Correcting Bidirectional Effect for Multiple-Flightline Aerial Images Using a Semiempirical Kernel-Based Model

Due to the intrinsic observing characteristics of airborne sensors, the bidirectional effect is inevitable and happens regardless of the number of flightlines being considered. This affects the quantitative use of aerial data over large regions. In this paper, a simple "two-step" bidirectional effect correction scheme based on Ross-Li model (RLM) is developed for multiple-flightline aerial images. First, the localRLMcoefficients and local correction factors (K1) for each flightline were derived independently based on original observed reflectance; next, the global RLM coefficients and global correction factors (K2) for all flightlines were derived based on simulated directional-to-nadir reflectance. Nadir view bidirectional reflectance distribution function (BRDF)-adjusted reflectance (NBAR) from multiple-flightline for fixed illumination condition of "base flightline" was produced using the combination of the two correction factors (K1, K2). Correction experiments conducted using multiple-flightline push-broom compact airborne spectrographic imager aerial images and the NBAR produced by the "two-step" BRDF correction method were compared to other "one-step" methods. The results show that the "two-step" method gave a better BRDF correction performance-a slower trend change for average of NBARs at a constant viewing angle as varying sun-target-sensor geometry, indicating that the trends of bidirectional effects within a given flightline and between flightlines were effectively normalized. It is concluded that our normalization scheme can be applied to remove bidirectional effects of multiple-flightline aerial images without multiangular observations if reasonable land-cover types in the aerial images were determined.