Optimal dimension

In this experiment we take SIFT of certain topology (15x9x13) and variant and do the classification using the FLD and 1-NN in projection space.

Table 4.12: We applied the FLD projection to various number of dimensions, using 1-NN as a classifier in that projected space we obtained these results

	Classification Rate [%]
Feature Space Dimension	1-NN	nearest $\mathbf{\mu}$
1	31.52	31.52
2	48.91	47.82
3	66.30	60.87
4	66.30	71.73
5	81.52	80.04
6	81.52	83.69

Figure 4.18: The optimal feature space dimension selection for car images