基于机器视觉的建筑工程施工现场违规作业智能监测方法

Intelligent monitoring method of illegal operation on construction site of construction project based on machine vision

当前施工现场违规作业监测过程中,主要通过主成分分析法获取现场图像中的目标对象,容易受到复杂背景的干扰,使得最终监测结果ACC(准确率)较低。因此,提出应用机器视觉技术的建筑工程施工现场违规作业智能监测方法。通过灰度化和直方图均衡处理,实现建筑工程施工现场图像的增强处理。从机器视觉领域的目标检测原理入手,结合YOLOv3(端到端的目标检测)网络,构建施工现场图像目标检测模型,面对复杂施工背景快速提取目标对象。设计以TensorFlow(张量流)框架为核心的违规作业智能监测网络,基于此生成智能监测结果。实验结果表明:所提方法得到的智能监测结果ACC值为92.17%,满足施工现场违规作业监测要求。

In the current monitoring process of illegal operation on construction site,the target object in the site image is mainly obtained by principal component analysis,which is easily interfered by complex background,which makes the final monitoring result ACC(accuracy)low.Therefore,an intelligent monitoring method for illegal operations in construction projects using machine vision technology is proposed.Through grayscale and histogram equalization processing,the enhanced processing of construction site images of construction projects is realized.Starting from the principle of object detection in the field of machine vision,combined with YOLOv3(end-to-end object detection)network,a construction site image object detection model is constructed,and target objects are quickly extracted in the face of complex construction background.An intelligent monitoring network for illegal operations is designed with the TensorFlow framework as the core,and intelligent monitoring results are generated based on this.The experimental results show that the ACC value of the intelligent monitoring result obtained by the proposed method is 92.17%,which meets the monitoring requirements of illegal operation on the construction site.