Smoothing algorithms for clip-and-paste model-based video coding
Shu-Sheng Hao, Ming-Feng Lee, Jar‐Ferr Yang · IEEE Transactions on Consumer Electronics · 1999
Model-based video coding has been adopted as a core experiment in the ISO MPEG-4 standard. The clip-and-paste technique for putting video objects in line is an important tool to reduce the transmission rate. To assist the clip-and-pasting method fitting into the 2-D model, we propose several smoothing algorithms for improving the quality of the reconstructed images. In this paper, the proposed smoothing algorithms can adjust the deformation of the zoom, tilt, and rotation of the object images. A luminance smoothing algorithm is also applied to compensate the light source variations. Simulation results show that the smoothing methods help to improve clip-and-paste images to achieve a satisfactory quality in visual perception.