基于小波的超分辨率算法研究及FPGA实现

Research and FPGA Implementation of Super-resolution Algorithm based on Wavelet Transform

由于超分辨率研究较火热,且在各个方面应用广泛,尤其是在家庭娱乐方面需求较大。如果利用纯软件的方法,处理速度会过慢,不适合于具体应用。利用基于小波效果较好的超分辨率算法,提出硬件实现方式,并给出各部分的实现方法以及设计实现。最后将整体以及各部分仿真,利用FPGA验证其方法的正确性。利用此方法增加图像分辨率可以在每秒30帧的情况下处理1024×1024大小图像,然而1024×1024大小的图像通过软件实现则需要较长时间。得到的硬件架构简练,使其更加容易应用到实际中。通过升级(并行处理)可以处理4 K图像。

The research of super-resolution is hot and it is widely used in many fields,especially in family entertainment.If you use a pure software approach,the processing speed will be too slow,so it is not suitable for specific applications.In this paper,a method of hardware implementation is proposed based on the super-resolution algorithm with better effect,and the implementation method and design implementation of each part are given.Finally,the whole and each part of the design are simulated,and the correctness of the method is verified by FPGA. This method can process 1024 × 1024 size images at 30 frames per second when increasing image resolution.However,1024×1024 size images require a much more longer time by software implementation.The hardware architecture of this article is concise,making it easier to apply to practice.4 K images can be processed by upgrading（ parallel processing）.