The Image Segmentattion Method Research Based On Visual Significance And C-V Level Set Model

Image segmentation is an important research content of the image engineeringand its related fields，it has broad application prospects,and has been attracting theattention of a large number of scholars.This paper focuses on the automated imagesegmentation,combined with the visual salience and C-V level set,The main workcompleted is as follows:Firstly,analyses and compares the existing classic visual perception model,meanwhile study detailedly Itti visual perception model algorithm principle andcontent, combined with the objectives of this article on the exsisted shortcomings, themain improvement work are as follows: in the primary visual feature extraction partintroducing the image texture feature consistency measure,to enhance the significantdegree of the saliency regions; Adding the image pixel features discrimination,according to the image pixels’ size to establish corresponding gaussian pyramid, anddo related central surrounding differential operation; Improved saliency mapcombined strategy,to decrease the pseudo-focus attention in image,in order to reducethe subsequent workload;in addition,as the paper uses the C-V level set model forsingle level set,just to obtain the most significant region contours as the initialevolution curve.Secondly,in order to reduce the complexity of the C-V level set modelalgorithm,to avoid constantly re-initialization of the initial curve,adding constraintsitems in C-V level set;using laplace operator as the fixed operator to improve theaccuracy of image segmentation.Finally,using C-V level set model to do curve evolution,whiche initial contourcurve getted by Itti pre-segmentation model,the related experimental validation workresults prove that the method used in this paper is not only suitable for the use ofsingle objective object, but also apply to multiple target object segmentation, incomparison to the traditional methods, this image segmentation method from eitherthe image segmentation accuracy or efficiency achieved better effect.