一种基于概率密度分布的图像边缘检测技术研究

A method of image edge detection based on probability density distribution

图像边缘检测是机器视觉和目标提取的重要步骤,直接决定目标检测的准确性,传统边缘检测方法大都只利用像素的梯度信息和局部邻域信息,使提取的图像目标边缘存在断点,文中引入像素的概率密度分布函数描述像素在整个图像空间的全局分布信息,提出一种新的图像边缘检测方法.仿真试验结果表明,与传统方法相比,新方法具有较好的图像边缘检测效果.

Image edge detection is an important step in machine vision and object extraction, and it directly influences the accuracy of object detection. Most of traditional methods of image edge detection only make use of the gradient information and local neighbourbood information of pixels, leading to the discontinuous points in the edge of extracted image. This paper introduces the probability density distribution function of pixels to describe distributed information of pixels in the whole image space, and proposes a new method of image edge detection. The results of simulated experiments show that the proposed method outperforms traditional ones.