCOL,and the process of geographic target recognition and spatial location.(4)Experimental verification for the validity of geographic target recognition and spatial location combination model.In this paper,the simulation experiment of combination model is carried out with the excavator as the target.The pre-calibrated binocular camera is used to simulate the high-tower shooting scene,and the excavator targets are recognised and spatially located in an integrated way.Finally,the calculated coordinate values of the excavator target are obtained by the combination model in the custom coordinate system,and they are compared with the real values to verify the effectiveness of the method in this paper.