Similar Audio And Video Distributed Retrieval

Nowadays,people's demand for multimedia content has increased.Compared with the traditional text retrieval,multimedia content retrieval has been more difficult to obtain good results.Audio,image and video are most used multimedia content formats.How to design a reasonable and effective audio and video retrieval system has become a hot research topic in the field of computer vision.The main contributions of this paper can be summarized as follows:1.I propose an image retrieval method based on local feature description,which measures the similarity between video clips by calculating similarity on frame level.It combines the traditional SIFT+BOW with binary description based on the detected interesting points to describe image feature.By comparing recall and precision,experimental results show the good performance of the proposed method.2.I improve the performance of our available audio retrieval system by adding format normalization,OpenMP and loacal sensitive hash index structure.Experiments on a large number of internet dataset show the effectiveness of our imporved method.3.My proposed image retrieval method is transplanted to Hadoop platform.Decompose the task to each worker by slicing reference list.The experiment result shows that distributed processing approach can improve the speed of retrieval system.4.A Uyghur text detection method based on learning of baseline feature is proposed.Experimental results show that the detection accuracy is higher than 95 percent in well-binarized image.