3D-view Generation and Species Classification of Aquatic Plants Using Acoustic Images

抄録

In this study, we used a new integrated measurement system that combines the acoustic imaging sonar of DIDSON (Dual-frequency IDentification SONar) with concentrator lenses, motion sensors, and GPS receivers to find and identify different species of aquatic plants. Two types of experiments were performed at two lakes in Japan: Lake Yamanaka and Lake Yunoko. In the first experiment at Lake Yamanaka, the image was captured with a one-degree concentrator lens. Multi-beam image processing was used to generate 3-D images of aquatic plants. The lens used in this experiment concentrated the vertical beam width within 1 degree. The second experiment at Lake Yunoko DIDSON with a 3-degree concentrator lens was applied, and histogram and spatial spectrum analyses were performed for plant species classification. Three species of aquatic plants, Myriophyllum spicatum, Chara globularis, and Elodea nuttallii, were classified by the parameters obtained from the current methods. The ability to visualize the features of each aquatic plant for species classification, e.g., leaf and branch dimensions, was attributed to spatial spectrum and scattering analyses of the acoustic images. The high spatial resolution of the integrated DIDSON measurement system will contribute protection of endangered species in rapidly changing underwater environment.