【作者】 孙丽娅；

【导师】 石光明； 黄道君；

【作者基本信息】 西安电子科技大学 ， 模式识别与智能系统， 2004， 硕士

【摘要】 与线性滤波器相比较，非线性滤波器对于脉冲类、乘性类噪声的滤除，具有较好的效果，因此近年来受到众多学者的关注。堆栈滤波器是一类非线性数字滤波器，在数字信号和数字图像处理中有广泛的应用，展开这方面的研究具有实际意义和应用前景。堆栈滤波器的特点是具有阈值分解和堆栈特性，这两个重要性质将滤波过程转化到二进制运算，易于 VLSI 并行实现。本文从理论上具体研究了堆栈滤波器的原理结构和进行噪声滤除的过程，并通过仿真验证了所给算法的有效性。本文针对非线性阈值分解结构，进一步结合图像本身的特征，提出了一种新的动态非线性分解的方法，能够在保证图像信息的同时很大程度上降低滤波过程的运算量。仿真结果表明，与其它的算法进行比较，本算法具有一定的实用价值。堆栈滤波器应用于 SAR 图像噪声的去除，也是一个新的尝试。与已有的统计类滤波算法进行比较，堆栈滤波器也有较好的效果。结合堆栈滤波器的易于集成实现和并行处理的优点，这部分的研究为开发实时 SAR 图像滤波器提供了一定的理论基础。更多还原

【Abstract】 Nonlinear filters do a good job in suppressing impulsive and multiplicative noisecompared with linear filters. So they are widely used in digital signal processing andimage processing and worthy of research. Stack filters are a class of nonlinear digital filters possessing thresholddecomposition and stacking property. These two important properties ensure that thefiltering go on in binary filed only with easy comparison, addition and multiplication,which allow an efficient VLSI implementation. This paper studies the architecture andtheory of stack filters and the adaptive algorithms for designing stack filters. Someresults of simulation for the proposed algorithms are given to prove their validity. Then a new nonlinear threshold decomposition architecture is introduced to reducethe complexity of algorithms. By combining with the character of image, a dynamicnonlinear decomposition architecture is proposed to implement stack filters which canshorten the running time in the filtering procedure and improve the quality of the outputimage simultaneously. Comparisons on some criterion between this new algorithm andothers are provided to show the validity of our algorithm. Results of the innovative application on SAR images filtering are given in the end,which show that stack filters are approximately with noise filtering algorithms based onlocal statistics property. And as their superior in integration implement, researches inthis part supply for exploring the real time SAR filters.更多还原