【摘要】 本文研究一种新的2—D卡尔曼滤波算法。该算法将广义似然比(GLR)检测方法由一维扩展到二维,结合Woods—Radewan缩减更新卡尔曼滤波(RUKF)模型设计GLR边缘测检器,根据图象边缘变化的情况在线地调整工作滤波器的参数或状态估计量。实验结果表明:该算法在增加很少运算量的条件下,可以有效地去噪并较好地保护图象边缘。更多还原

【Abstract】 A new 2-D Kalman filtering algorithm for eliminating edge smearing aswell as removing noise is developed. This work is an extention of previouswork on the detection and estimation of jumps in 1-D processing. Theapproach is to design a generalized likelihood ratio (GLR) edge detectorin an implementing Kalman filter, which adjusts the filter on-line accordingto the variation of image edges. Experiment results indicate that thisalgorithm is efficient and feasible.更多还原