传统形态学滤波器滤波后的图像灰度值产生偏移,滤波算子抑制噪声效果较差.基于双路对偶形态学算子的混合滤波器解决了传统形态学滤波器存在的灰度值偏移问题,但运算次数多,处理速度较慢.提出了广义混合交变序列滤波器,该滤波器继承了一般混合交变序列滤波器的重要性质,保持了对偶滤波器的强对称性,且运算量减少了一半.仿真实验表明,该混合滤波器不仅去除了图像中的高斯噪声,而且保留了图像细节,最终滤波后的图像具有较高的峰值信噪比和较小的均方根误差.在保证峰值信噪比的前提下,广义形态学混合滤波器较一般形态学混合滤波器的处理速度提高了一倍多.%Traditional morphological filters perform poorly in reducing noise, and the grey values of the images filtered by them have a heavy deviation. Although the morphological hybrid filters can eliminate the deviation, they require much more operations and time. A new class of generalized hybrid alternating sequence filters (GHASF) is proposed, which can inherit the important properties of the general hybrid alternating sequence filters, maintain a strong symmetry of the dual filters, and have less than half computational complexity. Simulation results show that the new hybrid filters can not only effectively remove the Gaussian noise in the image, but also preserve the edge details. Besides, the images filtered may have a higher peak signal-to-noise ratio and a smaller root mean square error. Compared with the general hybrid morphological filters, on the premise of the peak signal-to-noise ratio, the generalized morphology hybrid filters can improve the processing speed by more than twice, on the premise of the peak signal-to-noise ratio (PSNR).
展开▼
