针对最优双线性插值算法进行精度优化,采用ARM A9处理器软件定点实现比浮点提高13.5倍处理速度.并提出一种高效的硬件实现结构,在Xilinx Zedboard开发板上验证了该结构,实验结果表明:在典型情况下可以比ARM A9软件定点实现提高9倍速度.
The paper optimize the accuracy of the opt-shifting bilinear interpolation.Implementation uses the ARM A9 processor software fixed-point implementation than 13.5times floating point processing speed.Then the paper puts forward a highly efficient hardware implementation structure and implemented on Xilinx Zedboard.The experimental results show that in typical cases:hardware can 9times faster than implementation by the ARM A9 software fixed-point.