Rotation invariant texture recognition by using Neighbor Discriminant Feature Transform and reinforcement learning

This paper presents the method of image texture recognition by Volume Trace Transform-VTT, based on several Trace function. All of 10 trace functions will be selected by the reinforcement learning process and constructed to produce noticeable features. The next process is to sum the results of each function together by “NDFT - Neighbor Discriminant Feature Transform”, this process will all the results from the different positions of image results and construct 2-D histogram. The histogram is evaluated by chi-square statistic test and the process works on the basis of brodatz texture and vision texture database.