Improved Active-mesh-based Motion Estimation by using Bidirectional Motion Estimation algorithm
Publish place: First National Conference on Electrical Engineering Isfahan
Publish Year: 1391
Type: Conference paper
Language: English
View: 852
This Paper With 5 Page And PDF Format Ready To Download
- Certificate
- I'm the author of the paper
Export:
Document National Code:
ISFAHANELEC01_142
Index date: 14 March 2014
Improved Active-mesh-based Motion Estimation by using Bidirectional Motion Estimation algorithm abstract
Motion estimation and compensation is an important part of existing video coding systems. Mesh motion estimationprovides smoother estimated intensity fields, better su~ective quality and higher PSNR, resulting in most cases in amore realistic motion field and smaller estimation error. However, In MME, 1he computation of a motion vector isaffected by its neighbouring nodes vectors 1hat lead to tedious computations and unfeasible me1hod in realtimesystems. In 1his paper, we utilize integral projection algori1hm in order to obtain preliminary translation estimation andshorten algori1hm search range. In 1he propose scheme, hi-directional motion estimation is proposed to construct 1hemotion vector field for 1he frame to be interpolated. The motion vectors inside 1he non-complete patches 1hatconnected to 1he frame borders are computed by 1he same MME interpolation patterns as in 1he interior patches, wi1htaking 1he two or one remained nodes into account. Fur1hermore, in case of bilinear motion, an activ<>-mesh-basedinterpolation outperfurms 1he current mesh-based approach by up to 5 dB for frames. After coding, gains of up to 4.5dB PSNR in comparison to current mesh-based motion estimation and 2 dB compared to 1he mesh based motionestimation by using integral projection algori1hm are achieved by using 1he Bi-directional motion estimation.
Improved Active-mesh-based Motion Estimation by using Bidirectional Motion Estimation algorithm Keywords:
Improved Active-mesh-based Motion Estimation by using Bidirectional Motion Estimation algorithm authors
Keihane Keikhosro Kiani
Shahid Bahonar University of Kerman, Kerman, Iran.
Siamak Talebi
Shahid Bahonar University of Kerman, Kerman, Iran