| Popis: |
Rice whole crop silage is used to feed cattle in Japan. The silage contains grains of rice, which are not easily digested because the chaff is indigestible. Partially removing the rice chaff by mechanically scratching the unhulled rice surface was used to make the rice in the chaff digestible, however, it is not easy to inspect the scratching effect objectively. This paper describes the process of distinguishing between damaged chaff and undamaged chaff using image processing, programmed using Java2 SDK. To identify the damaged part on the surface of the chaff, color difference (brown and light white) was used. The chafed rice images were labeled individually for application to image processing one by one. RGB (red, green and blue) channels in a 24-bit color image were divided into three stacks. In the three stacks, the red slice owned more pixel value than any other slice, which reflected the difference between the chaff and damaged part. By adjusting the threshold appropriately, it could distinguish between damaged chaff and undamaged chaff with a probability of more than 90%. Furthermore, the rate of discernment became about 100% by adding shape analysis. This program will be integrated into rice whole crop silage processing equipment. These results are useful not only for the effective use of cattle feed, and maintaining healthy milk and beef cattle, but also for various other image processing uses, for example to distinguish small defects on the surface of grain. |
