华南理工大学学报(自然科学版)

• 机械工程 • 上一篇    下一篇

基于双目结构光与融合约束的超薄微热管厚度检测

李勇1 周仔涵 周文杰 欧其罐 冯丹艳3   

  1. 1.华南理工大学 机械与汽车工程学院,广东 广州 510640;

    2. 广东技术师范大学 机电学院,广东 广州 510450;

    3. 广东理工学院 智能制造学院,广东 肇庆 526100

  • 发布日期:2026-04-03

Thickness Measurement for Ultra-Thin Micro-Heat Pipes Based on Binocular Structured Light and Fused Constraints

Li Yong Zhou Zihan  Zhou Wenjie Ou Qiguan1   Feng Danyan3   

  1. 1. School of Mechanical and Automotive Engineering, South China University of Technology, Guangzhou 510640,Guangdong, China;

    2. School of Mechatronic Engineering, Guangdong Polytechnic Normal University, Guangzhou 510450, Guangdong, China;

    3. School of Intelligent Manufacturing, Guangdong Technology College, Zhaoqing 526100, Guangdong, China

  • Published:2026-04-03

摘要:

为实现超薄微热管厚度的非接触、自动化检测,提出了一种基于双目结构光与融合约束的超薄微热管厚度检测方法。首先,针对超薄微热管的结构尺寸与表面特性,对外接矩形边长不超过150 mm的超薄微热管构建双目结构光视觉测量系统,通过高精度标定将重投影误差控制在0.08像素以内;其次,采用三频四步相移结构光对被测表面进行编码,基于多频外差原理实现相位阶数与绝对相位的稳定求解;进一步,在极线矫正基础上提出一种融合相位阶数一致性约束与视差连续性约束的立体匹配优化算法,构建联合代价聚合函数,有效提升弱纹理区域的匹配精度与稳定性;最后,依据最小聚合代价逐点计算视差,结合标定参数恢复深度信息,并通过平面拟合实现厚度测量。实验结果表明,系统对标准阶梯块高度测量误差在±0.05 mm内,拟合平面高度标准差不超过0.022 mm,相对误差不超过1.4%,综合不确定度为0.024 mm;对100支超薄微热管进行检测,结果与人工分类一致,进一步对比分析各分类结果的检测数据与人工检测数据,绝对误差为0.011 mm,最大误差不超过0.017 mm;精度方面,该方法的平面拟合标准差较绝对匹配方法降低0.117 mm以上,效率方面,该法的平均用时较窗口匹配方法缩短43.6%,验证了该方法在超薄微热管厚度检测中具有良好的适应性。

关键词: 超薄微热管, 厚度检测, 双目视觉, 结构光, 融合约束

Abstract:

To achieve non-contact and automated thickness measurement of ultra-thin micro-heat pipes with an external rectangle side length not exceeding 150 mm, a method based on binocular structured light and fused constraints is proposed. First, a binocular structured-light vision measurement system was constructed according to the structural dimensions and surface characteristics of ultra-thin micro-heat pipes, with reprojection error controlled within 0.08 pixels through high-precision calibration. Second, the surface was encoded by using three-frequency four-step phase-shifting structured light, and a stable solution for the phase order and absolute phase was achieved based on the multi-frequency heterodyne principle. Furthermore, based on epipolar rectification, a stereo matching optimization algorithm integrating phase order consistency and disparity continuity constraints was proposed, and a joint cost aggregation function was constructed to effectively improve the matching accuracy and stability in weak-texture regions. Finally, the disparity was computed pointwise according to the minimal aggregated cost, depth information was recovered using calibration parameters, and thickness measurement was achieved via plane fitting. Experimental results demonstrate that the system achieves a height measurement error within ±0.05 mm on a standard step block, with a standard deviation of the fitted plane height not exceeding 0.022 mm, a relative error within 1.4%, and a combined uncertainty of 0.024 mm. The detection results of 100 actual ultra-thin micro heat-pipes were fully consistent with manual classification. Further comparative analysis was performed between the measured data of each category and the corresponding manual measurement results, which showed an absolute error of 0.011 mm and a maximum error not exceeding 0.017 mm. In terms of accuracy, the standard deviation of the plane fitting achieved by the proposed method was reduced by more than 0.117 mm compared to the absolute matching method. Regarding efficiency, the average processing time was shortened by 43.6% compared to the window-based matching method. These results validate the strong adaptability of the proposed method in thickness measurement of ultra-thin micro-heat pipes.

Key words: ultra-thin micro-heat pipes, thickness measurement, binocular vision, structured light, fused constraints