Abstract: Canny edge detection is a relatively new edge detection operator which has the very good edge detection performance. So it is a very wide range of application. However when it uses classic Canny edge detection algorithm, there is a large limit in removing noise and determining the thresholds. So in view of the above weaknesses, this paper puts forward a new adaptive Canny edge detection algorithm based on modified bilateral filtering. First bilateral filter which is modified is used to remove noise instead of traditional gaussian filter. This can remove the noise while retaining edge information better. In addition, the high threshold is got by making the rate of between-cluster difference and within-cluster difference largest in determining the thresholds, this can make the algorithm more adaptive. Theory and experimental results show that the algorithm has very good accuracy and practicability.

Key words: Canny edge detection, bilateral filtering, within-cluster difference, between-cluster difference

摘要： Canny边缘检测是一种比较新的边缘检测算子，具有良好的边缘检测性能，得到了广泛运用。然而经典Canny边缘检测算法在去除噪声和门限确定上还有很大的局限，针对以上弱点，提出了基于改进的双边滤波的自适应边缘检测算法。用改进的双边滤波方法代替传统的高斯滤波平滑图像，这样在更好地去除噪声的同时保留了边缘信息；在门限确定上，用类间方差与类内方差之比最大的方法确定高门限，该方法具有较强的自适应能力。理论推导和实验证明该算法具有很好的准确性和实用性。

关键词: Canny边缘检测, 双边滤波, 类内方差, 类间方差