Species classification of submerged aquatic plants using acoustic images in shallow lakes

Abstract

Acoustic imaging sonar is used for species classification of the submerged aquatic plants at lakes. We used a measurement system that combined a high resolution acoustic imaging sonar, i.e., a Dual-frequency IDentification SONar (DIDSON) with two types of concentrator lens, motion sensors, and GPS receivers to quantify the underwater status of the lake. Our survey system could efficiently accumulate information from beneath the lake surface and it could be used to reconstruct the spatial state of the lake as a three-dimensional (3D) image at any time. In addition, we developed an image processing program to classify the aquatic plants using acoustic features of each aquatic plant. Two types of the field experiments were performed at Lake Yamanaka and Lake Yunoko in Japan. In the first experiment at Lake Yamanaka, 3D views of Myriophyüum spicatum were generated. In the second experiment at Lake Yunoko, we conducted a species classification of three aquatic plants, Myriophyüum spicatum, Chara globularis, and Elodea nuttallii.