摘要
图像分割是模式识别、计算机视觉等领域的重要研究内容,也是图像信息处理的难点和热点之一.以孪生支持向量机(TSVM)与极坐标复指数变换(PCET)理论为基础,提出了一种基于局部像素特征分类的图像分割算法.该算法首先对局部像素窗口进行PCET,并将PCET系数幅值作为图像的像素级特征;然后利用指数交叉熵阈值技术确定训练样本,并进一步训练出TSVM分类模型;最后利用已获得的TSVM分类模型对原图像像素进行分类处理,从而获得图像的最终分割结果.实验结果表明,该算法可以获得较好的图像分割结果.
Image segmentation is a classic inverse problem w hich consists of achieving a compact re‐gion‐based description of the image scene by decomposing it into meaningful or spatially coherent re‐gions sharing similar attributes .Image segmentation is of great importance in the field of image pro‐cessing .In this paper ,we present an effective image segmentation approach based on local pixel fea‐tures classification .Firstly ,the pixel‐level image feature is extracted via Polar Complex Exponential Transform (PCET) .Then ,the pixel‐level image feature is used as input of twin support vector ma‐chine (TSVM ) model (classifier) ,and the TSVM model (classifier) is trained by selecting the train‐ing samples with the exponential cross entropy thresholding .Finally ,the image is segmented with the trained TSVM model (classifier) .This image segmentation not only can fully take advantage of the local information of image ,but also the ability of TSVM classifier .Experimental evidence shows that the proposed method has very effective segmentation results in comparison with the state‐of‐the‐art segmentation methods proposed in the literature .
出处
《辽宁师范大学学报(自然科学版)》
CAS
2014年第4期479-485,共7页
Journal of Liaoning Normal University:Natural Science Edition
基金
国家自然科学基金项目(61472171
61272416)
辽宁省教育厅科学技术研究项目(L20134070503)
关键词
图像分割
极坐标复指数变换
孪生支持向量机
指数交叉熵
image segmentation
polar complex exponential transform
twin support vector machine
exponential cross entropy