CONCLUSION
In this paper a total subjects are taken three. Each
person with 9 variant images. Hence total images taken
are 27. Here 17 images are of changed angle, 4 images
are of front faces, 4 are of changed expressions and 2
are of changed complexions. It is observed that the
Euclidian distance method does not take into account
the variability of the values in all dimensions
particularly for images with changed angle, and is
therefore not an optimum discriminant analysis
algorithm for this case. The L1 norm (Manhattan
distance) distance gives the better results as Euclidean
distance method, it has recognized total 8(29.69%)
images out of 27 while L1 norm (Manhattan distance)
has recognized total 11(40.74) images out of 27. Hence
using L1 norm distance there is an improvement in the
performance.