Application Research Of Machine Vision In The Field Of Engineering Inspection

With the upgrading and strengthening of computer hardware and software and the rapid development of artificial intelligence in recent years,machine vision is widely used in various fields,driving the development of digitalization and intelligence in all walks of life.In particular,the continuous upgrading of current smartphones has brought more and more powerful image acquisition technology.This makes portability and application of machine vision technology for the masses possible.At the same time,the number of infrastructures such as buildings and transportation facilities has increased year by year in the context of economic development.These construction facilities are subject to internal and external factors that cause their structural damages.These damages need to be discovered and dealt with in a timely manner,otherwise it will cause hidden safety hazards,and affect the service status and service life of the building.In severe cases,it will cause a loss of people's lives and property safety.Therefore,these construction facilities need to be regularly detected and maintained.Existing detection methods are inseparable from the manual operation of the detection personnel,but due to the subjectivity of manual operation and other factors have certain limitations.Combined with the research direction of machine vision,structural inspection overcomes the limitations of manual operation,and is also the key to its safety,digitization,and intelligence.In order to optimize the inspection work and improve the accuracy of the inspection results,this paper proposes to apply machine vision and other related technologies in combination with specific construction facility inspection scenarios.The following text describes the exploration and research of the main two directions of this article:(1)Based on the deep learning theory of images in machine vision,the training model of the pictures taken by the smartphone is used to realize the rapid detection and classification of the maintenance tools.This paper establishes a tool feature recognition and classification model for the tool inventory in the track maintenance work.In the study,ten common tools were selected for model training and verification.The results show that the model can accurately and quickly determine the type and number of tools,which can be applied before and after the maintenance.This method can quickly complete the tool inventory work,ensure the smooth maintenance of the track,and reduce the hidden safety hazards of the track caused by missing tools.(2)Based on the three-dimensional reconstruction theory in machine vision,a damage detection method for structural damage model reconstruction is proposed.To achieve accurate and convenient detection of the surface damage of the building structure,this paper uses two different equipment and methods of depth camera and smartphone to reconstruct the surface damage of the concrete specimen in three dimensions,obtain point cloud data,and then use the random sampling consensus algorithm to simulate Close the plane and remove it to get the data of the damaged part,and extract the geometric features of the damage and calculate the damage volume.The accuracy of the damage volume obtained by the two methods is compared.By analyzing the data of the two methods,the accuracy and post-processing difficulty,equipment cost,data acquisition efficiency,and applicability of the two methods are compared and analyzed during the experiment.The method of data obtained through multi-view stereo reconstruction is more accurate,and the cost is low,so it is easy to carry and carry out inspection work.