模糊图像边缘精确定位的滤波算法

图像的边缘是图像最基本也是最重要的特征之一。实际应用中,光学系统离焦和图像传感器噪声源引起图像模糊,导致边缘自动定位算法无法精确确定边缘点。从边缘检测理论出发,根据噪声及图像模糊模型,提出了基于空间域和频域的滤波方法对模糊边缘图像去噪,从而消除模糊图像对于边缘定位的影响。实验表明,各种滤波方法对于边缘点的自动提取都具有一定的效果,离焦模糊和噪声可以得到有效抑制,尤其是频域低通滤波器能够精确地提取出模糊边缘位置。

Study of filtering methods for accurate edge detection of blurred images

Edge is one of the most fundamental and significant features in an image. In practical applications, the acquired images are often blurred due to de-focus and the inherent noise of the imaging sensor. Consequently, the image edge cannot be accurately located by automatic location algorithms. Based on the edge detection theory and the theoretical model of image blurness, this paper presents both spatial domain and frequency domain image filters for reducing noise effects on blurred image edge detection. The experimental results show that all the presented filters can be of some help to the blurred image edge detection. Especially, frequency domain low-pass filter method is shown to have the best effect in accurately detecting the blurred edge position.