Concrete dams are a dam type commonly used in large-scale hydraulic engineering projects; the detection of their reinforcement mesh configurations during construction is fundamental to quality control and the application of intelligent equipment. However

for multi-layer reinforcement in high-noise environments

previous studies have struggled to achieve high-accuracy perception and recognition. This study presents a new intelligent perception and recognition method of high accuracy for such reinforcement mesh structures utilizing 3D LiDAR technology. First

we develop a multi-stage data denoising and preprocessing method based on SOR-DBSCAN-Tensor Voting to enhance the quality and usability of raw data. Then

we adopt the MLESAC algorithm and weighted least squares to formulate a progressive procedure for refined fitting of reinforcement meshes. Finally

a new method for plane fitting of dual-layer reinforcement meshes based on 2D projection MLESAC is implemented to tackle data loss caused by occlusion. And

by integrating this method with the point cloud density maps

the spatial position of the mesh is determined

realizing an effective use of incomplete point cloud data. In a case study of the Tuxikou reservoir

numerical experiments demonstrate our method is effective in leveraging the LiDAR equipment and has achieved refined fitting and reconstruction of the dual-layer reinforcement mesh structures under high-noise conditions

useful for construction site quality control and intelligent equipment application.