(英) |
In Hokkaido, fishing salmon, squid, and other representative regional marine products could be better, and research and development in fisheries science utilizing information technology are required to sustain marine resources.
Furthermore, since forecasting the salmon catch and operation period is based upon the prediction for the amount of salmon migrations, a salmon-age estimation system with the high accuracy and efficient procedure has been required.
In this paper, we propose a method to determine the age of salmon by using false age rings and contour information of scale images to automate the age determination process using salmon scales. The method achieves stable age estimation by removing fake annual ring and contour regions that cause misjudgment when estimating age from resting zone images detected using a neural network. In experiments, the proposed method is applied to the detection results using the resting zone detection model, and its performance is evaluated to verify its effectiveness. |