The pseudo-perspective transformation for motion compensation in digital video compression.

摘要

Various image (transform coding) and video compression (motion estimation-motion compensation) schemes have been proposed and implemented to meet the growing demand for multimedia (audio-visual) applications, which involves processing huge amount of video information. These methods work by reducing the redundancies in the spatial, spectral, and temporal dimensions of the video data. The earliest and the most widely used motion compensation method is based on a block-matching translational motion model. This model is not adequate to capture video dynamics like rotation, zooming, and perspective distortion, which are common in video sequences.; One solution to overcome this problem is to use a deformable block motion compensation method with a more sophisticated model such as affine, bilinear, or projective mapping. Polynomial approximations of the perspective-motion model, such as the affine and bilinear models, have been widely reported in the literature. Nevertheless, the affine model lacks the correct degree of freedom required to capture the effect of perspective distortion between frames. Though the bilinear model has the correct degrees of freedom, it still cannot accurately model the perspective distortion, especially near the boundaries of an image block.; In, this research work, an eight-parameter pseudo-perspective mapping model is proposed for deformable block motion compensation. An efficient interpolation scheme based on the pseudo-perspective model is developed using the deformable block model and finite element shape functions. The proposed motion compensation algorithm is implemented and simulated under the MATLABRTM environment. The proposed motion compensation scheme is tested on benchmark video sequences and later applied to real-time aerial surveillance video sequences, 720 x 480 pixel NTSC frames gathered by an aerial surveillance vehicle. The implementation results show that the proposed motion compensation scheme yields an improvement of 1.5--2.3 dB in peak signal-to-noise ratio compared to a bilinear model-based motion compensation scheme. The design experiments and detailed implementation results are presented in the report.