This technology is a single-field deinterlacing method that establishes a conditional probability model by modeling the statistical characteristics of the original image as a Gaussian distribution, and restores missing lines based on weighted least squares error by setting weighted local windows on the interlaced image.
Spatial deinterlacing methods such as line averaging, ELA, and EELA are easy to implement but fail to reproduce fine structures because they rely on a limited number of candidate edge directions, and covariance-based techniques suffer from sharp drops in image quality if the estimation is inaccurate. Ultimately, they have limitations in visual quality and PSNR due to an inability to accurately estimate the underlying structure of the original image.
This technology approximates the relationship between high-resolution and low-resolution pixels using local statistical properties and performs least squares estimation by assigning different weights to each pixel within the window, allowing it to estimate a high-resolution image close to the original using only a single field. It can be applied to real-time video processing chips and low-latency broadcast monitoring equipment that operate without motion information, enabling sharp-edged images without the need for separate frame memory.
N/A
