Research On Image Processing Method Of Manganese Nodule

Manganese nodules mainly distributed on the seabed of two thousand to six kilometers,contain dozens of precious metal elements.We need to analyze the images obtained by autonomous underwater vehicle(AUV)to determine the reserves and distribution of manganese nodules in an area.This often requires a series of processing procedures for the images taken by AUV to understand the reserves and distribution of manganese nodules in this area.In this process,underwater images play an important role.In this paper,the process of image processing and image segmentation of manganese nodule are studied,and the method is applied to sonar image processing.First,in order to solve the practical problem,this paper takes manganese nodules image as the research object and studies the processing method of manganese nodules image.We design the process and verify the feasibility and stability of it.Firstly,there is a blurring problem caused by camera shake before image mosaic.At the same time,there is a problem of viewpoint difference due to different angles of image capturing.The above two problems can be solved by Wiener filtering and SIFT(Scale Invariant Feature Transform)algorithm,respectively.Secondly,after image mosaic,the two images have a boundary line,which can be processed by a weighting algorithm.Finally,although the weighting algorithm eliminates the boundary between the two images,it does not meet our recognition requirements for manganese nodules in the image.Therefore,in the process of this paper,a method is proposed to make most of the manganese nodules clearly displayed in the image.Second,aiming at two problems in the process of target and background segmentation in the image of manganese nodules,the uneven illumination and the morphological defects of manganese nodules caused by white sand cover,this paper proposes an improved method.After the image is processed by this method,the above two problems are solved and segmentable manganese nodule images are obtained.Finally,the processed image is segmented to verify the feasibility and stability of the method.Thirdly,aiming at the segmentation problem of sonar images blurred by noise reduction,this paper proposes a fast segmentation method suitable for blurred images according to the characteristics of gray value of sonar images.In binarization,set the threshold less than 255 to zero.Most of the artifacts will disappear,and the remaining small noise can be removed from the image by morphological processing.Therefore,this method does not need to calculate the threshold value of segmentation,which saves a lot of time for sonar image segmentation.