A Fast Implementation of the Isodata Clustering Algorithm

Abstract

Clustering is central to many image processing and remote sensing applications. ISODATA is one of the most popular and widely used clustering methods in geoscience applications, but it can run slowly, particularly with large data sets. We present a more efficient approach to ISODATA clustering, which achieves better running times by storing the points in a kd-tree and through a modification of the way in which the algorithm estimates the dispersion of each cluster. We also present an approximate version of the algorithm which allows the user to further improve the running time, at the expense of lower fidelity in computing the nearest cluster center to each point. We provide both theoretical and empirical justification that our modified approach produces clusterings that are very similar to those produced by the standard ISODATA approach. We also provide empirical studies on both synthetic data and remotely sensed Landsat and MODIS images that show that our approach has significantly lower running times.

Remote Sensing of Glacier Change in the Central Qinghai-Tibet Plateau and the Relationship with Changing Climate

“…As shown in Figure 2, the absolute percent difference in the total mapped glacier area between our choice of threshold and the thresholds in 0.35-0.45 is mostly within 0.8%. To retrieve glacier area from the MSS image without short-wave bands, we used the ISODATA unsupervised classification method that iteratively clusters pixels class using minimum distance techniques [29]. In this process, the scene was firstly clipped around the glacier region and then was classified into four classes based on all the four bands (Table 1) with up to four iterations.…”

Acoustic mapping and classification of benthic habitat using unsupervised learning in artificial reef water

“…Unsupervised clustering algorithms are fundamental tools in image processing for geoscience and remote sensing applications (Memarsadeghi et al, 2007). The iterative self-organizing data analysis technique (ISODATA) is a widely used classification methods (Micallef et al, 2007) that approximate the natural construction of a multidimensional dataset by iteratively passing it through defining classes to minimize pixel separation distance (D) and sum of squared error (SSE) (Ball and Hall, 1967;Jones et al, 2014).…”

Real-Time Plane Segmentation Using RGB-D Cameras

“…For being able to merge multiple clusters, we keep track of the conducted merges. In case cluster a should be merged with cluster b that was already merged with cluster c, we check if we can merge a and c, or if a + b is a better merge than a + c. Although this procedure is less adaptive (and complex) as sophisticated clustering algorithms like k-Means, mean-shift-clustering or, e.g., ISODATA [7], this simple approach allows for reliably detecting larger planes in 3D point clouds at high frame rates. In all modes and resolutions of the camera, plane segmentation is only a matter of milliseconds.…”