摘要
智能成像方法的研究有利于装配式建筑钢结构布局的更加合理,及时检测和排除装配式建筑钢结构中的构架缺陷,保证其应用过程中的安全性能。针对采用当前方法得到的装配式建筑钢结构布局成像效果图,存在成像视觉效果不佳、丢失了较多细节特征,无法有效抑制噪声的问题,提出一种基于Markov模型约束最大似然法的智能成像方法,对装配式建筑钢结构布局原图像块和被噪声污染的图像块进行估计值计算。采用局部自适应回归内核和主成分分析法对图像块结构进行特征提取和聚类分析,得到原图像块的均值和协方差矩阵估计值,对获得的均值和协方差矩阵进行加权平均处理,即可实现每个图像块的去噪估计。基于各个图像块的去噪估计结果,采用最大似然法对其进行理想共焦非相干成像,同时引入Markov随机场模型对其进行约束,能够恢复图像高频成分,获得最终成像效果图。仿真证明,所提方法能够极大地抑制原始目标图像中引入的噪声,同时保留边缘锐度以及细节特征,获得良好的视觉成像效果。
An intelligent imaging method based on Markov model restricted maximum likelihood method was presented. At first,we calculated the estimated value of the original image block and the image block with noise pollution in steel-structural layout of prefabricated construction. Then,we used the local adaptive regression kernel and principal component analysis for the feature extraction and cluster analysis of image block structure. Thus,we could obtain estimated value of the mean and covariance matrix of the original image block. Moreover,we performed the weighted mean processing of the mean matrix and covariance matrix,so as to achieve denoising estimation of each image block.Based on the results of denoising estimation,we used the maximum likelihood method to perform the ideal confocal incoherent imaging. Finally,we introduced Markov random field model to restrict it. Thus,the high-frequency components of image could be restored and the final imaging effect could be obtained. Simulation proves that the proposed method can greatly suppress the noise introduced from the original target image. Meanwhile,it can preserve the edge sharpness and detail feature and obtain good visual imaging effect.
作者
殷乾
YIN Qian(Wesleyan College,Guangzhou University,Gaungzhou,Guangdong 510000,China)
出处
《计算机仿真》
北大核心
2018年第11期310-313,共4页
Computer Simulation
关键词
装配式
建筑
钢结构
布局
智能成像
Prefabricated
Building
Steel structure
Layout
Intelligent imaging
