| Popis: |
This paper proposed a content adaptive de-interlacing algorithm (CADI) for conversion of interlaced video to progressive video. The proposed CADI algorithm interpolates the missing field based on the local image characteristics (smooth, horizontal edge, vertical edge and texture) and chooses the most appropriate de-interlacing algorithm. In addition, a novel vector matching algorithm is introduced for interpolating missing pixels in the horizontal edge region. It has three features to increase the accuracy of estimating edge directions: an adaptive search range for different edge orientations, combining consecutive pixels for directional correlation estimation within the same edge region and consistency checking along the candidate directions. Experimental results show that the proposed CADI algorithm outperforms existing algorithms in various video sequences, and produces high quality de-interlaced frames. |
