基于改进OSV分解模型的图像分割  

Image segmentation based on improved OSV decomposition model

作　　者：郭雨莹 许建楼[1] 尚婉清 尤少培 王海军[1] GUO Yuying;XU Jianlou;SHANG Wanqing;YOU Shaopei;WANG Haijun(School of Mathematics and Statistics,Henan University of Science and Technology,Luoyang 471023,Henan,China)

机构地区：[1]河南科技大学数学与统计学院,河南洛阳471023

出　　处：《西安工程大学学报》2022年第5期53-60,78,共9页Journal of Xi’an Polytechnic University

基　　金：国家自然科学基金(U1504603);河南省重点高校基金(21A510003,22A120006);河南省科技攻关项目(222102210053)。

摘　　要：针对自然图像结构和纹理成分并存现象,提出一种新的基于改进OSV图像分解的多区域图像分割模型。该模型保留OSV图像分解中被忽略的刻画纹理图像的零散度向量场,结合多区域图像分割模型,使得分解后的纹理图像和结构图像更加准确,进而提高图像分割的准确性。为求解本文模型,设计了交替方向乘子法,提出的方法可以同时完成图像分解和图像分割任务。实验结果表明:与相关经典模型相比,改进的模型能明显提高图像分割的主观视觉效果,且精确率、召回率以及Jaccard相似系数平均至少分别提高了3.42%、0.77%、4.89%。A new multi-region image segmentation model based on improved OSV image decomposition was proposed for the coexistence of image structure and texture components.The model preserved the divergence-free vector field of texture image,combined multi-region image segmentation model,which made the decomposed texture and structure image more accurate and improved the accuracy of segmentation.In order to solve the proposed model in this paper,the alternating direction multiplier method was designed.The proposed method could accomplish both image decomposition and image segmentation.The experimental results show that compared with the related classical models,the proposed model can significantly improve the subjective visual effect of image segmentation,and the precision,recall and Jaccard similarity coefficient are at least 3.42%,0.77%and 4.89%higher respectively.

关 键 词：OSV分解模型 零散度向量 图像分割 交替方向乘子法 图像分解 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]