CIGR VI 2019

Presentation information

Poster Session

Others (including the category of JSAM and SASJ)

[5-1130-P] Other Categories (5th)

Thu. Sep 5, 2019 11:30 AM - 12:30 PM Poster Place (Entrance Hall)

11:30 AM - 12:30 PM

[5-1130-P-26] Classification of Salinity Damaged Spring Potato (Solanum tuberosum) using Hyperspectral Imagery based on Decision Tree Classifier

*KyungSuk Kang1, Sae Rom Jun1, Si Hyeong Jang1, Jun Woo Park1, Hye Young Song1, Ye Seong Kang1, Chan Seok Ryu1, Su Hwan Lee2 (1. GNU(Korea), 2. RDA(Korea))

Keywords:Hyperspectral imagery, Potato, Salinity, Decision tree, Classification accuracy

Salinity which is detected on reclaimed land is a major obstacle factor to crop growth. Currently, salinity is determined by experts directly examining the salinity of water and soil on farmland suspected of salinity. However, if salinity can be identified in real time and non-destructive way on the vast landfills, it can quickly respond to salinity to ensure stable cultivation. Accordingly, the objective of this paper is to verify the possibility of saline determination of non-destructively spring potatoes (Solanum tuberosum) through decision tree classifier using hyperspectral imagery of spring potatoes. In each vegetative period (VP), root formative period (RFP) and root growing period (RGP), the potatoes deal with treatment of normal watering, no-watering(drought) and salinity watering. The hyperspectral imagery of the treated potatoes was acquired at every midday. Individual potatoes canopies in hyperspectral imagery were extracted by a spectral imagery processing software (ENVI 4.7, Exeils Visual Information Solution Inc., USA). Reflectance data in the extracted canopies areas was used to classify each treatment. Calculated classification accuracy was evaluated by overall accuracy (OA) and kappa coefficient (KC). As a result, in all growth stage and treatment, the Rpart shows the highest classification accuracy. In particular, the classification accuracy was the highest between treatments OA 93.3% and KC 87.3% in the RFP that highly absorbs the moisture, and the lowest below OA 90.5% and KC 82.7% in the VP. As a classification of normal, drought and salinity using hypersepctral imagery, it showed that the possibility of salinity is different with spring potatoes in all the growth stage and it is also judged that these results can be applied as important basic results for further research to qualify and quantify salinity.