Silkworm egg image analysis using different color information for improving quality inspection

To increase the productivity in agricultural production, speed and accuracy are key requirement. In this paper, we proposed image analysis technique for silkworm egg quality inspection. We focus on silkworm images from the last incubation period because it can fully provide statistics of successfully hatched silkworms. Those statistics are useful information for both quantity and quality aspects. The images from the last incubation period contain three different types of egg including shells, defect eggs and unhatched eggs. As a consequence, it is intuitively obvious that the images of last incubation period have higher complexity than the other periods. Our technique use different color images obtained from each incubation period to tackle the problem. We demonstrate a simple method in object detection and classification. The experimental results show that our approach can improvement accuracy for both detection and classification.