【摘要】 有限Ridgelet变换(FRIT)能够有效利用有限Radon变换(FRAT)和小波变换处理二维信号的一维奇异性。但是在有限Radon变换中,直线是以一种有别于几何方式而特殊定义的周期性"缠绕"直线。这种定义降低了FRIT的检测效果而且不能够检测直线的端点。提出一种无缠绕FRIT,从根本上解决FRIT的"缠绕"问题,并利用àtrous小波代替Mallat小波算法对改进的Radon变换进行检测提高了奇异点的检测精度。基于该方法的边缘检测能够对线段进行筛选与连接,并能有效检测目标的方向、端点、长度及宽度。实验结果表明,在具有一定噪声干扰的情况下该算法能够较精确地实现机场跑道及港口边缘的检测,克服了传统有限Ridgelet变换无法定位线段端点的不足。更多还原

【Abstract】 The finite Ridgelet transform(FRIT) deal well with the 1D singularity of the 2D signal by finite Radon transform(FRAT) and wavelet.However,the lines in the FRAT are the periodic wrap ones which are different with the geometrical.The definition has an influence on the detection purpose of the FRIT and makes it can not detect the points of the line.A novel algorithm is presented for solving the "wrap" of the FRIT radically by the no wrap lines,and could advance the detection precision by using à trous wavelet on the improved FRAT instead of Mallat wavelet.The method could choose and connect the lines,detect the points the length the width and the direction of the objects.The experimental results show that our method is precise to detect the edges of runways and harbor with the noise and comes over the FRIT defect.更多还原