Video short course

Abstract

The invention relates to a video short course for analyzing and extracting a video abstract according to the human visual characteristic. The video content is analyzed from the human vision by adopting a dynamic video abstract mode, and the video content clip is extracted according to the influence degree of the video content on the vision. Video shot classification is carried out, different shot types have different attractions on vision, and meanwhile, a face identification technology is integrated. The video content is analyzed from the vision of audiences, and video clips are extracted according to the color contrast of a video, which accord with a human vision system.

Description

technical field [0001] The invention relates to the technical field of multimedia video analysis and extraction, in particular to a video compendium for analyzing and extracting video summaries based on human visual characteristics. Background technique [0002] In the past few decades, the field of computer vision has made many major theoretical breakthroughs and technological advances in the acquisition, storage, operation and transmission of video data, but it is still in its infancy in the query and representation of visual information. With the rapid development of science and technology, in the past ten years, many universities, research institutes, military and large groups at home and abroad have begun to invest a lot of manpower, material and financial resources in the research and development of content-based video information retrieval, and strive to use the information that can be objectively reflected. The low-level features of video information itself are used ...

AI Technical Summary

Problems solved by technology

[0010] Although there are a variety of algorithms, there are still many problems due to the richness of video content: such as the distinction between large objects and camera movements, high-speed object or camera movements, illumination and flickering of lighting backgrounds, complex editing effects (such as Fragmentation, ripples, deformation, etc.) and other complex situations, the misidentification and missed recognition rates are generally high; due to the large amount of video data and long processing time, the recognition speed of many algorithms is still far from the actual application; the extraction is based on The key to the video analysis technology of the content, the previous research features mainly focus on low-level features such as color, texture, shape, etc., but these features cannot fully and accurately describe the meaning of the image