|RAO, XIUQIN - Zhejian University|
|YANG, CHUN-CHIEH - University Of Maryland|
|YING, YIBIN - Zhejian University|
|Chao, Kuanglin - Kevin Chao|
Submitted to: Transactions of the ASABE
Publication Type: Peer Reviewed Journal
Publication Acceptance Date: 3/15/2014
Publication Date: 6/1/2014
Citation: Rao, X., Yang, C., Ying, Y., Kim, M.S., Chan, D.E., Chao, K. 2014. Differentiation of deciduous-calyx Korla fragrant pears using NIR hyperspectral imaging analysis. Transactions of the ASABE. 57:1875-1883.
Interpretive Summary: Korla Fragrant pears are small oval pears characterized by light green skin, crisp texture, and a pleasant perfume after which they are named. Deciduous-calyx pears contain fewer stone cells than persistent-calyx pears do; the stone cells are a type of thick-walled plant cell that can give the pear flesh a gritty texture. The deciduous-calyx fruits are considered more desirable in taste and texture attributes, and are sometimes identifiable by the shape of the calyx-end of the fruit (more concave than convex). In this study, near-infrared hyperspectral imaging was investigated as a potential method for automatic sorting of Korla Fragrant pears to differentiate the consumer-preferred deciduous-calyx pears from the persistent-calyx ones. Hyperspectral image analysis showed that the wavebands of 1190 nm and 1199 nm could be used for a two-waveband intensity ratio for such differentiation. The multispectral differentiation algorithm using the ratio of the relative intensity at 1190 nm divided by the relative intensity at 1199 nm, within the region of interest on each pear, was developed to derive the differentiation threshold of 0.9986. The algorithm correctly classified 89.3% to 94.0% of the deciduous-calyx pears. This performance demonstrates another multispectral inspection algorithm that could be implemented by an online multitask hyperspectral imaging system used to simultaneously perform multiple quality and safety inspection tasks on an fruit processing line. These fruit inspection algorithms would benefit fruit industry processors in helping to optimize quality and minimize safety risks for fresh products according to consumer demand.
Technical Abstract: Near-infrared hyperspectral imaging was investigated as a potential method for automatic sorting of pears according to their calyx type. The hyperspectral images were analyzed and wavebands at 1190 nm and 1199 nm were selected for differentiating deciduous-calyx fruits from persistent-calyx ones. A multispectral differentiation algorithm using the ratio of the pears’ relative intensities at 1190 nm and 1199 nm was developed. The results showed that the algorithm could effectively differentiate pears such that the number of persistent-calyx pears misclassified as deciduous-calyx pears would comprise only 2.4% to 4.9% of all pears classified as deciduous-calyx pears, while 89.3% to 94.0% of deciduous-calyx pears were correctly classified. The result showed the potential of the algorithm for the real-world application.