A Technique for Object Movement Based Video Synopsis
Jeeshna P.V1, Kuttymalu V.K.2
1Jeeshna P.V., MTech Student, Computer Science and Engineering, Sree chitra Thirunal College of Eng. Pappanamcode, University of Kerala/ India.
2Kuttymalu V.K., Assistant Professor, Computer Science and Engineering, Sree chitra Thirunal College of Engineering. Pappanamcode, University of Kerala, Thiruvananthapuram (Kerala), India.
Manuscript received on 15 August 2015 | Revised Manuscript received on 25 August 2015 | Manuscript Published on 30 August 2015 | PP: 258-264 | Volume-4 Issue-6, August 2015 | Retrieval Number: F4231084615/15©BEIESP
Open Access | Editorial and Publishing Policies | Cite | Mendeley | Indexing and Abstracting
© The Authors. Blue Eyes Intelligence Engineering and Sciences Publication (BEIESP). This is an open access article under the CC-BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/)
Abstract: Video synopsis is the process of preserving key activities and eliminates the less important parts to create a short video summary from the long original videos. These techniques are used for fast browsing, extracting big data, effective storing and indexing. The main application of video synopsis is video surveillance. Video synopsis techniques are broadly classified into two types: object based approaches and frame based approaches. Important frames and objects are extracted and viewed as the basic building block of the synopsis, while other less important frames and objects are removed. But these approaches cannot handle the complexity of the dynamic videos. The object movement based video synopsis method focus on the movement of a single video object, and removes the redundancies present in the object movement. It helps to generate the video synopsis. The proposed method is the combination of frame based and object movement based video synopsis will generate more accurate and compact video synopsis. First frame based method will remove the nonmoving frames and object movement based video synopsis handles the moving objects. In object movement based video synopsis method the moving parts are considered as important and nonmoving parts are less important parts. The moving parts are preserved and nonmoving parts are to be eliminated. The basic aim is to work at the level of object part, and to remove the nonmoving parts. This method consists of three stages: Object Movement Partition, Assembling and Stitching. In object partition first segment and track the moving objects with the help of Kalman Filter algorithm and extracts the parts. Partition each object into several semantic parts, which produces several part movement sequences. Remove the object and repair the hole by structure completion method. In assembling select the same number of part movements from each part sequence and moving parts are then assembled frame by frame and nonmoving parts are removed. Finally stitch the assembled parts to eliminate the gaps between the frames, thus the synopsized video is produced. Many researches are going on this computer graphics and computer vision area related to video synopsis.
Keywords: Video Synopsis, Frame And Object Based Method, Object Movement Based Method (OMBVS), Kalman Filter, MRF, Object Movement Based Method.
Scope of the Article: WSN