Skip to main content
ARS Home » Research » Publications at this Location » Publication #118377
Title: DETECTION OF CONTAMINATION ON SELECTED APPLE CULTIVARS USING REFLECTANCE HYPERSPECTRAL AND MULTISPECTRAL ANALYSIS

Author
item MEHL, PATRICK - CATHOLIC UNIV. AMERICA
item Chao, Kuanglin - Kevin Chao
item Kim, Moon
item Chen, Yud

Submitted to: Applied Engineering in Agriculture
Publication Type: Peer Reviewed Journal
Publication Acceptance Date: 11/19/2001
Publication Date: N/A
Citation: N/A

Interpretive Summary: Apples are one of the most important fruit commodities in the U.S. market, with wide processing applications. Contamination of dropped apples leads to high incidence of non-pasteurized cider contamination resulting in possible illness outbreaks. Detection of apple contamination can be provided at two stages after harvesting: either before processing (e.g., fungal treatment, waxing, etc.) or after processing. This study proposes to detect contaminations during the most initial stage before any further apple processing. Hyperspectral imaging analysis was applied for the design of a multispectral detection system for rapid detection of diseased or defected apples. An adaptable three-channel camera was utilized in the multispectral imaging system. Three apple cultivars, Red Delicious, Golden Delicious, and Gala, were chosen for their wide popularity on the market and differences in the visible spectral range. Good separation of defects was obtained for Gala (95%) and Golden Delicious (85%) apples. However, separations were limited for Red Delicious (76%). This is the first time that simple statistical functions have been used to achieve good separations between normal and abnormal apples. This information is useful to the Food Safety and Inspection Service (FSIS), and Food and Drug Administration (FDA).

Technical Abstract: Contamination of apples is a food safety concern touching the general public and strongly affecting this commodity market. Accumulations of human pathogens are usually observed on surface lesions. Detection of lesions and pathogens is essential for assuring the quality and safety of commodities. We present the application of hyperspectral image analysis to the development of multispectral techniques for the detection of defects on three apple cultivars, Golden Delicious, Red Delicious, and Gala. Separate apple cultivars possess different spectral characteristics leading to different approaches for analysis. General preprocessing analysis with morphological treatments was followed by different image treatments and condition analysis for highlighting lesions and contaminations. Good isolation of scabs, fungal and soil contaminations, and bruises was observed with hyperspectral imaging processing using either principal component analysis or the chlorophyll absorption peak. Applications of hyperspectral results to a multispectral detection are limited by the spectral capabilities of our RGB camera using three band pass filters. Good separation of defects was obtained for Golden Delicious apples. However, separations were limited for the other cultivars. Having an extra near infrared channel will increase the detection level utilizing the chlorophyll absorption band for detection as demonstrated by the present hyperspectral imaging analysis.