Multi-View Image Processing With Adaptive Filtering for Crosstalk Control

The patent applies different filtering strengths to different sub-pixels based on their individual characteristics. Specifically, it calculates a representative value for each sub-pixel from the difference values between adjacent view images, and selects filtering strength according to a correspondence map that maps representative values to appropriate filtering strengths. This local adaptation ensures that each sub-pixel receives the optimal filtering strength to reduce crosstalk while keeping the filtered result within the valid value range.

The patent dynamically adjusts the filtering strength parameter based on the representative value of each sub-pixel. By changing the filtering strength parameter according to the correspondence map (which relates representative values to filtering strengths), the system optimizes crosstalk reduction for each sub-pixel while preventing overflow. This parameter adaptation resolves the contradiction between strong filtering and value range compliance.

Instead of applying a uniform low filtering strength to all sub-pixels, the patent calculates individual representative values for each sub-pixel and applies locally optimized filtering strengths. This allows some sub-pixels to receive higher filtering strength when needed (when their representative values indicate high crosstalk risk) while others receive lower strength, achieving effective crosstalk reduction overall while maintaining value range compliance.

The patent makes the filtering strength dynamic by selecting it based on the representative value of each sub-pixel through the correspondence map. This dynamic adjustment allows the filtering strength to adapt to the actual crosstalk conditions of each sub-pixel, ensuring effective filtering where needed while preventing overflow where not needed, thus resolving the contradiction between filtering effectiveness and value range maintenance.