亚像素渲染
质心
校准
材料科学
光学
体积热力学
噪音(视频)
激光器
干扰(通信)
直线(几何图形)
点(几何)
准确度和精密度
计算机科学
杂散光
信号(编程语言)
激光扫描
计量系统
人工智能
计算机视觉
测量不确定度
理论(学习稳定性)
萃取(化学)
观测误差
信噪比(成像)
降噪
结构光
固定模式噪声
声学
近红外光谱
图像质量
作者
Ling Cao,R. L. Zhou,Xinhua Wang,Wei Pan
标识
DOI:10.1088/1361-6501/ae1a9f
摘要
Abstract Accurate 3D volume measurement of transparent adhesives is essential for ensuring dispensing quality and production efficiency in precision manufacturing. We present a robust line laser scanning system enhanced by automatic camera calibration and complementary spectral background optimization (CSBO), which improves 3D reconstruction accuracy by suppressing transmitted light and internal reflections—without complex ray-tracing models. A novel subpixel laser stripe centroid extraction algorithm is further introduced, integrating a candidate point mechanism with a dynamic weighted multi-factor scoring strategy to enhance stability and noise robustness. Experiments show that the proposed centroid extraction method surpasses conventional approaches in precision and interference resistance, while the CSBO method achieves a minimum relative volume error of 1.62% against ground truth, enabling real-time industrial measurement of transparent materials.
科研通智能强力驱动
Strongly Powered by AbleSci AI