【摘要】 为了能够在硬件上实现高质量、高效果的视频图像超分辨率,提出了一种基于FPGA的自相似性超分辨率算法的实现。首先介绍了基于自相似性的超分辨率算法的基本原理和算法流程,接着提出了视频超分辨率引擎硬件设计的系统结构,并详细叙述了超分辨率引擎系统的各个模块的设计及实现,分析了系统的速度、带宽及资源。最终,在Xilinx K7 FPGA上实现了该视频超分辨率引擎。结果表明,本设计可以实时处理1 080p@25 Hz视频放大为4K@25 Hz视频并在图像质量上取得明显的效果。更多还原

【Abstract】 In order to achieve high-quality,high-effect on hardware video super resolution,a design method of video super-resolution engine is presented based on FPGA. First,the basic theory and algorithm flow of local self-example algorithm is introduced.Then,on the basis of this theory,a new hardware design method for video super resolution is presented in detail. Meanwhile,the speed,band width and the resource usage of the system are analyzed. Finally,this design is implemented in K7 FPGA of Xilinx.Experiment results show that the design can process 1 080p@ 25 Hz video scaling to 4K@ 25 Hz video in real time and achieve significant performance.更多还原