HardwareSoftware Architectures for Low-Power Embedded Multimedia Systems by Muhammad Shafique & Jörg Henkel

Author:Muhammad Shafique & Jörg Henkel
Language: eng
Format: epub
Publisher: Springer New York, New York, NY

4.5.1 Adaptive Motion Estimator with Multiple Processing Stages

Now the constituting processing stages of the proposed adaptive Motion Estimator will be explained.1. Initial Search Point Prediction : An adaptive Motion Estimator starts with an Initial Search Point Prediction stage that provides a good guess of the vicinity where the best match has a high probability to be found. A good predictor provides a good starting point to converge quickly to the best match (i.e., near-optimal Motion Vector , MV). Based on the assumption, the blocks of an object move together in the direction of the object motion, spatial and temporal neighbors are considered as good predictor candidates. Therefore, MV of the current MB is highly correlated with MVs of the spatially and temporally adjacent MBs that have been previously calculated. A set of predictors is selected by analyzing the MV difference distribution between various predictors (see Eq. 4.12 and the predictor set below) and the optimal MV (i.e., obtained by using the Full Search algorithm). Figure 4.27a–c shows that spatial median predictor (Median Spatial , Eq. 4.12) has a higher correlation with the optimal MV compared to the temporal median predictors (Median Temporal1 , Median Temporal2 , Eq. 4.12). This implies that Median Spatial needs to be examined first as it has high probability to be the True Predictor 2 (i.e., to find a near-optimal MV). It is noticed that Median Temporal1 and Median Temporal2 are also highly probable to be the True Predictors especially when the MB is moving vertically or horizontally with a constant velocity.

Download

Copyright Disclaimer:
This site does not store any files on its server. We only index and link to content provided by other sites. Please contact the content providers to delete copyright contents if any and email us, we'll remove relevant links or contents immediately.