基于GPU的红外传感器频域物理效应实时仿真方法研究

A real-time imaging simulation method of infrared sensors physical effects based on GPU parallel computing

随着计算机仿真技术的发展,由于具有抗干扰、低成本、高可靠性等特点,红外成像技术已经成为红外成像研究领域中一种强有力的技术研究手段,可广泛地运用于目标探测识别、武器装备研制、操作员训练等多个方面。本文针对红外传感器频域物理效应的建模与仿真,提出了一种基于GPU的改进图像卷积算法用于实时红外图像生成,在GPU顶点处理器和片段处理器实现算法的并行处理。通过与采用调制传递函数法实现的频域效应仿真结果进行对比分析,证实了此方法既可保证多种频域物理效应影响下红外图像的仿真精度,又大幅提高了渲染效率。

Due to advantages on anti-jamming,low cost and high reliability,infrared imaging technology is widely used in military field,such as target detection and recognition,weapon equipment development and operator training. In this paper,we take sensor physical effects in frequency domain into account and propose a novel image convolution algorithm for real-time infrared image generation which can be parallel achieved in GPU's vertex and fragment processors. By comparing the simulation results with images pro-cessed by MTF method,this method has been demonstrated to have high simulation accuracy and can significantly improve the rendering efficiency.