基于多光谱图像融合技术的麦粒图像增强方法研究  被引量:1

Research of Grain Image Enhancement Method Based on Multispectral Image Fusion Technique

在线阅读下载全文

作  者:张红涛[1] 孙志勇[1] 田媛[1] 侯栋宸 

机构地区:[1]华北水利水电大学,河南郑州450045

出  处:《华北水利水电大学学报(自然科学版)》2014年第6期89-92,共4页Journal of North China University of Water Resources and Electric Power:Natural Science Edition

基  金:国家自然科学基金资助项目(31101085);河南省基础与前沿技术研究计划(122300410145);河南省高等学校青年骨干教师资助计划(2011GGJS-094);华北水利水电大学2014年大学生创新计划项目

摘  要:部分麦粒在单一波段的光谱图像中,常规的形态学及纹理特征难以完全体现出来,这样对于后续的特征提取会造成信息的丢失,降低识别率,为此提出了一种基于多光谱图像融合技术的图像增强方法.首先,采集同一组麦粒的近红外、红色、绿色和全色图像,采用IR+R,IR-R,IR+G+R,IR/R,IR/G,G/R,r+ir,g+ir,g+r,ir-r,2g-r-ir,(IR-G)/(IR+G)等融合方式;然后将全色图像转换到HSI彩色空间,同时采用IR/I,R/H,G/S等7种融合方式对其处理;最后使用主成分分析,进行数据的分析.在试验中,进行了多种融合方式的对比研究,并与原始波段的光谱图像比较.结果表明,IR-R的融合方式效果最好.Conventional morphology and texture of some grains is difficult to be completely reflected in spectral images of the single band, which would cause losses of information for the following feature extraction and reduce the recognition rate. Thus, an image enhancement method based on muhispectral image fusion technique was proposed in this paper. Firstly, the near infrared, red, green, and full-color images of same group of grains were acquired respectively, and these fusion methods were used, such as IR + R, IR-R, IR + G + R, 1R/R, IR/G, G/R, r + ir, g + it, g + r, ir-r, 2g-r-it and (IR-G)/( IR + G). Then the full-color image was converted into HSI space, meanwhile 7 fusion methods including IR/I,R/H,G/S and so on, were used to deal with the image. Finally, the principal component analysis was used to analyze the data. In the experiment, a comparative study of a variety of fusion methods was carried out, and the spectral images of the original band were compared. The result shows that IR-R fusion images get the best performance.

关 键 词:多光谱图像 图像融合 图像增强 HSI空间 

分 类 号:TP751[自动化与计算机技术—检测技术与自动化装置] S123[自动化与计算机技术—控制科学与工程]

 

参考文献:

正在载入数据...

 

二级参考文献:

正在载入数据...

 

耦合文献:

正在载入数据...

 

引证文献:

正在载入数据...

 

二级引证文献:

正在载入数据...

 

同被引文献:

正在载入数据...

 

相关期刊文献:

正在载入数据...

相关的主题
相关的作者对象
相关的机构对象