Research On Image Segmentation Algorithm Based On Saliency Detection Model

Image segmentation is a process that divides images into specific,unique regions and extracts interest objects.As the bottom part of the image processing,the segmentation results are used in many fields,such as image understanding,semantic recognition,image search and so on.In modern society,as an important way to obtain information,images are widely accepted by the public for their easy to understand and easy access.The segmentation of the image can make the computer get the object faster and facilitate the further analysis and understanding of the image.In this paper,based on the significance of the image,the image segmentation model based on the saliency detection is mainly studied in combination with the pulse coupled neural network and the level set model.Aiming at the problem of sensitive seeds in the region growing model,we combine the saliency detection model to determine the initial seed points,and realize the region based image segmentation using the coupling characteristics of pulse coupled neural network.In addition,aiming at the problem that the initial contour is sensitive to the level set model,we combine the saliency detection model to determine the initial contour of the level set,which accelerates the evolution speed of the level set contour and improves the efficiency of image segmentation.Specific contents are as follows:Aiming at the problem that complicated images are interfered by background,a saliency pulse coupled neural network method of image segmentation based on region growing was proposed.Firstly,using the saliency filtering algorithm and the method of maximum between-class variance,the saliency map and the object image are obtained.With this method,the interference which comes from the background for the initial seed point selection is eliminated.Secondly,according to saliency values in saliency map,the most saliency region is captured and the initial seed points are produced.Finally,the operations of object image segmentation are achieved via the improved RG-PCNN model.Moreover,in order to improve segmentation effect of images with multi-object and images with intensity inhomogeneity,an image segmentation method based on region growing with local coupled neural networks(RG-SLPCNN)is proposed.First,the saliency map of the original image is extracted by using saliency detection algorithm.Second,the object and the background of the saliency map are coarsely segmented by histogram thresholding method.And centroid of the object is calculated as the initial seed point of RG-SLPCNN.In addition,convolution results of Gauss kernel and original image are used as amplification coefficients,so that local characteristic is introduced into the dynamic threshold.Finally,the RG-SLPCNN method is implementing by segmenting images with multiobject and images with intensity inhomogeneity.In order to improve the edge segmentation effect of the level set image segmentation and avoid the influence of the initial contour on the level set method,a saliency level set image segmentation model based on local Renyi entropy is proposed.Firstly,the saliency map of the original image is extracted by using saliency detection algorithm.And the contour of the saliency map is used as the initial contour of the level set.Secondly,the local energy and edge energy of the image are obtained by using local Renyi entropy and Canny operator respectively.At the same time,new adaptive weight coefficient and boundary indication function are constructed.Finally,the local binary fitting energy model(LBF)as an external energy term is introduced to enhance the segmentation ability for the intensity inhomogeneous image.Although a series of researches have been done in the field of image segmentation based on saliency,considering the complexity of natural images,the salient objects extracted are often difficult to meet the requirements of preprocessing.And,for the segmentation of edge and intensity inhomogeneous images,further research are needed.