Search

Individual tree detection and counting in unmanned aerial vehicle (UAV) imagery constitute a vital and practical research field. Vegetation remote sensing captures large-scale trees characterized by complex textures, significant growth variations, and high species similarity within the vegetation, which presents significant challenges for annotation and detection. Existing methods based on bounding boxes have struggled to convey semantics information about tree crowns. This paper proposes a novel deep learning network called VrsNet based on the density map information. The proposed work pioneers the segmentation and counting application by utilizing the semantic information of Gaussian contour. Besides, we sample and create the UAV vegetation remote sensing density dataset TreeFsc for experiments. In quantitative comparison across multiple datasets, the proposed method demonstrates high performance, with a 3.45 increase in MAE and a 4.75 increase in RMSE. Experiments demonstrate superior cross-region, cross-scale, and cross-species target detection capabilities of the proposed approach compared with the existing object detection methods. Our code and dataset are available at: https://github.com/luotiger123/VrsNet/tree/main/VrsNet.

In computer-controlled optical surfacing (CCOS), the paths of the lap tools are limited inside the optical surface; this restricts convolution in the dwell-time algorithm and causes mid-spatial-frequency surface errors. An optimized strategy ensuring relatively complete convolution of the dwell-time algorithm is developed to control the mid-spatial-frequency surface error and simultaneously ensure high optical manufacturing efficiency. Different-sized lap tools are then introduced to correct the surface error in different areas of the optics. Simulations and experiments using a large off-axis SiC mirror demonstrate the validity of the strategy, and it could be widely applied to CCOS in grinding or polishing processes.

Magnetorheological finishing (MRF) fabrication can effectively facilitate a rapid fabrication of an aspherical 2 m reaction-bonded silicon carbide mirror. The dwell-time algorithm and tool path supported tool-mark mitigation and virtual-axis employment are analyzed. A rapid-fabrication strategy alternates MRF and a large polishing lap to converge surface error profile before MRF alone is used to high-precise finishing. In the alternate-use process, large polishing lap’s dwell time map for low-order space frequency is calculated before MRF’s for high-order one, but executed later in reality. The fabricated mirror with a silicon modification layer showed accuracy convergence from 0.098 λ rms to 0.019 λ rms at 84.6 h, demonstrating the strategy’s validity for large optical-surface processing. Keywords: Computer controlled polishing, Silicon carbide mirror, Surface error, Magnetorheological finishing, Large polishing lap