The key frame extraction technology is the basis of content-based video retrieval, and a key frame extraction algorithm is based on the lens for compressed video sequence. Its core is based on the compression encoding characteristics of video sequences, only need to be part of the decoding, use the DC component of Ⅰ-frame information to construct DC thumbnail, and combine the important difference of different areas of the image frame's DC information to make the similarity measure, and thus achieve key frame extraction. The experiments show that the algorithm is significantly improved than the traditional one about the two indicators of the recall and retrieval time, especially for the more dramatic news documentaries, films, and other local sports video sequences.
展开▼