Research Of Key Techniques In Virtual Marine Biological Swarm Simulation

The aggregate motion is one of the most common phenomena in natureespecially marine ecosystems. Either on land or in the sea, the aggregate motion has acommon feature: the behavior of the individual organisms is relatively simple. Whenthey cluster, however, it will exhibit complex group behavior. In recent decades, dueto the continuous development of film animation, virtual reality, computer games andother areas, crowd simulation has become a hot research direction. Although abundantgreat achievements in crowd simulation are being reached in many application fields,it is still insufficient to meet our expectations and requirements in the flexibility ofcontrol, reality of simulation, and the efficiency of computation in the simulation ofmarine crowd behavior.Based on the above research background, this paper selects marine crowds as theresearch object, designs and implements the crowd behavior model to simulate thecomplex group behavior. The main research contents of this paper are to buildhigh-precision models of marine life by skeletal animation technology, and on thisbasis to design the marine life crowds-node class and related properties-grid classwhich is used to high-level edit and control group attributes intuitively, flexibly,effectively with controller mechanism. At last, this paper implements the complexgroup behavior simulation algorithm in Graphic Processing Unit, which markedlyincreases the computing speed by using graphics processor parallelism. The maincontents and results of the research in this dissertation are summarized as follows:(1) The simulation of marine life modelThis paper converts the marine life model produced by modeling software intoFBX format. The animation of model is driven by the internal skeleton movement. Itrenders the model by combining the controller mechanism with the data which areread from bones and skin of the model. It also implements the complex computations of skeletal animation in GPU, which significantly increases the speed and accuracy ofthe model simulating.(2) GPU-based group behavior algorithmIn this paper, group behavior simulation algorithm and related features ofmodern graphics processors are fully analyzed and studied. On the basis of priorstudies, the entire simulation algorithm is decomposed into several parts. Each part iscompiled into an effect file, to implement the group behavior algorithm in GPU and toreduce the burden of CPU and to improve the rendering speed of the entire group.(3) Real-time interaction of marine biological crowdsThis paper designs the interactive program and the specific behavioral controlsof marine crowd model. It implements the flexible control and editing functions tocrowds’ properties with screen selection function based on the off-screen and the GPUreal-time transfer-parameter function based on the crowd node’s properties-grid,which makes the crowds move according to the operator’s intention and thusachieving real-time dynamic control to entire crowd model.This paper implements a complete marine crowd simulation model based on theabove theories and methods. The experimental results show that this model cansimulate the marine crowd behaviors realistically and get a high rendering efficiencywith a large number of sophisticated individual models and meet the requirements ofreal-time and interaction. This paper provides a new idea for large-scale crowdsimulation by implementing the computation of skeletal animation, crowd behavioralgorithm and rendering of individual model in GPU. Besides, the research resultshave a high application value, which can be applied in various aspects, such as virtualaquarium, marine teaching and demonstration, ocean scene simulation and so on.