CAJ | 학술논문

为提取马铃薯高光谱图像的外部缺陷特征，结合高光谱成像技术、主成分分析、小波变换以及改进的最大类间方差法对马铃薯外部缺陷进行提取。通过对马铃薯高光谱图像数据做全波段（468~960 nm）主成分分析（PCA）和独立主成分分析（ICA），以全波段缺陷特征明显的主成分图像作为特征提取图像，然后对其分别进行4尺度的sym4、Haar、db4小波变换，从纹理信息客观的评价利用小波重构经PCA和ICA处理的主成分图像，再使用改进的Otsu算法对重构的图像进行目标图像分割，获得马铃薯的缺陷特征。结果表明，结合高光谱成像技术、PCA分析、小波变换和改进的Otsu算法提取马铃薯病斑、机械损伤和孔洞缺陷特征效果明显，正确率达94.2%。
위제취마령서고광보도상적외부결함특정，결합고광보성상기술、주성분분석、소파변환이급개진적최대류간방차법대마령서외부결함진행제취。통과대마령서고광보도상수거주전파단（468~960 nm）주성분분석（PCA）화독립주성분분석（ICA），이전파단결함특정명현적주성분도상작위특정제취도상，연후대기분별진행4척도적sym4、Haar、db4소파변환，종문리신식객관적평개이용소파중구경PCA화ICA처리적주성분도상，재사용개진적Otsu산법대중구적도상진행목표도상분할，획득마령서적결함특정。결과표명，결합고광보성상기술、PCA분석、소파변환화개진적Otsu산법제취마령서병반、궤계손상화공동결함특정효과명현，정학솔체94.2%。
To get feature extraction of potato hyperspectral image external defect, a method is proposed based on hyperspectral imaging technology, principal component analysis, wavelet transform and the advanced Otsu. The fundamental principle is that using principal component analysis (PCA) and independent principal component analysis (ICA) for al band (468-960 nm) of potato hyperspectral image to gain principal components image, and then using of 4 scale sym4, Haar, db4 wavelet transform to reconstruct higher resolution image and assessing the effects of the method with wavelet transform to process principal components image, final y, a advanced segmentation algorithm of Otsu was applied to obtain potato external defect. Experimental results showed that the accuracy was 94.2% for extracting characteristics of potato sickness spot, the mechanical damage and holes based on hyperspectral imaging technology, principal component analysis, wavelet transform and the advanced Otsu.