TSML: A XML-based Format for Exchange of Training Samples for Pattern Recognition in Remote Sensing Images

Abstract

The availability of large and complex data sets has shifted the focus of pattern recognition towards developing techniques that can efficiently handle these types of data sets. For example, Multiple Classifier Systems claim their ability in reducing the error and complexity of classification by partitioning the data space and combining classifiers predictions. However, it is not an easy task to generate several partitions and moreover to use them in an efficient manner. Another difficult aspect is related to the exchange of training data in different formats among systems to combine classifiers of different and heterogeneous systems. This paper presents a model and structure of training samples based on XML (eXtensible Markup Language) to facilitate the partitioning and exchange among different image classification system. The main contribution is to apply the flexibility of XML that addresses interoperability and communication among heterogeneous systems in partitioning data sets as well as to facilitate interchange of such sets among image processing and pattern recognition systems.