论文部分内容阅读
针对籽棉表层多类难检异性纤维,包括纸屑、白发、丙纶丝、化纤和地膜等5种白色物质,采用高光谱技术和最小噪声分离(minimum noise fraction,MNF)分析方法对含有异性纤维的籽棉图像进行研究。该文在400～1 000 nm的光谱范围内采集高光谱图像,根据光谱曲线选取子区域,应用最小噪声分离分析方法降维、去噪。取MNF变换后的前4幅分量图像,通过视觉评价,选定最佳成分图像并融合中值滤波、灰度变化等图像处理的方法确定最佳分割图像,提取异性纤维。试验结果表明,对于以上5种异性纤维,该方法的识别率达到91.0%。该研究可为棉花异性纤维检测系统的开发提供参考。
Aiming at the different kinds of difficult-to-detect foreign fibers on the surface of seed cotton, including five kinds of white matter such as paper dust, white hair, polypropylene filament, chemical fiber and plastic film, the hyperspectral and minimum noise fraction (MNF) The seed cotton image was studied. In this paper, the hyperspectral images are collected in the spectral range of 400-1000 nm, the sub-regions are selected according to the spectral curve, and the minimum noise separation and analysis method is used to reduce the dimension and denoise the spectrum. Taking the first 4 components of MNF transformed images, the best segmentation image was selected by visual evaluation, the best composition image was selected, and the median filtering and gray-scale changes were fused. Then the heterosexual fibers were extracted. The experimental results show that the recognition rate of this method reaches 91.0% for the above five kinds of heterosexual fibers. This study can provide a reference for the development of cotton heterosexual fiber detection system.