The conversion of images from interlaced to progressive format offers a lot of advantages in terms of subjective quality of the pictures at TV receiver and is much better suited for any pre- and post-processing.
In this thesis, a two-pass deinterlacing method using adaptive weighted median filter is proposed. In the first step, adaptive weighted median operation is done on each fields using current and two previous fields. Interpolation is based on the pixel differences of spatially neighboring pixels between current field and the field before previous fields. This operation detects motion existence and interpolates using spatial correlation if there is motion, using temporal correlation if there is no motion. In the second step, to compensate for the missing pixel in the moving region, motion compensaton technique with previous deinterlaced field and next future deinterlaced field is used. Motion vectors are estimated using the block matching algorithm with half pixel accuracy. For the interpolation of missing pixel, backward prediction and forward predictions are used. To avoid false interpolation at scene change or zoom region, selective adoption of adaptive weighted median interpolated pixel and motion estimated pixel is used.
The simulation results show that the performance of the proposed method is better than those of other conventional method(e.g.intra-field,inter-field) in terms of subjective quality of deinterlaced pictures and PSNR evaluated by comparing the interpolated value of missing pixel with its original value.