|
This paper introduces a practical color filter array (CFA) interpolation technique. Among a lot of technologies proposed in this field, the inter-color methods that exploit correlation between color planes generally outperform the intra-color approaches. We have found that the filtering direction, e.g. horizontal or vertical, is among the most decisive factors for the performance of the CFA interpolation. However, most of the state-of-the-art technologies are not so flexible enough in determining the optimal filtering direction. In this context, we propose an inter-color CFA interpolation using a local map called unified geometry map (UGM). In this method, the filtering direction is determined based on the similarity of the local map data. Thus, it provides more choices of the filtering directions, which enhances probability of finding the most appropriate direction. It is confirmed through simulations that the proposal outperforms the state-of-the-art algorithms in terms of objective quality measures. In addition, the proposed scheme is as inexpensive as the conventional methods with regard to the resource consumption.
|
