VQ Classification based on MPEG-7 visual descriptors for video endoscopic capsule localization in the gastrointestinal tract

The paper addresses the problem of localization of video endoscopic capsule in the gastrointestinal (GI) tract on the base of appropriate classification of images received from it. In this context usefulness of MPEG-7 image descriptors as classification features has been verified. Novelty of the presented approach consists in joint application of vector quantization technique to finding representative codebooks of MPEG-7 features for different parts of GI tract and Principal Component Analysis for reduction of feature vectors' dimensions. In this initial research recognition of the upper (esophagus, stomach, duodendum) and the lower (colon) part of the GI tract has been performed. It has turned out that using the Homogenous Texture (HT) descriptor offers about 95% classification accuracy.