In the recent years binary coding of image features such as local binary pattern and local phase quantization became popular in large variety of image quantification tasks. Lately some non-binary codings, such as local ternary pattern, were proposed to improve the performance of these binary based approaches. In these methods it is very important to correctly choose the thresholds applied for building the coding used to represent a given image and its features by a feature vector. In this work we compare several approaches for extracting local ternary/quinary pattern image features and ternary coding for local phase quantization on various types of biological microscope images using six image databases for sub-cellular and stem cell image classification. We use these image features for training a stand-alone support vector machine and a random subspace of support vector machines to separate the different classes present in each dataset . Moreover also some distances are tested. Our results showed that, on the chosen datasets, the best approach is a multi-threshold local quinary coding: the use of a more discriminating coding than the binary one, combined with a pool of thresholds, helps in distinguish between descriptive features and noise, providing improved classification results. The Matlab code is available at bias.csr.unibo.it/nanni/TernaryCoding.rar.
Keywords: machine learning, non-binary coding, stem cell image, sub-cellular image, support vector machine, texture descriptors.