一种贝叶斯和支持向量机相结合的相关反馈策略

A Relevance Feedback Strategy Combining Bayesian and SVM

相关反馈技术在提高图像检索性能方面发挥着重要作用,但图像检索过程中的相关反馈存在反馈次数过多,反馈效果不够理想等问题。为解决上述问题,提出一种贝叶斯和支持向量机相结合的反馈算法。实现方法是:用贝叶斯分类器对图像库进行分类,达到压缩图像库的目的,然后用支持向量机分类器对压缩之后的图像库进行分类,并反馈最终结果。研究结果表明,与支持向量机和贝叶斯算法相比,在很少的反馈次数下,该方法明显提高了反馈效果。

Relevance feedback technology plays an important role in improving image retrieval performance.However,the image retrieval process with relevance feedback technology also has many disadvantages such as too much feedback times or unsatisfactory feedback effect.In order to improve the relevance feedback method,we present a new relevance feedback strategy combining Bayesian and SVM technology.The main approach was achieved by firstly assorting the image library with the Bayesian classifier compressing the image library.Secondly,classifying the compressed image library with the SVM classifier,and lastly returning the worked out results.The presented algorithm was compared with SVM algorithm and Bayesian algorithm,the experiment results illustrated the accuracy of the feedback method significantly improved.