Abstract: Conventional edge detection algorithms suffer from the shortcomings of undetected and so on. This paper proposes an algorithm based on UGM model for improving considerably the performance of image edge detection. This paper just adds some pertinent remark to listing the two topics of explanation: how to realize image edge detection based on UGM model; the results of analysis. In the first topic, this paper adds an enhancement parameter [q] in the error image to improve the edge definition. This paper does experiments and analyzes their results. The analysis of the experimental results show preliminarily that the UGM model can not only effectively detect the information of image edges but also keep their details.

Key words: edge detection, Grey Model（GM）, Undulating Grey Model（UGM）

摘要： 由于传统的边缘检测算法会产生信息漏检等问题，为了更有效检测出图像的边缘信息，提出了基于UGM灰预测模型边缘检测算法。该算法分别在垂直和水平方向上选取与考察点相邻的4个灰度值作为建模数据，通过UGM模型对建模数据进行处理，得到2幅预测图像，用原始图像分别减去2幅预测图像，根据正负得到4幅误差子图像，将4幅误差子图像相加，检测图像边缘。在此基础上，通过对误差子图像加入调节因子[q]，提高边缘的清晰度。该算法与传统算法的结果比较表明该方法能清晰地检测出图像边缘点，图像的细节信息很好地保留下来，且对噪声图像具有一定的抗噪性，说明该算法是一种非常有效的图像边缘检测算法。

关键词: 边缘检测, 灰色微分方程模型, 灰色摆动微分方程模型