EM 1110-2-2907
1 October 2003
a pixel being accurately classified; it is a comparison to a reference. Commission deter-
mines the probability that a pixel represents the class for which it has been assigned. The
total accuracy is measured by calculating the proportion correctly classified pixel rela-
tive to the total tested number of pixels (Total = total correct/total tested).
Table 5-3
Omission and Commission Accuracy Assessment Matrix. Taken from Jensen (1996).
Reference Data
Classification
Residential
Commercial
Wetland
Forest
Water
Raw Total
Residential
70
5
0
13
0
88
Commercial
3
55
0
0
0
58
Wetland
0
0
99
0
0
99
Forest
0
0
4
37
0
41
Water
0
0
0
0
121
121
Column Total
73
60
103
50
121
407
Overall Accuracy =
382/407=93.86%
Producer's Accuracy (measure of omission error)
User's Accuracy (measure of commission error)
Residential= 70/73 = 964% omission error
Residential= 70/88 = 8020% omission error
Commercial= 55/60 = 928% omission error
Commercial= 55/58 = 955% omission error
Wetland= 99/103 = 964% omission error
Wetland= 99/99 = 1000% omission error
Forest= 97/50 = 7426% omission error
Forest= 37/41 = 9010% omission error
Water= 121/121 = 1000% omission error
Water= 121/121 = 1000% omission error
Example error matrix taken from Jensen (1986). Data are the result of an accuracy assessment of Landsat
TM data.
Classification method summary
Image classification uses the brightness values in one or more spectral bands,
and classifies each pixel based on its spectral information
The goal in classification is to assign remaining pixels in the image to a
designated class such as water, forest, agriculture, urban, etc.
The resulting classified image is composed of a collection of pixels, color-
coded to represent a particular theme. The overall process then leads to the
creation of a thematic map to be used to visually and statistically assess the
scene.
5-34
Previous Page
