Weld Seam Detection Algorithm Implementation Based On Android Platform

In the context of the rapid development of modern intelligent manufacturing,industrial vision inspection technology is widely used in automated welding production.The inspection of the workpiece,as well as the monitoring of the work conditions by the operator,has become more demanding than before.As the revolution of Industry 4.0proceeds,automatic weld seam tracking is still a critical problem that needs to be solved.And in recent years,due to the increasing computing power of mobile chips and the excellent human-computer interaction performance of Android system,it is also increasingly used in industrial production control,and the implementation of real-time weld seam detection algorithm on mobile devices has become possible.The title of this article comes from Wuhan Iron and Steel Group Co.,Ltd.welding automation transformation needs,the use of machine vision solutions instead of manual inspection,cost reduction and efficiency,used to improve the efficiency of welding production.This paper carries out the study of weld inspection algorithms.The main work and innovation points of this paper can be divided into four aspects.First,the current status of domestic and international research on weld seam tracking system is introduced.At the same time,the process procedure process of spiral pipe welding is also introduced.The practical use scenarios of the weld seam detection algorithm are described.And based on the scene positioning research of industrial environment,the use scenario and function positioning of the weld seam detection algorithm is discussed.Second,the selected camera was calibrated and corrected to collect image data.And the image processing methods were explored for different situations,after pre-processing,centerline extraction,feature point detection,and considering the shortcomings of the previous single algorithm such as low accuracy and robustness,the algorithm was improved.The centerline extraction algorithm is optimized to sub-pixel accuracy.Thirdly,the research on the application of Android system in industrial software is tracked to complete the implementation of weld detection algorithm in mobile by graphic image processing technology and embedded mobile application technology.The design and development of the weld inspection APP was completed by wrapping the algorithm with NDK and implementing it on the Android platform.Fourth,by building a weld detection platform and conducting welding inspection experiments,performance tuning is performed with reference to the experimental results.In the experiments,the hardware part focuses on the selection of light sources,cameras and other important sensors,the acquisition of image data under different environments,and the investigation of the impact on image data under different lighting conditions.The algorithm implementation part focuses on the image processing of the weld seam,which mainly includes pre-processing,centerline extraction and feature point detection.Then,the adaptability of different algorithms under different lighting conditions,such as structured and unstructured light,is verified.For the developed application,the time-consuming threads are explored using the Profile tool,which is used to provide directions for performance optimization.Finally,the functional decomposition of the detection system is used to clarify the functional orientation of the different modules and to ensure the reliability of the software development.In the process of algorithm implementation,the study used the Open CV graphics processing library.The Kotlin language was used for the development of the Android application.To ensure the algorithm and interface performance of the terminal software,the C++ algorithm module was encapsulated,and the application was interacted with using the NDK,and the embedded terminal system was mined for the best possible performance without replacing the hardware.After testing,it was able to meet the requirements of performance and stability.