5.2.2. Clustering in COLD Environments

The previous results have shown that the use of FS for clustering is less suitable. Considering this, only HOG and *gist* descriptors are analysed in the experiments with the COLD environment. Figure 11 shows the results using HOG depending on the parameter *k*<sup>2</sup> in the Freiburg environment. Figure 12 shows the results of the clustering methods using *gist* depending on the parameter *k*<sup>3</sup> and with *nmasks*=16 in the Freiburg environment. In the same way, for the Saarbrücken environment, Figure 13 shows the results using HOG, and Figure 14 shows the results with *gist*. Regarding the use of HOG with the second method (using SOM), it was not able to solve the clustering task for *k*<sup>2</sup> = [4, 16] when *nc* > 60.

**Figure 9.** Results of the two clustering methods: computing time vs. number of clusters, when using FS, HOG, and *gist* descriptors in the Quorum V environment.

Again, spectral clustering is the best method, and in this case, *gist* presents better clustering outcomes. Hence, through the experiments carried out in the environments of the COLD database, a confirmation of the results obtained in Quorum V is reached (see Figure 15). Therefore, the proposed method is generalizable despite the use of different types of models (linear or grid). As a conclusion, the best option to carry out the compression of visual maps is reached when spectral clustering with *gist* is applied.

**Figure 10.** Quorum V environment. Cluster obtained with spectral clustering and *gist* description (*k*<sup>3</sup> = 32, *nmasks* = 16).

#### -

**Figure 11.** Results of the two clustering methods: average moment of inertia, average silhouette of points, and average silhouette of descriptors vs. number of clusters, when using HOG in the Freiburg environment.

#### -

**Figure 12.** Results of the two clustering methods: average moment of inertia, average silhouette of points, and average silhouette of descriptors vs. number of clusters, when using *gist* in the Freiburg environment.

**Figure 13.** Results of the two clustering methods: average moment of inertia, average silhouette of points, and average silhouette of descriptors vs. number of clusters, when using HOG in the Saarbrücken environment.

-- 

**Figure 14.** Results of the two clustering methods: average moment of inertia, average silhouette of points, and average silhouette of descriptors vs. number of clusters, when using *gist* in the Saarbrücken environment.
