Research On The Image Segmentation Algorithm Based On Neutrosophic

Image segmentation is the first step in image analysis and image understanding,and also is a classic problem of computer vision. The result of image segmentation candirectly determine the quality of the image analysis and the final judgment result ofpattern recognition. Therefore, image segmentation has been the concern of manyscholars, has also been a hotspot of image processing research.In recent decades, many scholars have done a lot of research in the field of imagesegmentation, and has made many research results, but there are still many problems tobe solved and further improved. For example, there is uncertainty in the image. Whenperforming image segmentation, we cann’t use the uncertain information of image,which can lead to inaccurate image segmentation. so we introduce the notion ofneutrosophic theory. The uncertainty information in the image can be clearly quantified,then the neutrosophic theory be used in image segmentation.In this paper, we mainly studies image segmentation which affected by the inherentfuzzy and uncertainty of the image and noise interference in the image. The mainresearch contents and innovation as follows:(1) We proposed a new neutrosophic approach to image segmentation based onwatershed. Firstly, the neutrosophic image is converted by S function, which isdescribed by three subsets T、 I、 F. Then, we use the method of thresholdtransforming neutrosophic image to binary image. Finally, we employ watershedalgorithm to perform segmentation of the image in the binary image. The improvedalgorithm is based on the watershed algorithm, and can get a better segmentationresults. Theoretical analysis and the experimental results show that the algorithm cannot only greatly reduce the traditional watershed algorithm produces over-segmentationbut also eliminate noise of original image.(2) Given the interference from the noise, it is difficult to segment the noisy imageby utilizing traditional image segmentation method. Therefore, we propose a newneutrosophic image segmentation method integrated LPG&PCA (Principal ComponentAnalysis with Local Pixel Grouping). Firstly, the input image is converted intoneutrosophic images by using the neutrosophic set theory. Then, we fitter the noiseimage by appliing proposeda-LPG&PCA filtering operation, and advance thedenoised image by employing the b-enhancing operation, with the uncertainty information from the input image, to smooth the noisy points in the input image. Finally,we segment the denoised and enhanced iage with ag-means clustering method. Theexperimental results indicate that the proposed algorithm can eliminate the noiseeffectively as well as improve the PSNR, therefore, our method can achieve desirableresults in noise immunity and segmentation accuracy and show the effectiveness of theproposed algorithm.