Research, With Multiple Images Of The Scene Image Segmentation Algorithm

In recent years, the technology of image segmentation has been a hot spotin the field of the imaging technology and computer vision. Especially in thefield of medicine, the medical image has different details because of thedifferences between imaging mechanism and application environment ofmedical instrument. So the image segmentation technology is essential. Besidesthat, it is very important for computer-aided treatment and clinical treatment totake and analyze Multiple Copy Images. The use of complementary andredundant information of multi-source medical images, the images after imagesegmentation can provide users with more useful information and knowledge.The segmentatated results of the image segmentation algorithoms have a directimpact on the accuracy of the analysis results.Different methods of image segmentation have been studied and comparedin this paper; Watershed Algorithom is selected to deal with Multiple CopyImages segmentation. Watershed method advantadges of intuitive principle,high efficient algorithm, detecting the edge line continuously and closedly andcan respond weak edge. At the same time, the watershed algorithm is susceptibleto noise and fine-grained factors, and produces over-segmentation easily. For theover-segmentation of watershed, considered the forming reasons, the improvedICA algorithm is used on the image preprocessing for removing the impact ofthe noise which causes the over-segmentation of Watershed Algorithom, andthen it overcomes the over-segmentationg effectively. Improved WatershedAlgorithom is applied to segment the multiple copy images which are acquiredin real-time.Based on the previous studies, this method is used on CT hand multiplecopy images and MRI brain multiple copy images segmentation, and tried to usethis method to segment the sun images. The experimental results show thatwherther the medical images segmentation or sun images segmentation caninhibite the over-segmentation obviously, meanwhile, the accuracy of thesegmentation results are ensured.