Spectral clustering has gained importance in recent years due to its ability to cluster complex data as it requires only pairwise similarity among data points with its ease of implementation. The central point in spectral clustering is the process of capturing pair-wise similarity. In the literature, many research techniques have been proposed for effective construction of affinity matrix with suitable pairwise similarity. In this paper a general framework for capturing pairwise affinity using local features such as density, proximity and structural similarity is been proposed. Topological Node Features are exploited to define the notion of density and local structure. These local features are incorporated into the construction of the affinity matrix. Experimental results, on widely used datasets such as synthetic shape datasets, UCI real datasets and MNIST handwritten datasets show that the proposed framework outperforms standard spectral clustering methods.

In the field of clustering, spectral clustering (SC) has become an effective tool to analyze complex non-convex data using only pairwise affinity between the data points. Many novel affinity metrics have been proposed in the literature which use local features such as color, spatial coordinates, and texture. Some of these methods used SC for image segmentation [1, 2]. In this work, we have used the covariance matrix of the pixels in a patch and proposed an orientation based feature of a pixel called steering feature. This feature is robust and data-driven. The steering feature is used to enhance the construction of affinity metric for spectral clustering proposed by Shi and Malik [1]. Using the Nystrom framework [2] on images from BSD500 benchmark dataset, we have shown that the proposed affinity metric gives better result than Shi and Malik [1]

No abstract