基于神经网络的红外目标分类算法设计与应用

Design and application of infrared target classification algorithm based on neural network

卷积神经网络在图像处理领域取得了突出表现,但是由于算法庞大的计算量引起功耗高和实时性差的问题导致神经网络的实际应用受到一定限制。如果将神经网络移植在FPGA硬件平台,可充分发挥其高度并行的优势实现网络加速,降低功耗并提升算法实时性。基于上述描述,本文将用于目标分类的网络模型成功移植在FPGA上,通过对比加入分类模型前后的告警结果,说明分类模型设计的重要性。对比硬件实现与仿真结果,证明硬件实现的正确性。

Convolution neural network has made outstanding achievements in the field of image processing.However,the huge amount of computation in the algorithm leads to higher power consumption and poor real-time performance,which limits the practical application of neural networks.If the neural network is transplanted to the FPGA hardware platform,it can give full play to its highly parallel advantages to achieve network acceleration,reduce power consumption,and improve the real-time performance of the algorithm.In this paper,the network model for target classification is successfully translated to FPGA.By comparing the alarm results before and after adding the classification model,which shows the importance of the designed model.By comparing the hardware implementation with the simulation results,the correctness of the hardware implementation is proved.