Segmentation Of Salient Objects Based On Region Growing

With the rapid development of information technology,we can obtain a large number of images and videos through the network.On the basis of massive data,efficient and accurate computer vision processing becomes more difficult.In the complex environment,how to segment the interested objects from the image is a hot topic in the field of computer vision in recent years.Human vision system can locate salient objects in an unknown environment quickly and accurately and directly ignore the influence of complex background.In computer vision,the segmentation for salient objects has been used widely.In this paper,natural and medical images are our study targets.Based on region growing technology,an edge detection algorithm for salient objects is constructed.With the salient region detector,salient objects in target images are segmented to solve the problem of segmentation of salient objects from complex background regions in the real environment.The specific contents of this thesis are as follows:1.We propose a new edge detection algorithm of salient objects from images based on region growing.This algorithm could grow the salient objects' edges accurately,delete the interference edges from the background without additional interactive information.This algorithm still has a good effect on edge detection from the images with complex background.And this paper tests and evaluates our proposed algorithm compared with the traditional Canny edge detection algorithm and the improved edge detection algorithm based on Canny operator.Experiments verify the effectiveness of the proposed algorithm for salient objects' edge detection based on region growing.In addition,this edge detection based on region growing could also be applied to other edge detection,image segmentation and object recognition algorithms.2.After obtaining the edges of salient objects by proposed edge detection algorithm,in order to segment salient objects from the background,this paper uses the Itti visual saliency detector to get the salient region image.And this paper proposes a segmentation algorithm for salient objects by combining the detected edge image with the salient region image.Experiments show that the proposed segmentation algorithm could extract salient objects well.3.A segmentation algorithm of boney tissues is proposed based on region growing from medical images.Compared with the results of manual segmentation,the effectiveness and efficiency of the proposed segmentation algorithm are verified.The purpose of boney tissues' segmentation from CBCT images is achieved.