Research On Image Splice F Usion Based On Internal Inspeetion Of Pipeline

Industrial pipelines have been widely used in industrial and mining enterprises,among which industrial thermal pipelines,due to their special working environment,are prone to pipe blockage and burst due to the falling off and accumulation of coating materials inside the pipeline,thus causing equipment damage and more serious accidents that will cause casualties.At present,most non-destructive testing techniques are limited by the material,wall thickness,volume and cost of the equipment,and can not be applied to the non-destructive testing of this kind of pipeline.In order to solve the problem of pipeline detection effectively,this paper studies the nondestructive testing method of pipeline video image based on industrial endoscope.The video image of the inner wall of the pipe collected by industrial endoscope can not show the necessary flaw detection information intuitively and comprehensively.Therefore,it is necessary to expand and splice the inspection images obtained by video continuous frame capture,and finally realize the two-dimensional plane reconstruction and the statistics of the inspection area.This method not only improves the detection accuracy,greatly reduces the detection cost,but also is not limited by the pipeline's own attributes.This paper first introduces the process of image expansion,dimension reduction and image preprocessing.A fast SIFT algorithm based on improved feature descriptors is proposed to solve the problem of high dimension and high computational complexity of feature descriptors.Compared with the original algorithm,the overall running speed of the algorithm is doublled under the premise of ensuring the same scele of the original algorithm,which effectively improves the timeliness of the algorithm.Secondly,after detecting the feature points of the image,the matching feature points are elmlinated by us ing the text-based similarity preprocessing idea.It can be seen from the experimental results that it can effectively simply by Euclidean distance are matched,and the improved adaptive threshold RANSAC algorithm is applied to eliminate them.In addition to using the distance mean between the matching point and the relative strain exchange matrix,the method also uses the variance of the distance to obtain the best threshold.Finally,the simulation results show that the proposed algorithm is superior to the relevant algorithms in the evaluation parameters such as matching rate and matching total.Finally,based on the improved image splicing algorithm,the collected internal flaw detection image sequence is spliced into a complete flaw detection image to realize the reconstruction of the inner wall of the pipeline.And combined with image segmentation algorithm,statistics of its flaw detection area.From the experimental data,it can be seen that the statistical accuracy of the flaw detection area of this algorithm is obviously improved compared with the related splicing algorithm.