This technology is a demosaicing device that restores raw data from an RGBW panchro color filter array, which includes white pixels, into full-color images. A Bayesian conversion unit converts white pixels into green equivalents using neighboring pixel averages that reflect light intensity to create a Bayer pattern, while an estimation unit calculates final colors using pentomino path-based directional gradient weights.
Panchro arrays offer superior light sensitivity compared to Bayer arrays because some pixels accept white light directly. However, most existing demosaicing algorithms were developed for Bayer arrays, making them difficult to apply to RGBW arrays. In particular, low-resolution integrated cameras faced limitations where the Bayer method alone resulted in significant intensity loss and reconstruction errors.
This technology calculates light intensity using a modified averaging filter, combining color conversion and light intensity reflection into a single step. It preserves edges by determining interpolation weights based on quincunx green planes and directional gradients derived from polyomino blocks. Applicable to image signal processors in smartphone cameras for low-light environments, dash cams, and surveillance cameras, it enables high-quality imaging with reduced color bleeding while leveraging the high sensitivity of white pixels.
N/A
