Abstract
To select kernels for breeding from either naturally or artificially mutated samples, it is necessary to process batch samples in a short time. For this purpose, a continuous high-speed single-kernel brown rice sorting machine based on rice protein content has been developed. In Part 1, a measuring speed of approximately 500ms for the full spectrum range and overall sorting speed of approximately 2.8s for one kernel were obtained, indicating that this machine was ready for practical use. Then in Part 2, the accuracy of the estimation for protein content (DM) was found to be SEP=0.43%, which was concluded to be sufficient for breeding. In the present Part 3, improvements were made in both the feeding mechanism of the developed machine for obtaining the targeted sorting speed of 1.0s/kernel and in the optics for increasing accuracy. As a result, an overall sorting speed of 1.3s/kernel was achieved and the accuracy of the estimation of protein content (DM) was improved to SEP=0.39%. These results are sufficient for practical use, particularly for breeding and the detection of rice mixing.
