局部能量模型边缘检测方法研究

Study on Edge Detection Based on Local Energy Model

绝大多数边缘检测算法都是基于梯度的边缘检测,但是随着纹理复杂度加深,梯度会变得模糊,在边缘图像中很难找到有用信息。提出了一种采用局部能量模型来检测健全和有用的图像边缘的新方法。该局部能量模型与人类视觉系统相关,采用非标准的感受野(Non-CRF)抑制进行感兴趣边缘检测。实验结果表明,在感兴趣的边缘检测中,新方法比梯度模型(Canny)算法取得了更好的检测效果.

Most of the edge detection algorithms are based on the gradient method.When texture is complex,the gradient may be cluttered.And the useful information cannot be found in the edge map.In this paper,a local energy model to detect robust and useful edges is given.The local energy model is related to the Human Vision System（HVS）.And non-classic receptive field（Non-CRF） inhibition methods are used to detect interested edges.The results show the proposed method is much better than gradient methods in detecting edges of interesting.