Rice Spikelet Yield Determination Using Image Processing

摘要

There is a lot of Philippines rice varieties that need profiling according to Philippine Rice Research Institute (IRRI) [1]. While current methodology involves manual process and normally takes more than a week to complete, improving and developing a new variety is thus slower due to this aspect. This study proposes an automatic technique to perform profiling of rice spikelets using camera system and image processing technology. The system automatically counts and separates filled and unfilled rice spikelets using a camera, analyze the captured spikelet image by image processing, view and control the resulting data through a graphical user interface specifically designed to cater user's needs. Different varieties of rice samples are tested and validated with the current methodology. Result from 30 samples for each rice variety showed device accuracy of more than 95% was obtained.