Automatic medical image categorization and annotation using LBP and MPEG-7 edge histograms

This paper presents a global and local texture features combination as the description of medical images. The approaches use the block based Local Binary Pattern (LBP) texture histogram and MPEG-7 edge histogram together as the medical image description. The block histograms and the whole image texture histograms are accumulated to generate the image feature vector. Then, One-against-One SVM is adopted for automatic medical image annotation task in ImageCLEFmed2007. The experiments show that Local Binary Pattern texture and Edge Histogram are powerful for automatic medical image annotation.