Video space-time search method

Abstract

The invention discloses a video space-time search method. The video space-time search method comprises the steps of searching a video set satisfying a time constraint according to a time search condition; sampling a searching object into a plurality of sampling spatial points, and sampling the shooting direction into a plurality of direction units; determining whether the sampling spatial points are in visual threshold of video frames, obstructed by barriers and clearly imaged or not, and calculating perception intensity of the current spatial point by the current video frame if conditions are satisfied; calculating the direction of the video frame shooting the sampling spatial point, and merging each video frame to a corresponding angle unit; and calculating the sum of the perception intensity of the video frames belonging to the same video in each sampling spatial point and each angle unit, and selecting the video having the maximum perception intensity as the object video, a set formed by object videos involved in all sampling spatial points and direction units being a space-time search result. An ordered list of comprehensively describing spatial object information is obtained on this basis.

Description

technical field [0001] The invention relates to a video spatiotemporal retrieval method, in particular to a video retrieval method considering temporal and spatial information such as video shooting time, spatial object position information, and spatial object shooting direction. Background technique [0002] Video data is a streaming media that contains comprehensive information such as vision, hearing, time, and space. With the increase in the deployment of surveillance cameras and the popularity of video acquisition devices such as smartphones, video data has exploded. The explosive growth of acquisition, sharing and usage of video data has brought great challenges to video retrieval. [0003] Current video retrieval can be divided into two categories: content-based video retrieval and metadata-based video retrieval. With the integrated application of space-related sensors (GPS, electronic compass, gravity sensor, gyroscope, etc.) The method is automatically calculated...

AI Technical Summary

Problems solved by technology

[0005] However, video data has spatial aggregation and redundancy of information expression in space, because multiple videos are usually taken for a certain object of interest, and these characteristics of spontaneous video data collection and uploading are more significant, which This leads to a large number of retrieval result data sets and dense spatial distribution, which brings difficulties to the visualization and further analysis of the result video collection.

Images