基于随机模型的边缘检测

在图象处理和计算机视觉中图象分割是一个非常重要的步骤，分割的好坏将直接影响很多后续处理的结果。虽然目前已有很多不同的边缘提取算法，但是这些方法很少考虑图象结构上的特点，而且从某种意义上讲所有基于差分的方法都是病态的。本文提出了一种边缘检测的随机模型方法，在这种方法中建立了两个模型，成象模型和基于磁引力模型的先验模型。在这两个模型的基础上我们分别采用模拟退火和贪心法进行求解，并将结果与其它基于差分的

EDGE DETECTION BY THE STOCHASTIC MODEL BASED APPROACH

Segmentation is one of the most important steps in the computer vision system. Although a lot of algorithms have been presented to extract edges,few of them pay attention to the property of image structure. Moreover, all the algorithms based on difference method are ill-posed. This paper proposes a stochastic approach for edge detection. Two models are established in the approachthe imaging model and the prior model based on the magnetic field. Based on the models, simulated annealing and the greedy algorithms are adopted to extract edges respectively. The experiments show that the proposed approach can achieve an ideal edge detection operation.