Modeling of Fish Behavior on the Basis of Outdoor Water Tank Experiment Data

1. 以虹鳟鱼为原料，在东京水产大学大泉训练站进行了室外水箱实验。用录像带记录了鱼类行为的观察资料。为了在更大的空间（5mX5m）内研究鱼类的行为，在东京水产大学进行了一项内水箱实验。将视频图像数据转化为学校中每个个体的位置和速度的时间序列数据。获得了鱼群规模和鱼群方向的时间序列数据。为了从视频图像数据中获取时间序列数据，提出了一种部分自动化的操作方案。实验结果表明，单鱼的自动作用率为98%，双鱼的自动作用率为83%，表明了所提方案的有效性。在假设鱼群鱼群行为可分解为两部分的前提下，建立了鱼群行为的数学模型。一种是学校代表的动作，用物理模型来表达。另一种是学校形式的变化，用自回归模型表示。计算机仿真结果验证了该模型的有效性。

1. An outdoor water tank experiment was made ao Oizumi Training Station of Tokyo University of Fisheries by using rainbow trout. The observation data of fish behavior was recorded by video tapes.2. An inner water tank experiment was made at Tokyo University of Fisheries by using Japanese dace in order to investigate the fish behavior in a broader space (5mX5m).3. The video image data was transformed into the time-series data of the position and the velocity of each individual in the school. The time-series data of the size and the direction of fish school was also obtained.4. In order to obtain the time-series data from the video image data, a partly automatic operation scheme was developed. According to the experimental result for bitterling, the automatic operation rate was 98% for a single fish, and 83% for two fishes, which shows the effectiveness for the proposedscheme.5. A mathematical model of fish behavior was constructed under an assumption that the schooling behavior of fish can be decomposed into two components. One is the motion of a representative of the school and is expressed by a physical model. The other is the variation of the school form and is expressed by an autoregressive model. The validity of the proposed model was examined from the result of computer simulation.