Motion estimation and motion compensation are two interrelated and obviously important tasks in the total video coding process. Out of several motion estimation methods block matching algorithm (BMA) is the simplest, hence rightly adopted by several MPEGs and H.26X standards. Motion compensation is always associated with the motion estimation technique, which evaluate the error or residual image (block-wise) between the original frame and predicted frame. But the associated overhead of the error image requires more storage/channel bandwidth. In this paper we present an efficient motion compensation technique based on a predefined threshold value, we termed it as d-threshold value (γ). For applying our motion compensation technique we choose a recently reported modification of Hexagonal BMA, Threshold based Hexagonal (TBHEX) BMA [1] which has proved its efficiency in reducing the number of search points while maintains a good reconstructed image quality. Experimental results show that our approach requires maximum 29.60% of total block's motion compensation on standard video bench-marks.
[...] For applying our motion compensation approach we choose a recently reported Threshold-based Hexagonal BMA modification of Hexagonal BMA), which has less number of search points than that of Hexagonal BMA while maintains the quality of the Hexagonal search. Rest of the paper is organized as follows. In Section II we discuss the TBHEX BMA. Our proposed motion compensation model is presented in Section III. Experimental results and analysis are presented in Section IV. Finally the paper is concluded in Section V. [...]
[...] Such situation may arise due to the complicated motion like zooming, occlusion, rotation and arrival of new object Frame replenishment is a good technique which checks the active regions (which do not have significant correlation with the reference frame) of an image and stores/transmits the information of the all active regions while the passive regions PSNR d 2 ) = 10 log10 (which have significant correlation with reference frame) are represented by motion vectors. But the frame replenishment technique requires massive data storage or large bandwidth channel for storing or transmitting videos. [...]
[...] For a high motion video if γ is very less (say then in most of the cases algorithm evaluates that for the block motion compensation requires, hence the error information of that block needs to be taken care. While choosing a very large value (say 128 in gray scale image) will evaluate that for few numbers of blocks, motion compensation requires but the quality of the reconstructed image degrades a lot. Hence we can say that there is clear trade-off between choosing a high d-threshold value and the reconstructed image quality. [...]
[...] CONCLUSION A new method of motion compensation on top of a recently proposed BMA (TBHEX) is presented in this paper. Our proposed version requires (for high complex motion sequence, Garden) to (for slow translational motion sequence, Claire) of total number of blocks whose motion compensation is required. Rest of the blocks can be represented only with the motion vectors. This is a significant achievement over the traditional motion compensation techniques where for all blocks motion compensation is done. But in our approach motion compensation is performed only when the MAD value is larger than that of the d-threshold value. [...]
[...] Our motion compensation model is stated below. Before defining the motion compensation technique some terminologies used in our algorithm is defined below. Cdis = Counts the number of the search blocks where motion compensation requires. Cblock = denotes the number of the blocks evaluated for finding the motion of a block in current frame. mc = denotes whether for a block motion compensation requires or not represents not require, ‘1' represent motion compensation requires ) = block at spatial co-ordinate in current frame C. [...]
APA Style reference
For your bibliographyOnline reading
with our online readerContent validated
by our reading committee
