抄録
Since chirimenjako contains impurities in the fished stage, it is necessary to sort them out to improve the product quality. In this paper, we propose an algorithm for sorting them using AdaBoost, by making it learn the feature of appearance, such as color and shape of them. To extract the features, we used Chroma-key processing for distinguishing background region, and in machine training, we used the weights of obtained from AdaBoost for setting up threshold weak classifiers. As a result, about 90 percent of impurities could be sorted out using our method.
