HDR Image Quality Estimation Using Static-Dynamic Region Metrics

The patent divides the HDR image into static regions and dynamic regions using optical flow analysis. Static regions are processed using traditional HDR fusion methods while dynamic regions use motion-aware processing to reduce ghosting artifacts. This segmentation allows different processing strategies for different parts of the image, resolving the contradiction between maintaining detail and reducing artifacts.

The patent applies different quality metrics and processing techniques to different regions of the image. Static regions are evaluated using edge preservation metrics while dynamic regions use ghosting reduction metrics. This local quality approach ensures that each region is optimized for its specific characteristics, improving overall image quality while maintaining the benefits of HDR fusion.

The patent segments the quality assessment into multiple components: edge preservation in static regions, ghosting reduction in dynamic regions, and visual saliency weighting. This segmentation makes the complex perceptual quality metric computationally manageable by breaking it down into independent evaluable components that can be processed separately and combined.

The patent introduces visual saliency maps as an intermediary to bridge the gap between simple computational metrics and complex human visual perception. The saliency maps guide the quality assessment by highlighting important regions, allowing the system to focus computational resources on perceptually critical areas while maintaining alignment with human visual system characteristics.