1st International Conference on Mechanical Electronic and Biosystem Engineering | |
Selection of Morphological Features in Classifying Weedy Rice and Rice Seed Varieties using Discriminant Function Analysis | |
Ruslan, R.^1^2^3 ; Bejo, S.K.^1 ; Rukunuddin, I.H.^2 ; Aznan, A.A.^2 | |
Department of Biological and Agricultural Engineering, Universiti Putra Malaysia, UPM 43400 Serdang, Selangor, Malaysia^1 | |
School of Bioprocess Engineering, Universiti Malaysia Perlis, UniMAP, Arau, Perlis | |
02600, Malaysia^2 | |
Center of Excellence for Advanced Sensor Technology (CEASTech), Universiti Malaysia Perlis UniMAP, Arau Perlis | |
02600, Malaysia^3 | |
关键词: Classification accuracy; Discriminant function analysis; Malaysians; Morphological features; Rice seed; Stepwise discriminant; | |
Others : https://iopscience.iop.org/article/10.1088/1757-899X/557/1/012014/pdf DOI : 10.1088/1757-899X/557/1/012014 |
|
来源: IOP | |
【 摘 要 】
A comparison on classification accuracy and morphological features selection between stepwise discriminant function analysis (DFA) and enter all variables DFA were made for classification of Malaysian rice seed varieties namely MR297, MR263, MR284, MR219 and a group of weedy rice. Eighteen morphological features were used for the discriminant analysis. The classification accuracies of MR297 and MR263 against weedy rice group were maintained at 99.1% and 98.9% respectively, either using stepwise DFA or normal DFA. The classification accuracy for MR284 decrease from 95.0% to 93.7% using stepwise DFA with the reduction to five morphological features selected in the analysis. However for MR219, the classification accuracy increased by 0.3% using optimized features in stepwise DFA. Thus it can be concluded the optimization of morphological features through stepwise DFA classification does not necessarily increase the accuracy for discriminating weedy rice and Malaysian rice seed varieties.
【 预 览 】
Files | Size | Format | View |
---|---|---|---|
Selection of Morphological Features in Classifying Weedy Rice and Rice Seed Varieties using Discriminant Function Analysis | 11020KB | download |