Image classification using adaptive-boosting and tree-structured discriminant vector quantization

According to the principle of minimum description length, the best statistical classifier is the one that minimizes the sum of the complexity of the model and the description length of the training data. This paper focuses on improving the classification rate through correctly classifying the vectors that are misclassified by classifiers. For this purpose, a new tree-structured version of the algorithm, namely tree-structured discriminant vector quantisation, based on the BFOS algorithm. The major problem of the conventional algorithm is overcome by modifying the pdf of the training vectors using the adaptive-boosting algorithm. This new algorithm is implemented on a set of seven textures from the Brodatz data set.