| Summary: | 碩士 === 國立高雄應用科技大學 === 電子工程系 === 100 === Traditional analog display devices (e.g. Analog cameras, CRT televisions) use interlaced scanning methods which divides single frame data into two fields data. Each field has information of even or odd lines of the frame. Nowadays, a development of display technology and a demand of digital devices (e.g. Notebooks, Plasma displays, TFT-LCDs) that convert interlaced format into progressive format is growing fast. Some annoying artifacts such as blurring, line flickered and jagged line occur during conversion. De-interlacing schemes remove those artifacts and make high quality frames.
In this thesis, we present a local edge direction de-interlacing algorithm based on spatial and temporal information to interpolate the missing pixels. Our proposed algorithm consists of three steps：scene change detect, content can classified state into four categories and content adaptive interpolation. We analyze the local region feature using the DVD(Displacement Value Detector) to detect and classify unknown pixels. In addition, there are also four main interpolation methods. The proposed method shows different performances according to spatial-temporal local features.
Experimental results demonstrate that the proposed method performs better than previous motion-adaptive de-interlacing.
|
|---|
