中文摘要：

针对喷墨印花织物在噪声环境下缺陷检测精度低的问题,提出一种基于混态马尔可夫随机场（MRF）模型的喷墨印花运动纹理检测算法.该算法利用运动纹理的时-空域特征表示,引入运动纹理的混态MRF模型,构建同时包含运动状态和背景状态的运动纹理特征图.为了有效提高模型对复杂纹理背景的表征能力,建立基于混态MRF模型的运动纹理检测模型,并将运动纹理检测过程转化为特征能量最小化问题.采用改进ICM优化求解算法,实现运动纹理检测和动态背景重构,有效提高运动纹理检测精度.实验结果表明：该算法能够有效检测出喷墨印花织物缺陷纹理,并且具有较强的抗噪声干扰能力.

英文摘要：

A novel motion texture detection algorithm based on the mixed-state Markov random field（MRF）model was proposed to deal with the problem of low accuracy in defect detection of ink-jet printing fabric under noisy environment.The representation of spatio-temporal features was applied for motion texture.Meanwhile,a mixed-state MRF model was introduced to constructing a feature map of motion texture,where motion and background states could be jointly modeled.Furthermore,a mixed-state MRF detection model for motion texture was presented to enhance the capability representation of dynamic background texture changes.The process of motion texture detection was formulated into the feature energy minimization problem.A novel ICM optimization algorithm was employed to deal with the problem of simultaneous motion texture detection and dynamic background reconstruction to improve the detection accuracy of motion texture.The experimental results show that the proposed algorithm can effectively detect defect texture from ink-jet printing fabric and has strong anti-jamming ability against noise.