摘要
以鸡胴体为研究对象,探讨基于高光谱图像技术的鸡胴体内部粪便污染物检测方法。首先采集400～1 000 nm的鸡胴体高光谱图像;然后应用主成分分析(PCA)获得主成分图像,由第1主成分图像得到3个特征波长518.59,562.64,700.67 nm,并以700.67 nm特征波长下的图像作为鸡胴体内部粪便污染物检测特征图像;最后构建掩膜以消除特征图像背景噪声并将其置为白色,并运用阈值分割和数学形态学完成粪便污染物的提取。试验结果表明,对100个鸡胴体样本进行检测,检测总正确率为93%。应用高光谱图像技术结合主成分分析等数据处理方法能较好地完成对鸡胴体内部粪便污染物检测,为鸡胴体内部粪便污染物在线快速检测提供重要的理论依据。
Using chicken carcasses as the research subject,the detection method for internal fecal contaminants of chicken carcasses based on hyperspectral imaging technology was investigated.Firstly,hyperspectral images of chicken carcasses from 400 nm to 1 000 nm were acquired.Secondly,the principal component images were obtained by the principal component analysis（PCA）,three characteristic wavelengths（518.59,562.64 and 700.67 nm） were found according to the first principal component（PC1） image,and 700.67 nm wavelength images were selected as the detection images of the internal fecal contaminants of chicken carcasses.Lastly,the masks were constructed to eliminate the the background noise of characteristic images,and the fecal contaminants were extracted by the threshold segmentation and morphological imaging processing.The experimental results showed that the total accurate rate of the detection was 93% using 100 samples of chicken carcasses.The principal component analysis and hyperspectral imaging technology can meet the requirement for detecting internal fecal contaminants of chicken carcasses＇which provides a theoretical basis for the on-line rapid detection of internal fecal contaminants of chicken carcasses.
出处
《江西农业大学学报》
CAS
CSCD
北大核心
2011年第3期573-577,共5页
Acta Agriculturae Universitatis Jiangxiensis
基金
国家高技术研究与发展技术(863计划)资助项目(2008AA10Z209)
江西农业大学青年科学基金资助项目(2954)
关键词
高光谱图像
粪便污染物
鸡胴体
检测
hyperspectral imaging
fecal contaminants
chicken carcasses
detection
