Design Options of the Fuzzy Motion-Adaptive Algorithm

Abstract

The aim of this Chapter is to analyze the design options of the fuzzy logic-based system for video de-interlacing introduced in Chapter 2, which combines a spatial and a temporal interpolator accordingly to the presence of motion. Particularly, the research work presented herein is focused on studying two topics: 1) the convolution mask used to calculate the input to the fuzzy system; and 2) the fuzzy rules employed by the system. Regarding convolution mask, the influence of considering several neighborhoods has been studied by de-interlacing the battery of video and film sequences. The results are evaluated in terms of the outcoming efficiency and computational simplicity, which means facilitating the implementation. Once the influence of the convolution mask is analyzed and one of the options is chosen, the structure of the fuzzy system used to carry out the de-interlacing task is studied. Specifically, the rule base is extended from the initial approach with two rules up to a set of more sophisticated systems, among which the most complex one works with five rules. The results obtained by de-interlacing the battery of sequences leads to the selection of the most convenient approach in terms of efficiency and complexity. As a consequence of the research work developed in this Chapter, an improved fuzzy-logic based approach for motion-adaptive de-interlacing is obtained. The new solution uses the most attractive proposal for the convolution mask and the most effective rule base among the alternatives presented herein. This Chapter is organized in three Sections. In Section 3.1, the options for the convolution mask are presented and their performances are studied. After selecting one of the convolution masks, the rule base and the parameters of the fuzzy systems are studied in Section 3.2. Finally, the conclusions of this Chapter are expounded in Section 3.3.