一种新的Kirsch边缘检测算法

A novel Kirsch edge detection algorithm

目的:通过改进传统的Kirsch边缘检测方法,提供一种具有较好的抗噪声能力和自适应能力的边缘检测新算法。方法:首先利用模糊数学方法和中值滤波去除图像中的背景噪声和随机噪声。然后,基于三阶样条小波系数构造4个滤波模板并获得滤波图像和极值点图像。最后,根据最大熵算法自动获取的阈值和极值点图像得到边缘图像,连接不连续的边缘使用边缘跟踪算法。结果:新的Kirsch边缘检测方法同传统的方法相比,具有较强的抗噪声能力,边缘图像信噪比平均提高7.45 d B左右。结论:通过改进Kirsch算法,提出了一种具有较好抗噪声能力和自适应能力的边缘检测新方法。

Objective To provide a novel edge detection algorithm which can effectively reduce noise and adaptively extract much edge information by improving traditional Kirsch edge detection method. Methods The background and random noises of images were firstly eliminated using fuzzy mathematical method and median filtering. And then 4 filtering masks were constructed based on the wavelet coefficients of cubic spline wavelet, and the filtering images and local maximum images were obtained. Finally,the edge images were obtained based on the local maximum images and the thresholds which were automatically selected using the maximum entropy algorithm. The edge images were further processed utilizing the edge tracing and noise eliminating methods.Results The improved Kirsch algorithm showed stronger resistance to noise than Kirsch method, with a signal-to-noise ratio increased by 8.67 dB compared with the average value. Conclusion By improving Kirsch algorithm, we introduce a novel edge detection method with stronger resistance to noise and better adaptive ability.