Incorporating spatial association into statistical classifiers: local pattern-based prior tuning

Abstract

This paper proposes a new classification method for spatial data by adjusting prior class probabilities according to local spatial patterns. First, the proposed method uses a classical statistical classifier to model training data. Second, the prior class probabilities are estimated according to the local spatial pattern and the classifier for each unseen object is adapted using the estimated prior probability. Finally, each unseen object is classified using its adapted classifier. Because the new method can be coupled with both generative and discriminant statistical classifiers, it performs generally more accurately than other methods for a variety of different spatial datasets. Experimental results show that this method has a lower prediction error than statistical classifiers that take no spatial information into account. Moreover, in the experiments, the new method also outperforms spatial auto-logistic regression and Markov random field-based methods when an appropriate estimate of local prior class distribution is used.

Spatial dependence between training and test sets: another pitfall of classification accuracy assessment in remote sensing

“…A variety of techniques exist are available including morphological image analysis (Fauvel et al 2013) and textural analysis (Franklin et al 2000;Puissant et al 2005;Sheeren et al 2009), sometimes based on variograms (Atkinson and Lewis 2000;Berberoglu et al 2007) or on a local version of principal component analysis (Comber et al 2016). The adaptation of classifiers has also been proposed using the spatial locations of the training samples (Atkinson 2004) or local spatial patterns (Bai et al 2020) to estimate class probabilities. Other authors suggest using local spatial statistics (Myint et al 2007;Ghimire et al 2010) or interpolated spectral values and their degree of similarity with actual values to improve the classification (Johnson et al 2012).…”