脐橙糖度光谱图像检测技术研究  

A Study on Non-destructive Measurement of Sugar Content in Navel Orange Using Spectral Imaging

作　　者：刘木华[1] 程仁发[1] 林怀蔚[1] 周小梅[1] 

机构地区：[1]江西农业大学工学院,江西南昌330045

出　　处：《江西农业大学学报》2007年第3期443-448,共6页Acta Agriculturae Universitatis Jiangxiensis

基　　金：国家自然科学基金资助项目(30460059)

摘　　要：对采集到的不同波长的光谱图像灰度分布进行洛伦茨分布(LD)、高斯分布(GD)、指数分布(ED)函数拟合,通过比较发现洛伦茨分布为最优灰度分布拟合函数。将脐橙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。结果表明:利用光谱图像技术无损检测脐橙糖度是可行性的。The sugar content in navel orange was detected by using spectral imaging. The spectral images in wavelength of 632 nm,650 nm,670 nm,780 nm,850 nm and 900 nm were captured. 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 have the best function for fitting gray distri- bution of imaging. The multi - linear regression model relating Lorentzian parameters to sugar content was development by using the best single wavelength, double wavelengths, three wavelengths and four wavelengths. The best model with four wavelengths was able to predict navel orange sugar content with r = 0. 863. The results show that the muhispectral scattering imaging is nondestructive, fast and easy to implement, and it can provide a nondestructive means for measuring fruit sugar content.

关 键 词：脐橙 光谱图像 糖度 多元线性回归 

分 类 号：O434.13[机械工程—光学工程] TN911.73[理学—光学]