Rock Slice Image Stitching System Based On Hadoop/Spark

In recent years,with the development of image technology,there are more and more images with high resolution and a large number of images.Therefore,higher and higher requirements are put forward for the rapidity of image stitching technology.This article is devoted to the study of rock sheet image stitching based on the big data platform of Hadoop/Spark.The main research contents of the thesis are as follows:(1)For the problem that the SURF algorithm extracts too many feature points and causes the calculation amount to be too large,the SURF algorithm is improved based on Sobel edge detection operator.Firstly,the Sobel operator is used to detect the image edge,and then the SURF algorithm is used to detect the feature.Point,select the SURF feature points on the edge of the image for registration,thereby reducing the number of feature points and achieving fast registration of images;(2)The image stitching program based on the improved SURF algorithm is implemented.The program performs the initial matching of the feature points based on the nearest distance and the nearest distance.The improved RANSAC algorithm is used to complete the f-eature point re-matching to obtain the image transformation matrix.Finally,based on the improved A progressive fade-in weighted fusion algorithm performs image fusion to complete image stitching;(3)According to the working characteristics of each process of image stitching technology,Hadoop/Spark technology is applied to the mosaic of rock slices,and the rock slice image stitching system based on Hadoop/Spark platform is designed and implemented.The system is based on the rock based on Hadoop/Spark platform.The slice image registration module and the single-chip rock slice image fusion module are composed of two modules;(4)Constructing Hadoop/Spark platform,collecting rock slice image data,and verifying the effect of rock slice image stitching system based on Hadoop/Spark platform.Finally,the research work carried out in this paper is summarized,several existing problems are raised,and the future work of this research is further prospected.