GENERATING REPRESENTATIVE SETS AND SUMMARIES FOR LARGE COLLECTION OF IMAGES: APPROACH FOR GENERATING (2)

Overview of Proposed Solution
The proposed solution has three different phases. The first phase “Pre-cluster” is that we have three different techniques, followed by second phase “Clustering” phase which generates the representative set and the last phase is “Ranking mechanism” through which we generate the summary. The overview can be seen in the Figure 2.

In particular:

• We apply different pre-clustering techniques namely random and sequence in order to generate windows. The input will be an original large image set.

• We apply clustering to group appearance-wise similar images. The input will be windows generated from the pre-cluster phase. Then we fetch centroids of all clusters to generate a representative set.

• We apply ranking mechanism on the representative set and generate a summary by selecting any arbitrary number of top ranked images. So the input will be the representative set and the output will be a summary.

By applying these three phase, it will get the final outcome “summary”. One can see the graphical overview of the scenario in figure 3 for summary of 10 images.

Fig2Generating Representative_decrypted
Figure 2 : Overall Scenario of proposed solution.

As we described some heuristics in 4.1section, one may think that how can we obtain these heuristics by the proposed solution.

By clustering SIFT features; we can get groups of similar images. From each group (cluster) of images, we only fetch one image which is centroid of the cluster. So, now we have a set of images which are centroids of all clusters and should be visually distinct to each other. For more precise summary, we are applying ranking mechanism on top of the representative set in terms of highly dissimilarity.

Next we describe the algorithm in detail in the following sections. First, we discuss different preclustering techniques. Then we discuss the clustering algorithm and fetching centroids. Finally we describe how to rank each image of the representative set in order to generate summary.

Fig3Generating Representative_decrypted
Figure 3 : Overall Output Scenario