Similar Video Search System based on Feature Data Extraction

摘要

This study introduces a similar video search system based on object detection data. The system detects feature objects of video thumbnails, obtained from a client using video edit and encoding program. The client transmits data such as an area of an object and a distance between feature points calculated from object outlines and features to a video server and, then, the server compares the data with object/video data of other videos in order to extract similar videos. The similar videos extracted are used to calculate a probability of a similar video and provide a list of similar videos in order to help users create original contents.