21世纪以来，制造业市场的竞争愈演愈烈，工业自动化领域日益蓬勃发展，随着人工成本的不断上升，对于需要大量劳动力的制造业来说，机器换人已然是当今的一大趋势。如今制造业蓬勃发展，计算机水平又不断攀升，而针对于工业检测方向的智能化应用研究却稍显不足。高精密加工制造是目前制造业前进的方向，然而高精密的制造亟需高精度的工业测量技术保驾护航。传统的测量手段由于效率低、成本高，逐渐被市场所淘汰，取而代之的是非接触测量手段，应用极为广泛的是激光测距方法和机器视觉法。然而机器视觉法由于图像较大导致传输速度慢、图像处理速度慢且不能实现较大视野范围内的高精度测量。而激光测距法具有高精度、高速度、空间小、能实现快速的数据处理等优点，更为重要的是，激光测距的方法能提供物体的空间信息，结合运动信息实现高精度的三维重构，实现三维特征的测量。激光三角法的自动化检测是近年来实现工业自动化检测的重要手段。由于工业场景复杂，对于传感器的应用以及相应的数据处理方法的稳定性、适用性、处理速度等都提出了较高的要求。本文主要针对其数据的误差分析与采集、系统搭建、以及尺寸测量应用和形貌测量应用等四个方面进行了研究。首先，本文从工业检测的需求与背景入手；然后，本研究针对于三角法传感器进行了深入分析，对于其原理和数据采集过程中的误差进行了系统分析；接下来，基于三角法传感器合理设计的装夹模块，配合高精度电动位移台搭建了高精度测量系统；之后，基于搭建的平台，系统地分析了其在尺寸测量过程中的校正方法以及对单一尺寸和多尺寸的测量实验的结果。最后，结合PCL开源库在数据采集后进行了预处理与潜在的工业测量应用研究，应用三维点云结合了多种降噪方法，实现了平面度以及异形面的测量。本研究的结果展示了激光三角法传感器在尺寸检测、表面形貌检测以及工业应用中的重要方法。实验结果表明对于点激光三角法传感器在单一尺寸测量时能够达到±1μm的精度，在多尺寸测量时其精度由传感器的线性度误差所决定。对于线激光三角法传感器能够实现高度信息<±5μm的测量精度。通过将传统滤波算法与分割算法相结合的降噪能够有效提高信噪比。结合RANSAC实现的平面度测量可以有效避免噪声信号对于测量结果的干扰，与传统平面度测量方法误差<±5μm。并通过多种匹配算法实现了对于异形面测量结果的评估。本论文的研究成果展示了激光三角法传感器在工业应用中的巨大潜力和广阔的应用前景。

Since the 21st century, competition in the manufacturing market has more and more fierce, and the field of industrial automation has been booming. With the rising labor costs today, machine substitution has become a major trend for manufacturing industries that require a large amount of labor. Nowadays, the manufacturing industry is booming, the computer level is rising as well, but researches on intelligent application for industrial measurement is slightly insufficient. High-precision manufacturing is currently the direction of the manufacturing development, but high-precision manufacturing requires high-precision industrial measurement technology to ensure the precision. Traditional measurement methods are gradually decreased in the market due to low efficiency and high cost. Instead, non-contact measurement methods are used, especially, laser measurement technology and machine vision methods are widely used. However, due to the large image size, the machine vision method has lost efficiency in transmission and image processing, which cannot achieve high-precision online measurement in a large field of view. The laser measurement technology has the advantages of high precision, high speed, small volume, and high efficiency of data processing. More importantly, the laser measurement technology can provide spatial information of objects and can achieve high precision in combination with motion information. It can realize 3D reconstruction and evaluation of 3D features.The automated detection and measurement of laser triangulation is an important methodology to achieve industrial automation measurement in recent years. Due to the complexity of the industrial scene, high requirements are placed on the application of the sensor and the stability, applicability and processing speed of the corresponding data processing. This paper mainly studied on four aspects: error analysis and data acquisition, system construction, dimension measurement application, surface measurement application. Firstly, this paper starts from the requirements and background of industrial measurement. Secondly, this study is aimed at in-depth analysis of laser triangulation sensor. And the error in the data acquisition process was systematically analyzed. Thirdly, the corresponding mechanical part based on the rational design of the triangulation sensor was used to build a high-precision measurement system combined with the high-precision electric stage. Fourthly, based on the platform built, its calibration method in the dimension measurement process and the results of measurement experiments on single size and multi-size was systematically analyzed. Fifthly, combined with the PCL open source project, the theoretical processing and potential application research after data acquisition. The application of 3D point cloud combined with a variety of noise reduction methods to realize 3D data reconstruction with low-noise. Moreover, flatness measurement and complex surface measurement are realized in this study.The results of this study demonstrate the importance of laser triangulation sensors in dimensional measurement, complex surface evaluation, and industrial applications. The experimental results show that the point laser triangulation sensor can achieve a precision of ±1 μm in a single size measurement, and the result of multi-size measurement is determined by the linearity error of the sensor. For line laser triangulation sensors, measurement precision of height information <±5μm can be achieved. The noise reduction combined with the traditional filtering algorithm and the segmentation algorithm can effectively improve the signal-to-noise ratio and improve the measurement precision for metal parts. The flatness measurement realized by combining RANSAC can effectively avoid the interference of the noise signal to the measurement result, and the error of comparison with traditional flatness measurement method is <±5μm. The evaluation of the complex surface measurement results is realized by a variety of registration algorithms. The research results of this thesis demonstrate the great potential and broad application prospects of laser triangulation sensors in industrial applications.