An Adaptive Joint Bilateral Interpolation-Based Color Blending Method for Stitched UAV Images

Abstract

Given a source UAV (unmanned aerial vehicle) image Is and a target UAV image It, it is a challenging problem to correct the color of all target pixels so that the subjective and objective quality effects between Is and It can be as consistent as possible. Recently, by referring to all stitching color difference values on the stitching line, a global bilateral joint interpolation-based (GBJI-based) color correction method was proposed. However, because all stitching color difference values may contain aligned and misaligned stitching pixels, the GBJI-based method suffers from a perceptual artifact near the misaligned stitching pixels. To remedy this perceptual artifact, in this paper, we propose an adaptive joint bilateral interpolation-based (AJBI-based) color blending method such that each target pixel only adaptively refers to an adequate interval of stitching color difference values locally. Based on several testing stitched UAV images under different brightness and misalignment situations, comprehensive experimental results demonstrate that in terms of PSNR (peak signal-to-noise ratio), SSIM (structural similarity index), and FSIM (feature similarity index), our method achieves higher objective quality effects and also achieves better perceptual effects, particularly near the misaligned stitching pixels, when compared with the GBJI-based method and the other state-of-the-art methods.

Keywords

• adaptive joint bilateral interpolation
• color blending
• stitched images
• subjective and objective quality comparison
• UAV (unmanned aerial vehicle) images