本論文提出一個以邊緣為基礎，適應動作變化的除交錯演算法。為提高視覺效果並避免影像閃動，一開始先對影像做動作的偵測，將影像切割成動態和靜態的區域；同時也對整張影像做邊界的偵測，切割出強邊界，弱邊界和非邊界的部份。若屬於靜態的區域，則會檢查之後數張畫面的同一位置是否也是靜態，以減少雜訊的影響。在強邊界區域，先使用適應性搜尋，依據之前內插的方向，向周圍尋找可能的邊界。在濾除不相似的邊界後，對於剩下的每個強邊界中，各自尋找最相似的位置做內插。弱邊界區域如同強邊界區域做處理，但是內插時，是對於所有剩下的弱邊界中，只尋找一個最相似位置做內插。另外，切換器會偵測周圍區域的移動資訊以及鋸齒效應，並依此對動態和靜態區域做最後調整，並對靜態區域使用時間資訊做內插。畫面中屬於動態且非邊界的部份，則是使用一個結合空間跟方向資訊的內插法做內插。最後，將所有資訊混合輸出成一張完整的去交錯畫面。實驗結果證實，本演算法可以有效的提高影像的品質，尤其在水平邊界的部份有卓越的表現。

In this thesis, a edge-based motion adaptive de-interlacing algorithm is proposed to promote visual quality of video sequences. At first, a block-based motion detection is utilized to divide a field into moving and static regions. At the same time, the field is divided into strong edge, weak edge, and non-edge regions using edge detection. Static region will be checked upon the result of motion detection in the following frames to reduce effect of noise. In strong edge region, the nearby strong edges are adaptively searched in the direction of previous edge interpolation first. An edge interpolation is performed at the most similar position from each strong edge block after eliminating dissimilar ones and taking previous directional information into consideration. The operations of weak edge are the same as strong one except for performing an edge interpolation at the most similar position from all weak edge blocks. Pixel located at both motion region and non-edge region is interpolated by an interpolation method which refers to directional and spatial information. Pixel located at static region is interpolated by temporal interpolation. Finally, all interpolated pixel will be mixed to generate the final output.

Experiment results show that the proposed algorithm can promote image quality, especially in horizontal edges. And the proposed algorithm also outperforms previous motion adaptive de-interlacing algorithms.

[1] Fung-Jane Chang and Shen-Chuan Tai. A motion and edge adaptive de-interlacing algorithm. Master's thesis, National Cheng-Kung University, Tainan, Taiwan, R.O.C. Thesis for Master of Science Degree, 2003.
[2] Yu-Lin Chang, Shyh-Feng Lin, Ching-Yeh Chen, and Liang-Gee Chen. Video de-interlacing by adaptive 4-field global/local motion compensated approach. IEEE Transactions on Circuits and Systems for Video Technology, 2005.
[3] Tak-Song Chong, Oscar C. Au, Wing-San Chau, and Tai-Wai Chan. A content adaptive de-interlacing algorithm. IEEE International Symposium on Circuits and Systems, 2005.
[4] Tak-Song Chong, Au O.C., Wing-San Chau, and Tal-Wai Chan. A de-interlacing algorithm based on texture classi cation. ICCE. Digest of Technical Papers. International Conference on Consumer Electronics, Jan 2005.
[5] G.D. Haan and E.B. Bellers. Deinterlacing - an overview. Proceedings of the IEEE, 1998.
[6] Taeuk Jeong, Younghie Kim, Kwanghoon Sohn, and Chulhee Lee. Deinterlacing with selective motion compensation. IEEE Transactions on Optical Engineering, July 2006.
[7] You-Young Jung and Seungjoon Yang. An efficient de-interlacing technique using two types of motion information. IEEE Transactions on Consumer Electronics, 2003.
[8] Gwo-Long Li and Mei-Juan Chen. High performance de-interlacing algorithm for digital television displays. IEEE Journal of Display Technology, 2006.
[9] Shyh-Feng Lin, Yu-Ling Chang, and Liang-Gee Chen. Motion adaptive interpolation with horizontal motion detection for deinterlacing. IEEE Transactions on Consumer Electronics, 2003.
[10] Fabrizio Russo and Annarita Lazzari. Color edge detection in presence of gaussian noise using nonlinear pre ltering. IEEE Transactions on Instrumentation and Measurement, Feb 2005.
[11] Yanfei Shen, Dongming Zhang, Yongdong Zhang, and Jintao Li. Motion adaptive deinterlacing of video data with texture detection. IEEE, 2006.
[12] Yi-Shan Shiu and Shen-Chuan Tai. A de-interlacing algorithm based on true motion vectors. Master's thesis, National Cheng Kung University, Tainan, Taiwan, R.O.C. Thesis for Master of Science Degree, 2006.
[13] Ching tsorng Tsai, Chishyan Liaw, Chung chi Lin, Ming hwa Sheu, and Huann keng Chiang. An efficient video de-interlacing with scene change detection. IEEE ICICS, 2005.