基于FPGA的实时视频图像几何校正系统设计

Real-time Video Images Geometric Correction System Based on FPGA

基于后向映射查找表结构的几何校正算法的研究,针对双线性插值算法的特殊性,提出了四邻域像素的图像存储和插值方法,实现了图像插值时对静态存储器SRAM的快速随机读写。根据该方法设计的基于FPGA的实时视频图像几何校正系统的视频延时小于一帧,只需要改变视频图像的后向映射查找表,就可实现不同映射的视频图像几何校正。

Based on research of backward mapping lookup table architecture for geometric correction and the particularity of bilinear interpolation, the image storage and interpolation method of four neighborhood pixels is put forward. And it realizes how to make use of the frame buffer technology of four neighborhood images to achieve fast random access to SRAM. According to this method, the video delay of real-time image geometric correction system based on FPGA is one frame less than before. Only modification on video image backward mapping lookup table architecture is required to realize different video image projections geometric correction.