A new model for the automatic relative radiometric normalization of multiple images with pseudo-invariant features

Relative radiometric normalization (RRN) with multi-sensor images is required for land-cover change detection. However, there are only a few RRN studies using multiple sensors. This article presents a new method for normalizing multiple images with pseudo-invariant features (PIFs) (MIPIF), which includes automatic selection and step-by-step optimization of PIFs. The normalized difference water index (NDWI) was used to select the original PIFs, and statistical rules with iterative control were used to fix the final PIFs. The method was tested on multiple images from a single sensor and multiple sensors in four groups of experiments with different land-cover areas. The results show that the normalization coefficients exceeded 0.90 at a significance level of 0.01. For the reference and normalized subject images, the root mean squared error (RMSE) values of the PIFs were much smaller than those of the reference and original subject images. The difference histogram curves of the reference and normalized subject images in the PIF pixels had roughly narrow normal Gaussian distributions with one pick around the zero position. The results demonstrated that the MIPIF method considers the physical definition of the PIFs and is effective, stable, and applicable for multiple images from a single sensor and from multiple sensors.