摘要
以GrabCut算法为基础,引入多尺度分析方法,以塔式分解的多尺度图像序列代替固定尺度的原始图像序列估计GMM参数,将粗糙尺度的易分割性与精细尺度的精确性互补,使得算法以较少样本快速确定GMM参数,分割精度不减而效率显著提高。实验表明了算法的有效可行性。
On the basis of GrabCut algorithm, combined with muhiscale analysis method, estimated the GMM parameters with pyramidal decomposited muhiscale serial images instead of fix scale original image. The algorithm estimated GMM parameters rapidly with less samples, obtained equivalent accurate and significantly effective result by utilizing of the complementarity between segmentation accuracy of fine scale and segmentation easiness of coarse scale. The results of experiment show that the proposed method is effective and feasible.
出处
《计算机应用研究》
CSCD
北大核心
2009年第10期3989-3991,共3页
Application Research of Computers
基金
陕西省自然科学基金资助项目(2005A12)
陕西师范大学研究生培养创新基金资助项目(2008CXS025)
关键词
图像分割
多尺度分析
图割
高斯混合模型
image segmentation
muhiscale analysis
graph cuts
Gaussian mixture model（GMM）
