一种基于贝叶斯理论的目标验证方法。 提出了一种验证图像中候选目标的新方案. 综合目标的先验知识,将目标验证转化为给定候选目标的条 件下,图像特征观察集合...

一种基于贝叶斯理论的目标验证方法。 提出了一种验证图像中候选目标的新方案. 综合目标的先验知识,将目标验证转化为给定候选目标的条 件下,图像特征观察集合的条件概率问题. 同时,提取图像的短线段特征,利用短线段特征的观察,给出了一种目标 验证方法排除复杂场景图像中虚假建筑物目标. 在自然图像集上进行的实验表明,所提出的方法能有效排除虚假 目标,满足处理自然图像的要求.-Based on Bayesian theory the goal of authentication methods. A candidate for the target image to verify the new program. Comprehensive prior knowledge of the target, the target to verify the candidate into a given target under the conditions of image feature observe the collection of conditional probability problems. At the same time, extract the image characteristics of short-paragraph the use of short-term observation of the characteristics above, given a goal of authentication method to exclude false images of complex scenes building goals. Set in the natural image experiments show that the proposed method can effectively rule out the possibility of false targets, to meet the deal with natural images requirements.