Video and Image Processing Suite User Guide

ID 683416
Date 4/04/2022
Document Table of Contents

24.3.2. Resampling and Interpolation

The Warp Lite IP uses an inverse mapping from the output pixel positions to the input pixel positions. Generally, these output positions do not map directly to the available input pixels so the IP implements a resampling and interpolation algorithm. Bilinear interpolation uses the four input pixels surrounding the required input position.

Even though the IP expresses output (X,Y) positions as integers, the corresponding input (U,V) values are fractional because the mappings do not coincide with the exact pixel positions. The pixel position F corresponds to the position (floor(U), floor(V)) The figure shows the fractional offsets (ΔU and ΔV) as the fractional parts of U and V.

Figure 81. Kernel Data Positions Required for Bilinear Interpolation

The IP implements the required interpolation of the input data samples as supplied by the l line store. The IP provides information from the line store for each pixel position in the output image and includes the kernel data and the fractional U and V positions.