多尺度MRF边缘检测

本文提出一种改进的多尺度MRF边缘检测方法．基于图像的八邻域结构在MRF模型中引入了Gibbs分布的线过程，有效地改善了MRF边缘检测算子的方向灵敏性．给出了对应于八邻域线过程结构的能量函数．基于该算法对检测边缘的精确定位性能，提出了一种容易的多尺度检测边缘合成规则．

MULTI-SCALE EDGE DETECTION BASED ON THE MRF MODEL

An improved multi-scale edge detection method based on the Markov random field (MRF) is proposed in this paper. The 8-neighborhood structure was introduced into the line-process that was employed in the Gibbs distribution in the MRF, which improves the directional sensitivity of the MRF edge detector efficiently. The constructing mechanism of the energy function correspondence to the 8-neighborhood line-process is given. On the basis of improving the location deviating of the detected edge in multi-scale, an easier method to synthesis the detected edge in multi-scale is given.