一种自适应二值化的毫米波图像边缘检测方法

Millimeter-wave image edge detection based on adaptive binarization

针对毫米波图像分辨率较低、噪声大、成像物体边缘提取困难等问题,提出了一种基于自适应图像二值化的毫米波图像边缘检测方法。该方法通过对图像直方图的分析,自动地确定两个峰值的位置,得出阈值,对图像进行二值化,再对二值化后的图像进行边缘检测。实验表明,这种方法能较好地检测出毫米波图像中人与物体的边缘信息,同时又去掉了大多数由噪声引起的伪边缘。该方法避免了大量的方差计算,计算量仅为大津法的 50%左右,能够满足自适应性和实时性的要求。

In view of the problems such as lower resolution, large noise and difficult edge feature extraction, an edge detection method based on adaptive binarization for millimeter-wave images is proposed. Two peaks are automatically located after analyzing the image histogram, then the threshold is obtained and the image will be binarized by it. The edge detection for the binarized image will be carried out. The experiments show that clear edges of human and object images in millimeter-wave length can be detected and most of the pseudo-edges caused by noise can be eliminated. The method avoids large amount of variance calculation and its calculation quantity is only about 50% that of Otsu method. It can fulfill the demands of adaptability and real time.