分数阶反卷积的高分辨力目标亮点提取
Target highlight extraction method based on fractional deconvolution
-
摘要: 提出了一种采用分数阶反卷积的目标亮点提取方法对目标亮点进行高分辨力提取。该方法依据分数阶域卷积公式, 引入反卷积理论去除窗函数的影响, 实现了时频域去模糊, 提升了时频分辨力。仿真和实验数据表明, 基于反卷积的短时分数阶傅里叶变换对线性调频信号的频域分辨力高于分数阶傅里叶变换, 可提高时频域上检测线性调频信号的性能。从而实现时频域上目标亮点的高分辨力提取, 提升目标亮点间时延差的估计精度。与传统方法相比基于分数阶反卷积的目标亮点提取方法具有更高的分辨力, 且对混响有一定的抑制作用。Abstract: A high resolution target highlight extraction method using fractional deconvolution is proposed to extract target highlights. Based on the fractional order domain convolution formula, deconvolution theory is introduced to remove the influence of window functions in this method. The time-frequency domain deblurring is realized, and the time-frequency resolution is improved. Simulation and experimental data show that the frequency domain resolution of the short time fractional Fourier transform based on deconvolution is higher than that of the fractional Fourier transform for linear frequency modulated signals. It can improve the performance of detecting linear frequency modulated signals in the time and frequency domain, thereby achieving high resolution extraction of target highlights in the time-frequency domain and improving the estimation accuracy of time delay differences between target highlights. Compared with traditional methods, the target highlight extraction method based on fractional deconvolution has higher resolution and certain inhibition effect on reverberation.