摘要
应用光谱图像技术进行油桃内部品质无损检测技术的研究。通过采集不同波长(分别为632、650、670、780、850和900 nm)的光谱图像,对所采集的光谱图像灰度分布进行洛伦茨分布(LD)函数的拟合。将油桃的糖度与洛伦茨分布函数拟合所得参数分别进行多元线性回归,建立最佳单波长、最佳双波长组合、最佳三波长组合和最佳四波长组合的校正方程,相关系数分别是最佳单波长r为0.789;最佳双波长r为0.893;最佳三波长r为0.970;最佳四波长r为0.943。试验结果表明:利用光谱图像技术无损检测油桃糖度是可行性的,为计算机图像对水果进行内部品质的无损检测提供技术依据。
The sugar content of nectarine was detected by using spectral imaging. The spectral imagings were captured in wavelength of 632 nm, 650 nm, 670 nm, 780 nm, 850 nm and 900 nm. The lorentzian distribution （LD）, gaussian distribution （GD） and exponential distribution （ED） with three parameters were used to fit scattering profiles for all wavelengths. LD was found to be of the best function for fitting gray distribution of imaging. The multi - linear regression model relating lorentzian parameters to nectarine sugar content was developed using best single wavelength, double wavelengths, three wavelengths and four wavelengths. The best model with three wavelengths was able to predict nectarine sugar content with r = 0.97. The results showed that the multi -spectral scattering imaging is nondestructive, fast and easy to implement, and it can provide a nondestructive means for measuring nectarine sugar content.
出处
《江西农业大学学报》
CAS
CSCD
北大核心
2007年第6期1035-1038,共4页
Acta Agriculturae Universitatis Jiangxiensis
基金
国家自然科学基金项目(30460059)
江西省教育厅科技项目(赣教技字[2006])
关键词
光谱图像
糖度
油桃
多元线性回归
spectral imaging
sugar content
nectarine
multi - linear regression
