| 摘要 |
An object-oriented filter for filtering video images prior to video encoding is disclosed. The object-oriented filter is configured to include an image parameter extractor, a filter selector, and a filter. The image parameter extractor analyzes the input video sequences to map the location of pixels associated with one or more image parameters. Once the pixels associated with the image parameters are mapped, the filter selector ranks the identified image parameters based on their overall importance to the video sequence, and then selects a filtering factor for adjusting the strength of the filter used to filter the pixels associated with such identified image parameters. Thereafter, the pixels associated with each of the image parameters are filtered using the filter strength selected by the filter selector. An edge detector, skin detector, eyes-nose-mouth (ENM) region detector, and a moving region detector are optionally incorporated in the image parameter extractor when the image parameters identified are edge parameters, skin area parameters, eyes-nose-mouth (ENM) region parameters and moving region parameters, respectively.
|
