In: Wagner W„ Szflcely, B. (eds.): ISPRS TC VII Symposium - 100 Years ISPRS, Vienna, Austria, July 5-7, 2010, IAPRS, Vol. XXXVIII, Part 7B 
338 
time-synchronous to receiving IRS images. Vegetative period 
for spring crops usually starts at the end of April - beginning 
of May. Fields are mainly bare at that time and vegetation 
has low value of dry matter as well as leaf area index. 
Phenological observation of crops during the spring and 
summer period (Table 1) were used for crop classification 
Very low llllllHlllllll Medium to dense 
Table 1. Vegetation development during the spring and 
summer period. 
based on IRS-P6 and IRS-1 D data. Multispectral data was 
obtained on May 18, June 16 and July 7. To evaluate the 
status of vegetation within the period of crop sampling we 
measured plant height, leaf area index, aboveground biomass 
and dry matter. 
Peculiarities of crop development were used to analyze per 
field vegetation development. Information on crop cover 
types and vegetation density (very low, low, medium, 
medium to dense, dense) based on ground truth data provided 
information for image classification. 
Ten training classes for crop classification were used in 
spite of that seven crop were cultivated. Peas, winter wheat 
and spring barley were located within two fields but they 
differed by the level of fertilization, crop variety and 
previous crop. It gains to differences in development of 
aboveground mass and yield components. 
After the training sites had been created, the three methods 
were used to determine if a specific pixel qualifies as a class 
member. The minimum distance procedure, the minimum 
distance method with standardized distances and the method 
of maximum likelihood known to be hard classifiers were 
applied. They made a definitive decision about the land 
cover class to which any pixel belongs (Figures 1 to 3). 
The minimum distance classifier uses the mean vectors of 
each training site and calculates the Euclidean distance from 
each unknown pixel to the mean vector for each class. All 
pixels are classified to the nearest class unless a standard 
deviation or distance threshold is specified, in which case 
some pixels may be unclassified if they do not meet the 
selected criteria. The method made some mistakes in 
classification results because of standard deviation of pixel 
spectral characteristics within the polygons. 
C 
Figure 1. Crop classification maps derived from IRS-ID image 
(May 18): A - minimum distance procedure; B - minimum 
distance method with standardized distances; C - maximum 
likelihood classifier