학술논문
融合加窗和自适应加权引导滤波改进的Canny算法 / An Improved Canny Algorithm Combining Windowed and Adaptive Weighted-guided Filtering
Document Type
Academic Journal
Author
Source
化工自动化及仪表 / Control and Instruments in Chemical Industry. 51(1):86-91
Subject
Language
Chinese
ISSN
1000-3932
Abstract
针对Canny算法滤波过程中图像细节部分缺失和无法有效去除噪声,并且无法动态识别图像边缘的问题,提出一种融合加窗和自适应加权引导滤波的Canny算法.该算法对Canny算法做出两点改进:一是使用加窗滤波器对图像进行处理,有效滤除噪声,然后对图像进行数学形态学滤波处理,有效增强图像边缘;二是引入分段权值模型对引导滤波算法的正则化参数自适应修正,使其能够根据梯度信息区分边缘和非边缘,从而达到动态识别图像边缘的效果.相比传统Canny算法,改进算法处理后的信噪比提高了1.05～5.22 dB,品质因数提高了0.16～0.35.实验结果表明:改进算法能有效滤除椒盐噪声,识别出比较清晰和完整的边缘.
Considering the fact that some image details are missing,the noise failed to be removed and the image edge failed to be dynamically identified in the filtering process of Canny algorithm,a Canny algo-rithm which combining windowing and adaptive weighted-guided filtering was proposed.In which,the algo-rithm makes following two improvements on Canny algorithm,i.e.adopting a windowed filter to process the image to filter out any noise effectively,and then making use of mathematical morphology filtering to pro-cess the image so as to effectively enhance image edge;introducing a piecewise weight model to adaptively modify regularization parameter of the guided filtering algorithm so that it can distinguish the edge and non-edge according to the gradient information and to achieve the effect of dynamic identification of image edges.Compared with the traditional Canny algorithm,the signal-to-noise ratio of the improved algorithm can be increased by 1.05 dB to 5.22 dB,and the quality factor increased by 0.16 to 0.35.The experimental results show that,the improved algorithm can effectively filter out impulse noise and identify relative clear and complete edges.
Considering the fact that some image details are missing,the noise failed to be removed and the image edge failed to be dynamically identified in the filtering process of Canny algorithm,a Canny algo-rithm which combining windowing and adaptive weighted-guided filtering was proposed.In which,the algo-rithm makes following two improvements on Canny algorithm,i.e.adopting a windowed filter to process the image to filter out any noise effectively,and then making use of mathematical morphology filtering to pro-cess the image so as to effectively enhance image edge;introducing a piecewise weight model to adaptively modify regularization parameter of the guided filtering algorithm so that it can distinguish the edge and non-edge according to the gradient information and to achieve the effect of dynamic identification of image edges.Compared with the traditional Canny algorithm,the signal-to-noise ratio of the improved algorithm can be increased by 1.05 dB to 5.22 dB,and the quality factor increased by 0.16 to 0.35.The experimental results show that,the improved algorithm can effectively filter out impulse noise and identify relative clear and complete edges.