Video Object Retrieval Method Based on Color and SIFT Features
Guobing Li · Science Technology and Engineering · 2013
A method of video object retrieval which is combined color and SIFT features is proposed and achieved. Firstly,color histogram is used to search the target and sort the results of preliminary retrieval. Secondly,the SIFT features can be extracted from the retrieval results which is similar to the reference images. They can be matched with the SIFT features in the object area which is cut out from the reference image. Finally,the preliminary retrieval results can be sorted again according to the matching results of SIFT features. The object location and migration information returns to the control interface as feedback. The experimental results show that this method has good adaptability of object brightness changes,position translation,scaling and shape changes. By using this method the object can be retrieved accurately and it has high practicability.