【摘要】 混凝土构件外观质量缺陷检测是施工质量验收的一项重要内容。基于机器视觉的数字化检测方法可以替代传统的观察法，得到兼具客观性和准确性的外观质量缺陷检测结果。基于文献调研，将现有的基于机器视觉的检测方法分为定性检测方法和定量检测方法。详细阐述了不同定性检测方法实现气泡及裂缝等缺陷的特征识别与提取的一般技术流程和优缺点，以及不同定量检测方法测量缺陷几何参数指标的基本原理和适用检测范围。针对现有机器视觉检测技术存在的混凝土外观质量缺陷检测种类不充分且工程现场适用性较差的问题，提出研发集成高清相机和结构光相机的便携式检测设备构思，该检测设备基于深度学习算法和三维点云数据可实现定光照、定角度、定距离的混凝土构件外观质量采样以及基于数字孪生的检测。更多还原

【Abstract】 The appearance quality defects inspection of concrete components is an important aspect of construction quality acceptance. The digital detection method based on machine vision can replace traditional observation methods and obtain appearance quality defect detection results that are both objective and accurate. Based on literature research, existing detection methods based on machine vision can be categorized into qualitative detection methods and quantitative detection methods. The general technical process and advantages and disadvantages of different qualitative detection methods for feature recognition and extraction of defects such as bubbles and cracks were elaborated in detail, as well as the basic principles and applicable detection ranges of different quantitative detection methods for measuring geometric parameter indicators of defects. Aiming at the problems of insufficient detection types of concrete appearance quality defects and poor applicability of engineering site in the existing machine vision detection technology, the idea of developing portable detection device integrating high-definition camera and structured light camera was proposed. The detection device based on the deep learning algorithm and 3D point cloud data, the detection equipment can realize the sampling of the appearance quality of concrete components with fixed illumination, fixed angle and fixed distance, and the detection based on digital twin.更多还原