摘要
基于PC的传统的视频压缩实时性较差,给视频的实时压缩和传输带来了困难。针对实时视频压缩传输的难点,提出了在FPGA平台上实现基于神经网络的视频压缩的策略。在建立了基于神经网络的视频压缩系统结构的基础上,给出了神经网络算法的设计,同时完成了非线性的神经元激励函数的线性逼近,并给出了典型FPGA模块的设计,最后通过DSPBuilder和Matlab工具对编写的verilog模块加以验证,给出了实验结果。
To address the difficulty of real-time video transmission and compression, a neural network based video compression strategy implemented on FPGA is proposed. After the building of the neural network based video compression system, the design of the neural network is completed. At the same time, the approximation of the nonlinear neuron activation fimction is completed and the designs of the typical FPGA modules are presented. At last, an experiment is implemented using Matlab and DSP Builder and the result is given out.
出处
《计算机工程与设计》
CSCD
北大核心
2009年第13期3082-3084,3132,共4页
Computer Engineering and Design