VIDEO SEARCH DEVICE, VIDEO SEARCH METHOD, RECORDING MEDIUM, AND PROGRAM

摘要

A video search device for video searches in which a user specifies the position and orientation of an object that should appear in a video. A receiver receives input of a still image, two reference positions in the still image and two target positions in a video frame. An extractor extracts a reference image containing the two reference positions from the still image. A searcher searches for similar frame images in which local images similar to the reference image are depicted, from frame images in the video, traces movement tracks of two noteworthy pixels at start positions corresponding to the two reference positions in a local image when time advances or regresses from a similar frame image in the video, searches for a target frame image where the two movement tracks approach two target positions, and produces videos containing the similar frame image and the target frame image.

主权项

1. A video search device, comprising:
a receiver for receiving input of a still image in which a reference object is depicted, a first reference position and a second reference position contained in a reference region in which the reference object in the still image is depicted, and a first target position and a second target position within a frame image contained in a video; an extractor for extracting a reference image depicted in the reference region based on the first reference position and the second reference position; and a searcher for searching for, similar frame images that are frame images contained in the videos and in which local images similar to the reference image are depicted, finding a first start position and a second start position corresponding to the first reference position and the second reference position in the reference region, in a local region in which the local image is depicted in the searched similar frame image, tracing a first movement track along which a first noteworthy pixel depicted at the first start position moves in the searched similar frame image and a second movement track along which a second noteworthy pixel depicted at the second start position moves, by scanning frame images contained in the video including the searched similar frame image in the chronological progress or retrogression direction from the searched similar frame image, searching a target frame image at which the first movement track reaches the neighborhood of the first target position and the second movement track reaches the neighborhood of the second target position, and producing as a search result a video including the searched similar frame image and the searched target frame image.