Motion trajectory based video content retrieval and delivery for small displays

Abstract

Adaptive Multimedia Content Retrieval and Delivery for small displays is one of the challenges faced by Multimedia Community. Input video is transformed to an output video by utilizing manipulations at multiple levels (signals, structural or semantics) to meet diverse resource constraints and user preferences with optimizing overall utility of the video. The proposed system is developed to display the retrieved video shot, by motion trajectories of individual object, in a small displays. This system needs video shots as the inputs whose motion vectors are extracted by using exhaustive search algorithm. This shot-level motion feature is linked across the consecutive frames of shot to form the motion trajectories. Remove redundant trajectories and preserve one motion trajectory from all the similar motion trajectories. The representative object motion trajectory is stored in a database. Query interface which allows users to search for similar video shots by giving query video clip as input. Similarity matching algorithm is used to retrieve similar video shot from the database by comparing their motion trajectories. In this paper, next, in order to display those retrieved video shots in a small display, shape information of moving objects are extracted using Region- Growing algorithm. The segmented foreground is scaled down and re-integrated with the repaired and directly resized background to deliver effective video shot for small displays. Keywords: Motion Trajectory, Exhaustive Search algorithm, Douglas - Peucker algorithm, Region Growing algorithm, Content-based video retrieval (CBVR).

View PDF