basis for further and more detailed investigations in using the 
textural parameter for testing its repeatability in signature 
extension studies over a. variety of crop/land use types. If 
successful, it may lead to substantial improvement in automated 
classification accuracies (very economically) and may have 
particular significance to relatively poorer and developing 
nations. 
Introduction: This paper is based on some research work carried 
out by me recently while I was at the University of Michigan at 
Ann Arbor in U.S.A. 
I directed this study to examine the potential role of texture 
in computer-aided signature analysis using five selected crop- 
types (uncut corn, cut corn, wheat fields, alfalfa and idle fields) 
as a case-study. The prime motivation for this research was the 
fact that in most of the studies carried out so far in signature 
analysis of multispectral reflectance data, only the tonal data 
has been extensively used. Experience, however, tells us that 
variations of spectral reflectance are often greater within a 
single crop type than between different crops. These variations 
limit the accuracy of identification and discrimination when only 
tone is used for such purposes. 
Tonal variations are caused by factors both internal and external 
to the scene. There are natural variations within a single crop 
type due to a large number of factors such as state of growth, 
soil, aspect and sun angle - view angle effects. Texture tends to 
integrate the effects of many of these internal and external 
factors in the form of 'variability of spectral reflectance within 
an agricultural field' and seems to possess a statistical distri- 
bution characteristic of the crop type and its condition. 
Measure of Texture evaluated 
  
In this study "Standard deviation of photographic density within 
a single field" was selected as the only measure of texture. This 
was done primarily due to the inherent simplicity of the measure, 
keeping in mind the convenience with which it can be used by users 
if its utility could be established. 
Experimental Design and Measurements 
  
To isolate the variabilities in tonal and textural values due to 
causes external to the crop types, the areas chosen appeared on a 
single aerial photograph at a scale of 1:14,700. Thirteen replicates 
of each of the five crops were selected for the study. To study the 
effect on tone and texture due to change of scale (and consequently 
of resolution), the same aerial negative was used to obtain two 
times and four times enlarged prints. The printing was carried out 
under sensitometrically controlled conditions and fields were 
selected from within the central region of the photograph to mini- 
mise the variability due to the photographic lens fall-off effects. 
The ground truth was collected within a few days of the aerial 
photography. Well distributed density measurements for all the 
fields of each crop type (13 fields of the five crop types) were 
carried out by a densitometer with a fixed aperture size of 1.5 mm. 
Wherever possible obvious discontinuities of crop cover (crop type) 
within a field were avoided to provide as faithul a signature of