摘要
为快速准确地得到贾卡经编织物花型图案,提出一种基于小波多尺度分层、马尔科夫随机场建模和贝叶斯最优准则分割理论的贾卡经编织物图像多纹理区域分割技术。首先通过小波变换将贾卡织物图像作多尺度分解等预处理,其次结合马尔科夫随机场理论,运用有限高斯混合算法构建图像灰度场分布模型,采用尺度逻辑算法构建标号场先验模型,最后在贝叶斯理论框架下采用连续最大后验概率准则对贾卡织物图像进行多纹理区域分割。实验结果表明,应用该算法对多类贾卡经编织物进行纹理分割,可以快速高效地得到贾卡织物花型图。
To obtain the pattern of jacquard warp-knitted fabric quickly and accurately,a new approach is proposed a to multi-texture region segmentation of jacquard warp-knitted fabric based on the theory of wavelet transform,Markov random field and Bayesian optimal criterion.Firstly a wavelet multi-scale transform is performed on the image of jacquard fabric.Secondly within the theoretical framework of Markov random field,the grey field distribution model and label field prior model were constructed by finite Gaussian mixture algorithm and multi-level logistic algorithm,respectively.Finally by using sequential maximum posterior probability of Bayes,the multi-texture region segmentation of jacquard warp-knitted fabric is obtained.Experiment results demonstrate that this method is suitable for obtaining the pattern of jacquard warp-knitted fabric quickly and accurately through texture segmentation.
出处
《纺织学报》
EI
CAS
CSCD
北大核心
2011年第12期51-55,共5页
Journal of Textile Research
关键词
贾卡经编织物
多区域
纹理分割
小波变换
马尔科夫随机场
jacquard warp-knitted fabric
multi-texture region
texture segmentation
wavelet transform
Markov random field
