Portable Weld Bead Film Detector Image Processing Functional Reconstruction

Pipeline transport is proved to be the most effective technical method which makes the world oil industry development in large scale, however, domestic pipeline failure accident is caused by weld defect in relatively large proportion. Because pipe laying depends entirely on the welding process to complete, the quality of the welding seam has a direct impact on the normal work of the oil and gas pipeline. X-ray film imaging method is the most widely detection technology used in weld detection at present, but it has obvious shortcomings. So we put forward the conception of the X-ray weld film intelligence assessment system, and we develop a portable weld film detection system, which can be efficient, intuitionistic and accurate to read the film information auxiliary, and convenient for application of the outdoor pipeline lying in long distance, in order to achieve the timely evaluation of soldering on the weld line.Based on the current research status, this paper is on the foundation of the above survey data analysis, the system adopted the software-type image processing system, the film industry digital input is the focus of hardware design; And compiling software can develop flexible and applicable image processing algorithm and pattern recognition.This paper combines with X-ray film weld character, employs the industry camera in line scanning to collect undamaged and undistorted digital image, proposes the high-efficiency image processing method and order in allusion to features that digital weld image exists and contains a lot of noise, low contrast ratio, edge slur, and then carries through edge detection after image processing and calculates the characteristic parameters of weld defect.The issue put forward to improve identifiability of the weld defects by embossment processing, the realization about the weld X film defects intelligent identification comes up with identification through the improved three layer feedforward BP neural network to the acquired characteristic parameters. And the number of network hidden nodes, momentum coefficient, error level and step length and other network parameters adopt experiments to gain the best value. VC++, MATLAB and Microsoft Office Access are employed to realize system function mixed programming. The establishment of the system and database can lay the foundation to the research of weld parameters and weld image processing, and conduct database management to the weld bead characteristics parameters.