A Extract and Measure Method of Image Multi-Feature Vector on CBIR
Zhibin Ren · Journal of Yanan University · 2007
The traditional image retrieval methods are mainly dependent on the single vision feature,so the results are not always ideal.To cure this problem,the paper propose a new kind of image extract and measure method which is based on the multi-feature vector normalization and uses it as the standard of image's similarity measure.The method which is based on multi-feature vector and relevance feedback should have a very good effect.