An improved the volume trace transform by a sub-micro pattern for solving the Gaussian noise image problem

This paper presents a new data compilation of the image transformation base on the series of the trace transform. The advantage of this new compilation is to solve the noise problem that usually appears in the pattern recognition. Submicro pattern analysis is employ to encode the series of the trace transform from a image which organizes by the shift-invariant sub-micro pattern scheme (SiSMP). Then, all data will sum to the final result by 2-D histogram with the Discriminant Feature Transform. The experiments of our approach show that the new compilation can increase the performance of the recognition compare with the LBP technique by using Brodatz texture database.