A CORRELATIVE TECHNIQUE FOR CORRECTION OF SHADING EFFECTS IN DIGITAL MULTISPECTRAL 
VIDEO IMAGERY. 
Cindy Ong 
CSIRO, The Leeuwin Centre for Earth Sensing Technologies 
Private Bag, P.O. Wembley, W.A. 6014. Tel: (09) 387 0243 Fax: (09) 387 0121 
KEY WORDS: Method, Correlation, Correction, Multispectral, Video 
ABSTRACT 
This paper describes a technique for the removal of 
shading effects, caused by solar geometry and 
shadows, from airborne Digital Multi-Spectral 
Videography (DMSV) images acquired over the 
Comalco bauxite mines in North Queensland, 
Australia. The approach involves use of near- 
simultaneously acquired Landsat Thematic Mapper 
(TM) satellite data. The correlative technique 
produces imagery that retains a stable dynamic range 
and has very high numerical coherence in the overlap 
areas of sequential images. Calibration of the images 
to reflectances, using field spectral data, is performed 
before production of the seamless mosaics. The 
results of classification, using the corrected dataset, 
confirmed the adequacy of the correction procedure 
for its intended use in minesite rehabilitation 
monitoring. This paper describes the development 
and implementation of the technique and gives 
examples of the correction. 
Introduction 
Shading effects across Digital Multi-Spectral Video (DMSV) 
images (Lyon et al., 1994), sometimes called the “hot-spot” 
effect, is a significant problem. It is often referred to as the bi- 
directional reflectance factor, cross-scene or cross-track 
brightness, or differential illumination. It is visible on most 
aerial photographs and video images, appearing at the point on 
the ground in line with the aircraft and the sun. (The sun is in 
effect eclipsed by the aircraft). Maximum image-brightness 
surrounds that point as there is no visible shadow from 
vegetation structure. As the radial distance increases, so also 
does the amount of visible shadow in the image, causing 
reduced total reflectance. The shading is also dependant on 
particles in the atmosphere, pathlength, wavelength and the 
nature of the component surface material. Figure 1 shows an 
example of shading across two DMSV frames. 
  
of 10,000 feet with a maximum 37? field of view. 
Methods for removing shading by calculating a surface of 
average brightness, from a sequence of “uniform” frames in the 
flight line, and subtracting that surface from each image were 
evaluated. However, with the Weipa data, we were unable to 
obtain a satisfactory result. Most of the DMSV images 
covering the Weipa and Andoom minesites contain highly 
variable land cover, making it impossible to obtain enough 
uniform frames to derive an appropriate correction. 
Figure 1: Example of two DMSV frames from a flight line over the Andoom minesite taken at 08:11 on 14 Nov 1994 from a height 
In his analysis of video data, King (1991) showed that, 
where the linear trends were consistent in all bands, 
simple band-ratios were a very effective means of 
reducing differential illumination caused by view angle. 
This method was also tried with unsatisfactory results. 
- Rayleigh scattering effects vary for each waveband, 
being greater in the shorter wavelength bands. Such 
effects, prominent in the Comalco data, seem to 
contradict King's assumptions. 
International Archives of Photogrammetry and Remote Sensing. Vol. XXXI, Part B2. Vienna 1996 
  
  
Pi 
vi 
Al 
of 
the 
D 
cu 
Fi;