摘要
基于模型设计方法实现基于细胞神经网络的红外图像边缘检测系统。将Simulink,Stateflow设计方法与CNN的设计理论相结合,建立CNN IP核模块。接着采用串行结构搭建图像输入、输出模块,达到可视化实时仿真及代码自动生成。仿真结果表明,基于模型的设计方法取得了较好的效果。在Xilinx公司Virtex-6系列的FPGA平台上,综合后占用极少资源的情况下达到了142.693 MHz的最高频率和7.927 Mpixel/s的处理速度。
Cellular neural network （CNN） achieves good results in the infrared image edge detection, but its application prospects are restricted by CNN hardware system. Considering the cycle is long and high-risk of CNN hardware development,the infrared image edge detection system based on cel- lular neural networks is achieved by model-based design method. Combining Simulink ,Stateflow with theory of CNN ,the [P core of CNN should be estab- lish. Then the input and output module is designed by the serial structure, the system is achieved the real-time simulation. The verilog-code of the system is automatically generated. The experimental results demonstrated ：the model-based design method obtain good results. Implemented target a Virtex-6 FPGA from Xilinx,system reach the 142. 693 MH2 of highest frequency and 7. 927 Mpixel/s of processing speed.
出处
《电视技术》
北大核心
2013年第3期29-32,共4页
Video Engineering
