Incremental Spectral Clustering via Fastfood Features and Its Application to Stream Image Segmentation

Abstract

We propose an incremental spectral clustering method for stream data clustering and apply it to stream image segmentation. The main idea in our work consists of generating the data points in the kernel space by Fastfood features and iteratively calculating the eigendecomposition of data. Compared with the popular Nyström-based approximation, our work accesses each data point only once while Nyström, in particular the sampling scheme, will go through the entire dataset first and calculate the embeddings of data points with a second visit. As a result, our method is able to learn data partitions incrementally and improve eigenvector approximation with more and more data seen from a stream. By contrast, the performance of the standard Nyström is fixed when the sample set is selected. Experimental results show the superiority of our method.

Fast Incremental Spectral Clustering in Titanate Application via Graph Fourier Transform

“…Inspired by the idea in [29], an online k-means algorithm is proposed in [30], that when processing a point, opts to start a new cluster if the point is far from the current centers. Closely related to our approach are several online methods for spectral clustering [10]- [13]. Reference [10] approximates Eigensystem and avoid re-computation as new points arrive by introducing the incidence vector/matrix.…”

“…In [11], [12], the online clustering methods are proposed by introducing the concept of representative set, they compress data set and only update the representative sets as new points come. Recently, an incremental spectral clustering method for image segmentation is proposed in [13], the main idea consists of generating the data points in the kernel space by Fastfood features and iteratively calculating the eigendecomposition of data.…”

An Image Segmentation Method Based on Improved Regularized Level Set Model

“…However, its drawbacks include the slow convergence of the iteration and over smoothing when suppressing noise [10]. He et al [11] investigated an incremental spectral clustering method for stream image segmentation. It is known that for traditional spectral clustering, a scaling parameter needs to be fixed artificially, and obtaining its optimal value is very difficult in a Gaussian kernel function.…”

K-Hyperline Clustering-Based Color Image Segmentation Robust to Illumination Changes

“…These methods are mainly based on watershed transform or clustering procedures [8]. The former computes the spatial gradient of image luminance; however, the chromatic information is ignored [9]. The latter depends on the balance of the pixel color distribution [10].…”