自适应阈值Harris算法遥感图像配准的FPGA实现

FPGA implementation of adaptive threshold Harris algorithm for remote sensing image registration

针对Harris角点检测器响应值R的阈值选择而导致角点失真问题,文章提出了一种基于现场可编程门阵列(FPGA)的自适应Harris角点检测器实现遥感图像的配准方式。该方式依据非最大值抑制(NMS)处理后的响应值对阈值进行实时变化。实验结果显示,优化架构在硬件资源仅增加2.76%的情况下,准确率相应提升了8.31%。因此,文章提出的遥感图像配准架构适用于硬件资源有限的平台。

Aiming at the problem of corner distortion caused by the threshold selection of the response value R of Harris corner detector,an adaptive Harris corner detector based on FPGA is proposed to achieve remote sensing image registration.This method changes the threshold in real-time based on the response value after Non Maximum Suppression(NMS)processing.The experimental results show that the optimized architecture achieved an accuracy improvement of 8.31%with only a 2.76%increase in hardware resources.Therefore,the remote sensing image registration architecture proposed in this article is suitable for computing on platforms with limited hardware resources.