Estimation of rice plant population using digital image processing technique

C.C. Teoh, A.R. Mohd Syaifudin, O. Muhamad Isa and B.H. Abu Bakar

Abstract

Monitoring of rice plants population density is important for crop setting and fertilizer management to achieve high target yield. Currently, the population density is determined by manually counting the tiller number of total rice plants in a 25 cm x 25 cm square frame. Generally, several random sampling locations of a paddy plot are selected to perform tiller counting. This is time consuming, labour intensive and costly. An automatic counting tiller number method using digital image processing technique was introduced to overcome the problem. PCI image processing software was used to process 3 rice plant digital images obtained from MADA, Bukit Besar, Kedah paddy plots. The captured images were classified into plant and non-plant regions by image processing technique. The area of plant region in frame of each classified image was calculated by the software and used to correlate with tiller number using linear regression analysis. The relationship analysis result showed that area of plant has high relationship with tiller number with correlation coefficient value of 0.8328. A linear model was developed to estimate tiller number in the analysis. The model was verified by 00 rice plant digital images obtained from FELCRA, Seberang Perak, Perak in terms of tiller number estimation accuracy. The verification result showed that, the model capable to estimate tiller number with 92. 7% average accuracy. As a result, the image processing technique is practical, feasible and effective in estimating tiller number for monitoring of rice plant population density.

Full Text (131 KB)