Quaternion wavelets based rotation, scale and translation invariant texture classification and retrieval

摘要

This study addresses problem of classifying rotated, scaled and translated texture image using two stage log polar quaternion wavelet energy signatures. Sum of squared distance is employed to classify a set of 38 distinct natural textures selected from Brodatz album and 48 from Outex texture database. Based on different testing data sets for images with different orientations scales and translation, proposed classification scheme using QWT outperforms other approaches, giving a good classification accuracy and retrieval efficiency (grey scale, > 85% and color textures, > 83%).