RiSe-IViS: rice seed inspection vision system

• Main Authors: Ruslan, Rashidah, Bejo, Siti Khairunniza, Rukunudin, Ibni Hajar, Ibrahim, Mohd Firdaus, Aznan, Aimi Athirah
• Format: Conference or Workshop Item
• Language: English
• Published: Malaysian Society of Agricultural Engineers 2019
• Online Access: http://psasir.upm.edu.my/id/eprint/76553/1/MSAEC-25.pdf; http://psasir.upm.edu.my/id/eprint/76553/

Rice seed production in Malaysia is greatly dependent on the purity of the cultivated paddy seed produced through the government certified paddy seed program. Under the program, the seeds to be marketed by the seed processors must undergo quality control protocol where the seed lots are sampled from the seed farms and seed processing plants for purity analysis by the Seed Testing Laboratory of the Department of Agriculture (DoA), the enforcing agency. The current inspection conducted by the laboratory is based on manual process which is laborious, time consuming (max 40 min for newbie while skilled operator takes about 15-20 min.). The process is also very subjective and error prone as it is influenced by the skills and experience of laboratory workers especially those involving segregating vague and indistinct special morphological or even textural and color features. A prototype (Patent ID: PI2018500018) of a machine vision-based paddy seed inspection system (RiSe-IViS) was developed to explore the possibility of replacing the existing manual method in distinguishing the weedy rice and cultivated rice seeds under the SJPM standard protocol with a modern, effective and efficient technique. The RiSe-IViS prototype developed consists of two parts i) hardware configuration ii) software protocol development. A user friendly graphical user interface (GUI) was developed to assist user for image acquisition and analysis. Analysis protocol was first developed based on the morphological features of the paddy seed and was tested for MR297 against weedy rice. The classification accuracy was achieved up to 99%. Validation of the protocol is to be carried out with local seed plant in Kedah to enhance the system. The RiSe-IViS is expected to classify major rice seed varieties available in Malaysia against the weedy rice variants with superior accuracy.