A Spatial Deinterlacing Algorithm Based on Edge Orientation Optimized in Local Area

In this paper, a novel de-interlacing algorithm in spatial domain is proposed based on an edge orientation optimizing strategy. Taking advantage of the resolution invariant character of edge orientation, we calculate the edge orientation in interlaced image by minimizing the cost function, sum of absolute difference (SAD). Under the assumption that edge orientation in local area is stationary, SAD stands for the possibilities of each edge orientation. Interpolation in the orientation with the least SAD will bring in the continuous value of missing pixels. In addition, a threshold is set to distinguish the texture area from edge or plain area, which makes the algorithm more robust.