Multi-View with Salient Depth Video via High Efficiency Video Coding Technique

Multi-view plus depth (MVD) video are scenes captured from multiple camera angles and often associated with its depth videos. MVD video is categorized as a type of 3D video. When dealing with MVD video delivery in a 3D video transmission, often it requires huge data transmission rates since multiple...

Full description

Saved in:
Bibliographic Details
Main Author: Mohd Noor, Norul U’yuun
Format: Thesis
Published: 2016
Subjects:
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:Multi-view plus depth (MVD) video are scenes captured from multiple camera angles and often associated with its depth videos. MVD video is categorized as a type of 3D video. When dealing with MVD video delivery in a 3D video transmission, often it requires huge data transmission rates since multiple videos are to be delivered simultaneously. The completion of the latest video compression standard known as High Efficiency Video Coding (HEVC) compression method was introduced recently in order to tackle the issue of delivering high quality 3D videos without affecting the transmission rate and quality of the 3D video. A method of depth video compression is proposed for the MVD video. The proposed method involves applying the Reduced Resolution Depth Coding (RRDC) method onto depth videos in MVD videos. The RRDC method reduces the resolution of the depth videos by utilizing the Down-Sampling Up-Sampling (DSUS) method. The proposed method manages to obtain bit rate savings that ranges from 13% to 32% for the synthesis view of the two views plus depth videos compared to the synthesis view of two views plus depth videos without applying the DSUS method. In a scene of an image or a video, the region where users tend to put more focus on while viewing the scene is known as the saliency region. The human visual system (HVS) tends to put more focus on salient region since the human brain does not have the capacity to process huge loads of information presented in the scene. A number of saliency models based on the HVS have already been developed, however very few focuses on the saliency model based on depth videos. A method of saliency depth video (SDV) is proposed, utilizing selected saliency map features and fusing them onto depth video sequences. The proposed method of SDV is used with MVD video resulting in a multi-view plus saliency depth video (MVD-SDV). The MVD-SDV is further compressed utilizing HEVC compression method.