Development of the recognition method based on computer vision for white spots in the carapace of Portunus trituberculatus and research on its characteristics

To study the morphological characteristics of white spots in the carapace of Portunus trituberculatus, a computer vision-based recognition method was developed in this paper with an accuracy of 100%. The molting process of thirty crabs from sea area body weight (2.8±0.56) g was tracked to investigate the growth changes of white spots in the carapace of P. trituberculatus. Results showed that the number and relative positions of white spots had no significant changes after molting, however, the white spot area and its distribution area grew with the carapace. After molting, the white spot area increased by 89.33%±8.61%, the distribution area increased by 90.51%±7.95%, and the carapace area increased by 94.66%±8.26%. P. trituberculatus from North Korea, Liaoning and Zhejiang were collected to compare six parameters related to white spots trait in the carapace, and results showed significant geographical differences. The white spot traits of North Korea had the features of small spots, small distribution area and accounted for a low proportion of the carapace area. The traits of Liaoning had the features of small spots, larger distribution area and accounted for a high proportion of the carapace area. While the traits of Zhejiang had the features of big spots, small distribution area and accounted for a low proportion of the carapace area. This study developed the recognition method based on computer vision for white spots in the carapace of P. trituberculatus, and illustrated the growth features of this trait, and revealed significant geographical differences among three populations.