Identification of Discrimination Factors for Development of Optical Soybean Sorter

대두의 광학적 선별장치 개발을 위한 선별 인자 구명

  • 노상하 (서울대학교 농업생명과학대학 생물자원공학부) ;
  • 김현룡 (서울대학교 농업생명과학대학 생물자원공학) ;
  • 황인근 (서울대학교 농업생명과학대학 생물자원공학부)
  • Published : 1998.08.01

Abstract

Spectroscopic analysis of soybean kernels were made in the wavelength range of 400 to 1100 nm to find effective discrimination factors which are required for developing an opitical soybean sorter. Soybean samples used for the test were the sound and five classes of the defective kernels such as the immature, discolored(brown and violet), damaged by insect and diseased. Effective discrimination factors to classify the soybean kernels into the sound and the defective were found to be $R_{640}$, $R_{580}$/ $R_{990}$, $R_{600}$- $R_{820}$ and ( $R_{590}$- $R_{820}$)/ $R_{990}$. with classification error of less than 4%. Mahalanobis distance was used as a criterion to select significant wavelengths involved in the discrimination factors.s.

Keywords