International Journal on Advanced Science, Engineering and Information Technology, Vol. 8 (2018) No. 1, pages: 93-98, DOI:10.18517/ijaseit.8.1.2752

Extraction of Morphological Features of Malaysian Rice Seed Varieties Using Flatbed Scanner

R. Ruslan, A. A Aznan, F.A Azizan, N. Roslan, N. Zulkifli


A high quality cultivated rice seeds are very important for Malaysian paddy industry to ensure a high yield of paddy production. Certified seeds that are mixed with other varieties and unwanted seeds such as weedy rice are considered as poor quality and faced rejection during a quality inspection by the Department of Agriculture. To ensure the seeds are cleaned from any foreign seeds, it is very important to develop a low cost and simple mechanism to classify the seeds according to its varieties. The use of a flatbed scanner is one of the alternative techniques for image acquisition of the seeds varieties. This study was carried out to evaluate morphological features of local rice seed varieties developed for Malaysian rice industry using image processing techniques. Image of four seed varieties, mainly are MR219, MR220, MR263, and MR269 were acquired and extracted using a normal desktop flatbed scanner. A LabVIEW program was developed to extract four main morphology features which are length, width, aspect ratio and rectangular aspect ratio. The extracted data were analysed in terms of its spread and variability. One-way ANOVA was done to compare the means of the morphological features. Further t-test analyses were done to distinguish between two seed varieties based on the variation in the morphological features of the seed kernel. The results indicated that seed length parameter extracted from the image acquired by the flatbed scanner is significant to differentiate the cultivated rice seed except for MR269 and MR220. Seed width can be used as a parameter to distinguish MR269 and MR220 pair. Thus, a combination of morphological parameters is necessary to classify the cultivated rice seed.


rice seed; image acquisition; machine vision; classification

Viewed: 1227 times (since abstract online)

cite this paper     download