加工番茄虫眼及霉变的可见近红外高光谱成像检测

为了探求一种快速有效识别虫眼和霉变加工番茄的无损检测方法,利用高光谱成像技术,从光谱和图像2个角度对其进行检测。先借助可见近红外高光谱成像系统获取408~1 013nm的加工番茄高光谱图像数据,提取并分析感兴趣区域的平均光谱曲线进行主成分分析,根据各波段权重系数优选了550,750,900nm 3个特征波长;然后通过特征波长下图像的主成分分析,选择缺陷部位与正常区域强度对照最明显的第一主成分图像,通过掩模、阈值处理和形态学开运算等图像处理方法对缺陷番茄进行检测判别。虫眼、霉变和正常三类番茄的识别率分别为93.3%,90%,100%。同时利用上述3个特征波长进行波段比图像运算,并选择波段比550nm/750nm图像进行缺陷识别,虫眼、霉变和正常三类加工番茄的识别率分别为93.3%,96.7%,100%。研究结果表明,二次主成分分析和波段比检测算法均可以有效地识别缺陷加工番茄。另外研究中仅选用了3个特征波段,数据量大大减少,为搭建开发适于加工番茄缺陷的多光谱在线检测系统提供了可能。

Detection of insect hole andmildew in processing tomato by visible near infrared hyperspectral imaging

The quality of tomato products is significantly degraded due to defects on raw processing tomatoes such as insect hole or mildew. This research aims to investigate the potential of using visible/ near infrared (Vis/NIR) hyperspectral imaging for detection of insect hole and mildew on raw processing tomato. Tomato samples were imaged using a hyperspectral imaging system that covers a spectral range from 408 to 1013 nm. To images, region of interests (ROIs) were manually selected to extract mean spectra on every individual samples. Principal component analysis (PCA) was performed on the extracted spectra to select three optimal wavelengths (550, 750, 900 nm) for defects detection. PCA and pair-wise band ratio analysis were conducted on the spectral images using the optimal wavelengths to generate PC and band-ratio images, respectively. Masking, threshold-based segmentation, and morphologic operations were applied on the generated images to identify defective areas on the tomato surface. The accuracies of identifying insect hole, mildew, and healthy tomato achieved 93.3%, 90%, and 100% in the PC images, and 93.3%, 96.7%, and 100% in the band-ratio images, respectively. Therefore, the Vis-NIR hyperspectral imaging could be an effective approach for detecting insect hole and mildew on the surface of raw tomatoes. In addition, online detection system could be benefit by using the wavelengths of 550 nm and 750 nm.