An image reconstruction algorithm based on semi-gradient edge-oriented interpolation

Aiming at the pseudo-color and edge jaggedness problems in the reconstruction of Bayer color filtered array images by traditional demosaicing algorithm, an image reconstruction algorithm based on semi-gradient edge-oriented interpolation is proposed. The algorithm uses the first-order differentiation of the green component, the second-order differentiation of the blue component and the red component to calculate the horizontal gradient and vertical gradient of the green channel, and recovers the green component along the direction with smaller gradient; it uses the semi-gradient as the weight parameter of the interpolation operation, and uses the recovered green component as the correction quantity in the interpolation operation to recover the red component and the blue component. The experimental results show that the algorithm improves at least 1.19db over the traditional bilinear interpolation algorithm and edge-oriented interpolation algorithm, and can improve the pseudo color effect and edge jagged effect of the reconstructed image more effectively.